Time Reversal Symmetry for Classical, Nonrelativistic Quantum and Spin Systems in Presence of Magnetic Fields
Extracted source text · arXiv 2104.13767
\documentclass[12pt]{article}
\pdfoutput=1
\usepackage[dvipsnames]{xcolor}
\usepackage[utf8]{inputenc}
\usepackage[english]{babel}
\usepackage{amsmath}
\usepackage{hyperref}
\usepackage{breakurl}
\usepackage[T1]{fontenc}
\usepackage{multicol}
\usepackage{mathtools}
\usepackage{float}
\usepackage{amssymb}
\usepackage{xfrac}
\usepackage{doi}
\usepackage[margin=0.75in]{geometry}
\usepackage[maxbibnames=99,sorting=none]{biblatex}
\addbibresource{references.bib}
\usepackage{amsthm}
\usepackage{ytableau}
\usepackage{braket}
\usepackage{csquotes}
\usepackage{titling}
\AtBeginBibliography{\renewcommand\finalandcomma{}}
\newcommand{\be}{\begin{equation}}
\newcommand{\ee}{\end{equation}}
\DeclareRobustCommand{\rchi}{{\mathpalette\irchi\relax}}
\newcommand{\irchi}[2]{\raisebox{\depth}{$#1\chi$}}
\setlength{\droptitle}{-3cm}
\usepackage{caption}
\usepackage{graphicx}
\usepackage{subfiles}
\newtheorem{example}{Example}[section]
\usepackage{abstract}
\usepackage[makeroom]{cancel}
\usepackage{tensor}
\usepackage{youngtab}
\renewenvironment{abstract}
{\small
\begin{center}
\bfseries \abstractname\vspace{-.5em}\vspace{0pt}
\end{center}
\list{}{%
\setlength{\leftmargin}{50mm}% <---------- CHANGE HERE
\setlength{\rightmargin}{\leftmargin}%
}%
\item\relax}
{\endlist}
\numberwithin{equation}{section}
\newtheorem{cor}{Corollary}[section]
\newtheorem{theo}{Theorem}[section]
\newtheorem{prop}{Proposition}[section]
\newtheorem{defi}{Definition}[section]
\newtheorem{assu}{Assumption}[section]
\newtheorem{rem}{Remark}[section]
\newtheorem{lemma}{Lemma}[section]
\begin{document}
\date{\vspace{-5ex}}
\title{%
\textbf{\textit{\large{Time Reversal Symmetry for Classical, Non-relativistic Quantum and Spin Systems in Presence of Magnetic Fields }}}\\
\large{\vspace{+2ex}Davide Carbone}\\
\small{\vspace{+0.1ex}\textit{INFN, Sezione di Torino, Via P. Giuria 1, 10125 Torino, Italy}}\\
\small{\vspace{+0.1ex}\textit{and Dipartimento di Scienze Matematiche, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy\\
\normalfont{davide.carbone@polito.it}}}\\
\large{\vspace{+2ex}Paolo De Gregorio}\\
\small{\vspace{+0.1ex}\textit{Dipartimento di Scienze Matematiche, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy\\
\normalfont{paolo.degregorio@polito.it}}}\\
\large{\vspace{+2ex}Lamberto Rondoni}\\
\small{\vspace{+0.1ex}\textit{INFN, Sezione di Torino, Via P. Giuria 1, 10125 Torino, Italy}}\\
\small{\vspace{+0.1ex}\textit{and Dipartimento di Scienze Matematiche and Graphene@Polito Lab Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy\\
\normalfont{lamberto.rondoni@polito.it}}}}
%\maketitle
%
% \begin{abstract}
% Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract Abstract
% \end{abstract}
% \noindent\rule{\textwidth}{1pt}
\leftskip=0cm
\rightskip=0cm
\maketitle
\noindent\textbf{Corresponding author}: Davide Carbone, davide.carbone@polito.it\\
\noindent\rule{\textwidth}{1pt}
\section*{\large Abstract}
\subfile{abstract}
\noindent\rule{\textwidth}{1pt}
\section{\large Introduction}
\subfile{introduction}
\section{\large Results outline}
\subfile{results}
\section{\large Theory}
\subfile{Theory}
\section{\large Results and Discussion}
\subfile{Conclusion}
\printbibliography
\end{document}
In this section we provide a summary of the results proven in Section \ref{TheorySect}, in particular:
\begin{enumerate}
\item the characterization of time reversal operations for Classical and
non-relativistic Quantum Mechanics systems, including spin 1/2 particles, in presence of an external magnetic field;
\item the extension of compatibility conditions between time reversal operations and magnetic field;
\item the extension to quantum mechanics and to non-constant magnetic fields of analytical results about correlators \cite{bonella2017time}, directly derived from the application of more than one time reversal symmetry.
\end{enumerate}
The experimental relevance of the Onsager reciprocal relations in absence of magnetic effects, is reviewed in a wide fraction of literature, see {\em e.g.}\
Refs.\cite{miller1960thermodynamics,miller1973onsager,letey1969movement,monroe2006onsager,gaeta1994phonons, miyatani1981experimental,rowley1978dufour,matthews2014experimental,kimura2007room,sola2019experimental},
which span from diffusion of various kinds, to quantum thermoelectric transport, and even spin and charge simultaneous transport.
In particular, Ref.\cite{jacquod2012onsager}
investigates theory and possible experiments about the reciprocal relations in spin-Hall and inverse spin-Hall effects, in spin-injection, in magnetoelectric spin currents, Seebeck, Peltier, spin Seebeck, and spin Peltier effects, in systems with and without coupling to superconductors. Or, on the other hand, Ref.\cite{marklund2007dynamics} specifically analyzes linear response in the context of Pauli equation and plasma dynamics.
Our results predict that the onset of an external magnetic field breaks neither the Onsager symmetry nor the validity of the Fluctuation Theorem a priori. Moreover, they predict that certain cross transport coefficients of {\em e.g.}\ superionic conductors are not merely small, hence hard to measure, they actually vanish. In fact, Refs.\cite{bonella2017time,rondoFT1,casa} have already verified numerically some of these predictions, using the previously found time reversal symmetries.
\subsection{Characterization of generalized time reversal operations}
%
First we identify the general structure of a time reversal operation that preserves the equations of motion. We begin with the classical mechanics a system of $N$ independent charged particles, each with 3 degrees of freedom, confined within a box of side $2L$, and coupled to an external magnetic field, with Hamiltonian
\begin{equation}
\label{H1}
H=\sum_{i=1}^N\left[\frac{(\boldsymbol{p}_i-q_i\boldsymbol{A}(\boldsymbol x_i))^2}{2m_i}\right]
\end{equation}
Our Theorem \ref{theo:1.1.2}
shows that a class of extended time reversal operations $\mathcal{M}$ for such systems is represented by $6N\times 6N$ matrices of the form:
\begin{equation}
\mathcal{M}=
\begin{pmatrix}
A&0\\
0&-A\\
\end{pmatrix}
\end{equation}
with $A\in O(3N)$, the orthogonal group, and $A^2=I$, the identity matrix, so that $\mathcal{M}(\boldsymbol X,\boldsymbol P)=(A\boldsymbol X,-A\boldsymbol P)$.
These constitute an infinite set of transformations, that includes the standard time reversal operation $(\boldsymbol X,\boldsymbol P) \mapsto (\boldsymbol X,-\boldsymbol P)$, where $A=I$, and includes a number
\begin{equation}
\Tilde{\Delta}_{even}(M)=\sum_{r_2=0}^{\lfloor M/2\rfloor}\frac{M!\, 2^{M-2r_2}}{(M-2r_2)!\,r_2!} \, , \quad \mbox{with } ~ M=3N ,
\end{equation}
of permutations of coordinates, cf.\ Corollary \ref{cor:1.1.1} below.
%In our work we want to stress that in principle one can use a huge number of time reversal operation as a direct consequence of the phase space formalism: coordinates and momenta have the same dignity and so any swap can be used to build time reversal operation. Retracing the exact trajectory in configuration space is not necessary to preserve time reversal hence Onsager reciprocal relations; what counts is that the integral contribution of all trajectory is preserved.
For the non-relativistic spinless quantum systems, time reversal concerns the Schr\"odinger equation, and is represented
by the Wigner operator $\mathcal{T}$ defined above, keeping in mind that a transformation of the coordinate operator $\boldsymbol X$ implies a modification of the momentum operator $\boldsymbol P=-i\partial_{\boldsymbol X}$.
We state that $\mathcal{T}$ must preserve the canonical commutation relations, as in Classical Mechanics the Poisson brackets must be preserved. Then, the matrix representation of the action of $\mathcal{T}$ on $\boldsymbol X$ and $\boldsymbol P$ must be symplectic \cite{anderson1994canonical}.
Considering that $K$ acts as the antisymplectic $6N\times 6N$ matrix $\operatorname{diag}(I,-I)$, where $I$ is the $3N\times 3N$ identity matrix, we conclude that $U$ must be antisymplectic.
%In fact, as stressed in the Introduction, the case where $U$ is the identity corresponds to the canonical time reversal operation thanks to the imaginary unit.\\
%The problem is then reduced to characterize the general structure of $U$ in the case of an
%the same we did in Classical Mechanics for $\mathcal{M}$. It would be convenient if we could recycle, at least in part, the characterization done in classical context. This is possible since thanks to Anderson
Our Theorem \ref{theo:1.2.1} characterizes $U$ similarly to the classical mechanical reasoning, apart from the fact that quantum mechanics also allows $\mathcal{T}^2=-I$, not only $\mathcal{T}^2=I$.
%The case $\mathcal{T}^2=I$ is perfectly equivalent to the classical situation.
%including the permutation.
%On the other hand, Proposition \ref{prop:1.2.6} shows that $\mathcal{T}^2=-I$ the formula that provides the number of matrices corresponding to swaps (up to minus signs) with square $-I$, whose building block is the $2\times 2$ matrix
%\begin{equation}
% Q=
% \begin{pmatrix}
% 0&-1\\
% 1&0
% \end{pmatrix}
%\end{equation}
Moreover, a simple example shows that $U$ has to be real valued, for the Hamiltonian $\eqref{H1}$.
%In conclusion, the largest part of this section is devoted to show the necessary tools to trace a parallel with classical situation; once we have done that, the characterization and the counting of the elements in swap subclass for Wigner operator is exactly the same showed for classical time reversal.
Finally, we consider time reversal invariance for particles of spin 1/2. In textbooks, such as Sachs \cite{sachs}, the action of the Wigner operator $\mathcal{T}=(U_{c}\otimes U_{spin})K$, where $U_{spin}$ acts on Pauli matrices $\boldsymbol \sigma_i=(\sigma^i_x,\sigma^i_y,\sigma^i_z)$, is defined considering spin analogous to angular momentum, which means:
\begin{equation}
\label{axio}
\mathcal{T}\boldsymbol\sigma\mathcal{T}=-\boldsymbol\sigma
\end{equation}
This implies $U_{spin}=\bigotimes_{i=1}^N \sigma_y^i$.
%The reason of this assumption is the analogy between spin and angular momentum $\boldsymbol L$: since $\boldsymbol{L}$ acquires a minus sign in Classical Mechanics under canonical time reversal, and since $\boldsymbol\sigma$ can be interpreted as an intrinsic angular momentum, then one imposes \eqref{axio}.\\
In Proposition \ref{prop:2.6}, we show that thanks to the generalized time reversal operations also condition \eqref{axio} can be relaxed. In particular, it suffices that $\mathcal{T}$ preserves the canonical commutation relation for the Pauli matrices algebra $\mathfrak{su}(2)$.
%In some sense, the path is the same we did in Classical Mechanics: there was a canonical structure of time reversal that is sufficient to preserve equations of motion, but that is way too restrictive for obtaining Onsager reciprocal relations.\\
%Using the isomorphism between the Lie groups $SU(2)/\mathbb Z_2$,
%Theorem \ref{theo:1.2.3} shows that %explicit expression of
%$U_{spin}$ in some particular situation provided that it separately acts on each particle: we study firstly the case
%\begin{equation}
% \mathcal{T}^{-1}(\sigma^i_x,\sigma^i_y,\sigma^i_z)\mathcal{T} =(s_1\sigma^i_x,s_2\sigma^i_y,s_3\sigma^i_z)
%\end{equation}
%with $s_1,s_2,s_3=\pm 1$; notice that the product $s_1s_2s_3=1$ since the algebra must be preserved (see the Theory section for the details). Then the case
%in which we have a swap of two of the matrices with appropriate signs.
%In conclusion, in this first part of the paper we provide plenty of new possibilities to preserve time reversal symmetry, at least in this generalized acceptation. As we showed in the introduction, this generalized T-symmetry is sufficient to imply the validity of Onsager reciprocal relations. Until now, the only information that we mentioned about magnetic field is the structure of the Hamiltonian, and in particular the presence of the minimal coupling term.
%The open problem treated in the second part of the paper is the compatibility condition between any of these new time reversal operations we found and a precise magnetic field.
\subsection{Extension of compatibility conditions}
In this part of the paper we extend the compatibility conditions of Carbone and Rondoni \cite{carbone2020necessary}
between the time reversal operations known at the time and both vector potential and magnetic field. These are the necessary and sufficient conditions for the Hamiltonian $H$ to be invariant under time reversal.
%In formulae, if they call $\mathcal{M}_m$ the representation of $\mathcal{M}$ on the three dimensional space of the coordinates of one particle. found by Carbone and Rondoni \cite{carbone2020necessary} For example
% \begin{equation}
% \mathcal{M}_m=
% \begin{pmatrix}
% s_x&0&0\\
% 0&s_y&0\\
% 0&0&s_z
% \end{pmatrix}
% \end{equation}
% for \eqref{1.59}.
The compatibility condition for the vector potential, that is defined up to a gauge transformation, was given by
\begin{equation}
\label{2.8}
\mathcal{M}_m\boldsymbol A(\mathcal{M}_m\boldsymbol x)=-[\boldsymbol A(\boldsymbol x)]_R
\end{equation}
where $\mathcal{M}_m$ is the representation of the time reversal map on the space of the coordinates of one particle, and $[\boldsymbol A(\boldsymbol x)]_R$ is the vector potential in the equivalence class of those with same magnetic field $\boldsymbol B=\nabla \times \boldsymbol A$. The compatibility condition with the magnetic field was given by
%more useful since there is not gauge freedom
\begin{equation}
\label{2.9}
\operatorname{det}(\mathcal{M}_m) \mathcal{M}_m\boldsymbol B(\mathcal{M}_m\boldsymbol x)=-\boldsymbol B(\boldsymbol x)
\end{equation}
We extend these conditions obtained for classical systems to the generalized time reversal operations derived here, and to the quantum case, including the spin 1/2 particles.
%where $\mathcal{M}$ is a classical time reversal operation and $\mathcal{T}$ is a Wigner time reversal operator. Going into details, we can divide the section in two parts:
%\begin{itemize
%\textbf{Classical Mechanics and Spinless Quantum Mechanics.} The first result of this section is the extension of \eqref{2.8} and \eqref{2.9} in Classical Mechanics for any generalized time reversal operation between the ones we characterize in the first part of the paper, then far beyond only the $20$ analyzed in \cite{carbone2020necessary}. In the case of vector potential, we found an expression similar to \eqref{2.8} but regarding the $3N$ dimensional vector field %\begin{equation}
% \boldsymbol{\mathcal A}(\boldsymbol X)\equiv(\boldsymbol A(\boldsymbol x_1),...,\boldsymbol A(\boldsymbol x_N))
% \end{equation}
% that is the compatibility condition
% \begin{equation}
% \label{2.12}
% \mathcal{M}_m %\boldsymbol{\mathcal A}(\mathcal{M}_m\boldsymbol X)=-[ \boldsymbol{\mathcal A}(\boldsymbol X)]_R
% \end{equation}
% where $[ \boldsymbol{\mathcal A}(\boldsymbol X)]_R=([\boldsymbol A(\boldsymbol x_1)]_R,...,[\boldsymbol A(\boldsymbol x_N)]_R)$. Notice that we used the same symbol $\mathcal{M}_m$ for the matrix representative of the time reversal operation, but in the present case it is a $3N\times 3N$ dimensional matrix. As a side note, we prove that \eqref{2.8} is a corollary of the more general result \eqref{2.12}.\\
% that compatibility condition on vector potential, although really general, is of difficult application since, first of all, in experimental situation one fixes $\boldsymbol B$ and $\boldsymbol A$ is often unknown; moreover, the gauge freedom makes \eqref{2.12} not so easy to check. In the light of this, we provide a proof of validity of the compatibility condition on magnetic field \eqref{2.9} also in the case of transformations that separately act on each particle subspace other than the $20$ previously analyzed in literature.\\
% The subsequent important result is the extension of these compatibility conditions to non-relativistic quantum mechanical systems; with respect to classical systems, the derivation of \eqref{2.12} and \eqref{2.8} is possible provided that the coordinate space is compact (\textit{e.g.} if the particles are confined in a cube). This is an important result, since \cite{rondoquantu}, as mentioned in the Introduction, is the only work on Onsager reciprocal relations and generalized time reversal that deals with quantum systems.\\
% Ultimately, in the light of our results, we show an exemplary case study: the case of constant magnetic field along $z$ axis; this is a fundamental example because the present research line originates from the "no-go" statement by Onsager \cite{onsager1931reciprocal} about the breaking of its reciprocal relations. In fact, in the present paper we prove the opposite, in some sense, statement: we have an infinite number of suitable time reversal operations that preserves T-symmetry. We can parametrize them via the matrix that acts on coordinate space of each particle:
% \begin{equation}
%\mathcal{M}_m=
% \begin{pmatrix}
% a&b&0\\
% b&-a&0\\
% 0&0& 1
% \end{pmatrix}
%\end{equation}
%It can be easily shown that $\mathcal{M}_m$ and $\boldsymbol B=B(0,0,1)$, with constant $B$, satisfies \eqref{2.9}.
In particular,
%\textbf{Invariance of
given the Pauli Hamiltonian
%. In the present subsection we prove, in our view, the most important result of the entire paper. We have already showed that we can generalize time reversal also in the context of particles with spin $1/2$. The interesting point regards the compatibility condition between generalized time reversal operators and magnetic field in the case of a system of particles described by Pauli Hamiltonian
\begin{equation}
H=\sum_{i=1}^N{\frac {1}{2m_i}}\left[(\boldsymbol {p}_i -q_i\boldsymbol {A}(\boldsymbol x_i) )^{2}-q_i\boldsymbol{\sigma}_i\cdot\boldsymbol{B}(\boldsymbol x_i)\right]
\end{equation}
we prove that, contrary to expectations that
%In fact, one would expect that the presence of
the additional term of spin field coupling $\boldsymbol\sigma_i\cdot \boldsymbol B$ reduces the number of available time reversal operators, %$\mathcal{T}=(U_{coor}\otimes U_{spin})$. On the contrary, we prove that
no additional compatibility condition other than \eqref{2.9} is needed.
In other words, %we prove that
if a time reversal operator preserves the spinless minimal coupling term, there is
%we can always find
an appropriate $U_{spin}$ that
%in order to
preserves also the spin field coupling. %Apart from technical details that are shown in the proof, the root of this fact is in the tensor product structure of the Hilbert space: briefly speaking, given any orthogonal transformation of the vector $\boldsymbol B$, we can always reabsorb it without constraints through a "coordinate transformation" in spin space thanks to the isomorphism between the Lie groups $SU(2)/\mathbb Z_2$ and $SO(3)$.
This result is fundamental because it means that from the point of view of statistical relations, such as the Onsager relations and the Fluctuation Theorem there is no effective difference between spinless and spin $1/2$ particles described via Pauli Hamiltonian.
%\end{itemize}
Finally, we extend a result proven in \cite{bonella2017time} showing that
%using generalized diagonal time reversal operations
the off diagonal entries of the diffusion tensor for classical particles subjected to a particular constant magnetic field obey:
\begin{equation}
\label{dd}
D_{xy}=-D_{yx}
\end{equation}
%Their proof is limited to systems of classical spinless charged particles subject to a constant magnetic field. Then, we take advantage of the result demonstrated in the present paper to extend
We show that the same holds
%the validity of \eqref{dd} to
in quantum systems, and also in presence of a non-constant magnetic field along $z$ axis.
\label{TheorySect}
%
As recalled in Sect.\ref{sec:Introduction}, sufficient conditions for the validity of Onsager reciprocal relations are known, and they include a set of generalized time reversal symmetries. On the contrary, very little is known about \textit{necessary} conditions.
First, in Sec.\ref{ClassMe}, we recall the main facts about time reversal
transformation in classical mechanics, introducing a whole new class of operations. Then, we illustrate quantum mechanical time reversal
operators, in the case of spinless particles, extending the treatment of Sec.\ref{ClassMe} to the quantum realm, cf.\ Sec.\ref{sec:The time reversal operator}. In Sec.\ref{sec:Time reversal with spin} we investigate the issue of time reversal in the context of quantum systems composed by particle of spin $1/2$. In Sec.\ref{sec:compatib} we summarize the concept of Kubo canonical quantum correlators, as well as the derivation of Onsager relations from a generalized time reversal symmetry obtained in Ref.\cite{rondoquantu}.
Subsequently, we illustrate the compatibility conditions between the different time reversal operations and a generic magnetic field. We recall the sufficient conditions of Ref.\cite{carbone2020necessary} that lead to Onsager relations in classical mechanics, and we extend them to the operations found in Sec.\ref{ClassMe} and to the case of particles with spin. The striking result of this Section is that a time reversal operator that works in the spinless case, works also for particles of spin $1/2$. Finally, in Sec.\ref{sec:example} we study the constraints imposed by TRI on the diffusion tensor in a particular physical setup, in the wake of the reasoning of Ref.\cite{rondom}.
\subsection{Classical mechanics}
\label{ClassMe}
The microscopic state of a system made of $N$ classical particles is represented by the collection of coordinates and momenta of all its particles, that constitutes a point $\Gamma$ in the phase space $\Omega$. In case of dynamics determined by a Hamiltonian $H$, the time evolution of the coordinates and momenta is prescribed by the following equations of motion:
\begin{equation}
\label{eq:2.2}
\begin{dcases}
\frac{\partial H}{\partial p_i}=\dot{x}^i\\
\frac{\partial H}{\partial x^i}=-\dot{p_i}
\end{dcases}\;\;\;\;\;\; i=1,...,3N
\end{equation}
Denote by $S^t : \Omega \to \Omega$ the operator that expresses the solutions of the equations of motion up to time $t$, {\em i.e.}\ $S^t \Gamma$ is the state at time $t$, if it was $\Gamma$ a time 0. Following Abraham Ref.\cite{abraham1978foundations}, the dynamics are called time reversal invariant (TRI) if there exists a linear antisymplectic involution
${\cal M} : \Omega \to \Omega $ (an operator such that ${\cal M}^2 = I$, the identity) such that:
\begin{equation}
{\cal M} S^t \Gamma = S^{-t} {\cal M} \Gamma \, ,
\quad \forall \Gamma \in \Omega \, , ~ \forall t \in \mathbb{R}
\label{TRIdef}
\end{equation}
The reason for this terminology is that right multiplication by $\cal M$ before application to $\Gamma$ yields:
\begin{equation}
S^{-t} = {\cal M} S^t {\cal M}
\label{TRIjusti}
\end{equation}
which means that the backward time evolution is conjugated to the forward time evolution {\em via} the involution $\cal M$, in such a way that one backward trajectory can be obtained by properly applying $\cal M$ to the forward trajectory.
Microreversibility corresponds to TRI under the involution $\mathcal{M}$ defined by ${\cal M}({\bf x},{\bf p})=({\bf x},-{\bf p})$,
%has been considered the only physically suitable time reversal operator, although other mappings with the same property \eqref{TRIjusti} have been identified.
but we call time reversal invariant
%not follow this approach, adopting that of Refs.\cite{rondoe,rondom,rondoquantu,carbone2020necessary}, that
all the dynamics for which one involution obeying Eq.\eqref{TRIdef} exists.
From the point of view of the equations of motion, we can define an extended operation $\mathcal{T}\equiv\mathcal{M}\circ\mathbb T$, where $\mathbb{T}$ maps the time parameter $t$ to $-t$ in Eqs.\eqref{eq:2.2}. The system verifies TRI under $\mathcal{M}$ if the application of $\mathcal{T}$ to the equations of motion leaves them unchanged. Note that
%how the application of the map $\mathbb{T}$ to Eq.\eqref{eq:2.2} breaks the structure of the equation of motion; in fact, to invert only the time parameter is \textit{never} sufficient to yield TRI in Hamiltonian mechanics. Moreover,
the concept of time reversal defined by Eq.\eqref{TRIdef} concerns trajectories in phase space.
%since $\mathcal{M}$ acts on $\Omega$, and so trajectories, by definition.\\
%We thus introduce the following notions:
\begin{defi}
\label{sym}
A matrix operator $P$ is symplectic or, respectively,
antisymplectic on a $2n$-dimensional phase space $\Omega$ if
\begin{equation}
P^T\omega P=\omega \quad \mbox{\rm or } ~~
P^T\omega P=-\omega
\end{equation}
where $\omega$ is the antisymmetric matrix
\begin{equation}
\label{anti}
\omega=
\begin{pmatrix}
0&-I_{n}\\
I_n&0
\end{pmatrix}
\end{equation}
and $I_n$ is the $n\times n$ identity matrix.
\end{defi}
\noindent
This can now be used to define the time reversal transformations of Hamiltonian systems.
\begin{defi}
\label{defi:1.1.2}
A time reversal transformation $\mathcal{M}$, for a Hamiltonian system, is a linear operator acting on the space $\Omega$ in such a way that:
\begin{itemize}
\item It is an involution, that is $\mathcal{M}^2=I$.
\item It is an antisymplectic linear operator.
\end{itemize}
\end{defi}
As in Ref.\cite{carbone2020necessary}, we begin considering transformations that separately act on the $6$-dimensional subspaces of $\Omega$ that concern single particles. In this case, a time reversal transformation on
$\Omega$ can be written as:
\begin{equation}
\label{eq:2.3}
\mathcal{M}(x,y,z,p_x,p_y,p_z)=P(s_1x,s_2y,s_3z,-s_1p_x,-s_2p_y,-s_3p_z)
\end{equation}
where $s_i=\pm1$, and $P$ is a permutation obeying $P^2=I$, which acts in the same way on coordinates and on the corresponding momenta. Our first result is the following theorem.
\begin{theo}
\label{theo:1.1.2}
Consider a system of $N$ particles subject to an external magnetic field, described by the Hamiltonian
\begin{equation}
\label{hamii}
H=\sum_{i=1}^N\left[\frac{(\boldsymbol{p}_i-q_i\boldsymbol{A}(\boldsymbol x_i))^2}{2m_i}\right]
\end{equation}
The matrix representation of the antisymplectic operator $\mathcal{M}$ representing a time reversal operation for such a system is block diagonal on the phase space $\Omega$ and takes the form:
\begin{equation}
\label{1.1.30}
\mathcal{M}=
\begin{pmatrix}
A&0\\
0&-A\\
\end{pmatrix}
\end{equation}
with $M=3N$, $A\in O(M)$ and $A^2=I$.
\end{theo}
\begin{proof}
We start from a generic matrix $\mathcal{M}\in GL(2M,\mathbb{R})$, {\em i.e.}
\begin{equation}
\label{1.1.31}
\mathcal{M}=
\begin{pmatrix}
A&B\\
C&D
\end{pmatrix}
\end{equation}
where $A$, $B$, $C$ and $D$ are $M\times M$ blocks. As stressed in Refs.\cite{rondoe,carbone2020necessary}, a time reversal operation defined as in Definition \ref{defi:1.1.2} yields TRI if and only if
\begin{equation}
H(\mathcal{M}\Gamma)=H(\Gamma) \, , \quad \forall \, \Gamma \in \Omega
\end{equation}
Thus, the Hamiltonian \eqref{hamii} requires $\mathcal{M}$ to not mix coordinates and momenta. In the case it does, the minimal coupling term $(\boldsymbol{p}_i-q_i\boldsymbol{A}(\boldsymbol x_i))^2$ is not preserved by the operation. In formulae, the matrix must be block diagonal:
\begin{equation}
\label{1.1.32bis}
\mathcal{M}=
\begin{pmatrix}
A&0\\
0&D
\end{pmatrix}
\end{equation}
As $\mathcal {M}$ must preserve the norm in the $2N$-dimensional phase space, because of the presence of the square in the minimal coupling term, $\mathcal{M}$ must be orthogonal and, consequently one has $A\in O(M)$ and $D\in O(M)$, with $M=3N$.
Moreover, the transformation must be antisymplectic by Definition \ref{defi:1.1.2}. Then, we impose the constraint:
\begin{equation}
\label{1.1.33}
\mathcal{M}^T\omega\mathcal{M}=-\omega \, ; \quad \mbox{i.e.}\ ~~
\begin{pmatrix}
A^T&0\\
0&D^T
\end{pmatrix}
\begin{pmatrix}
0&-I_n\\
I_n&0
\end{pmatrix}
\begin{pmatrix}
A&0\\
0&D
\end{pmatrix}
=
\begin{pmatrix}
0&I_n\\
-I_n&0
\end{pmatrix}
\end{equation}
which leads to the constraint $A^TD=-I$. Then,
\begin{equation}
\label{1.1.34bis}
\mathcal{M}=
\begin{pmatrix}
A&0\\
0&-(A^T)^{-1}
\end{pmatrix}
\end{equation}
Finally, because $\mathcal{M}$ is block diagonal, and $\mathcal{M}^2=I$, Eq.\eqref{1.1.32bis} implies $A^2=I$ and $D^2=I$. Ultimately, since $A$ is orthogonal, that is $AA^T=I$, and $A^2=I$, we obtain that $A$ is a symmetric matrix, {\em i.e.} $A=A^T$.
The proof is then complete, observing that
\begin{equation}
\label{1.1.35}
(A^T)^{-1}=A^{-1}=A
\end{equation}
%in Eq.\eqref{1.1.34bis} we derive the thesis of the theorem.
\end{proof}
This means that the number of time reversal operations is the number of matrices $A$ that are orthogonal and involutory. In particular, let us consider $3N \times 3N$ matrices whose entries are $0$ or $1$, called binary matrices (more general matrices will be considered later, in Section \ref{sec:compatib}). Among binary matrices we have the following:
\begin{defi}
A permutation matrix is a matrix obtained permuting the row of an identity matrix of the same dimension.
\end{defi}
\noindent
We are now going to find the time reversal matrices using some results of Group Theory \cite{tung1985group}. First, note the following proposition.
\begin{prop}
\label{prop:2.7}
The involutory binary matrices $A$ are in one-to-one correspondence with the linear representations of the cyclic group $\mathbb{Z}_2$ over a $3N$-dimensional vector space.
\end{prop}
\begin{proof}
By definition, the group $\mathbb{Z}_2$ is the set $\{e,a\}$ with $e$ the group identity and $a^2=e$. A linear representation $\mathcal{R}(a)$ on a $3N$-dimensional vector space is a $3N\times3N$ matrix such that $\mathcal{R}(a)^2=\mathcal{R}(e)=I_{3N}$.
%In fact, this is an involutory condition.
Moreover, $S_2$, the symmetric group on $2$ elements, coincides with $\mathbb{Z}_2$ and, in general, a representation of an element of $S_k$ must be a permutation, hence an orthogonal, matrix. Then, a matrix $\mathcal{R}(a)$ has the required property, and the possible $A$ and the $\mathcal{R}(a)$ are in one-to-one correspondence.
\end{proof}
Considering that one such matrix $A$ is a permutation, it is a representation of an element of $S_{3N}$, like $\mathcal{R}(a)$. Now, let us introduce the cycle decomposition of a permutation of $S_{3N}$, denoted by:
\begin{equation}
\label{cycle}
\underbrace{(\cdot)...(\cdot)}_{r_1}\underbrace{(\cdot\cdot)...(\cdot\cdot)}_{r_2}...\underbrace{(\cdot...\cdot)}_{r_{3N}}
\end{equation}
which is invariant under group conjugation, and allows us to uniquely label each conjugation class with a set of integers $\{r_1,...,r_{3N}\}$ satisfying the following constraint:
\begin{equation}
\label{1.1.49}
\sum_{l=1}^{3N} lr_l=3N
\end{equation}
There is a useful way to represent the conjugation classes:
\begin{defi}[Young tableaux]
\label{defi:1.1.6}
There is a one-to-one correspondence between a conjugation class $\{r_l\}$ of the group $S_k$ and the following graphical representations known as Young tableau,
\begin{equation}
\label{1.1.52}
\ytableausetup
{mathmode, boxsize=2em}
\begin{ytableau}
\scriptstyle 1 & \scriptstyle 2 & \scriptstyle 3 & \none[\dots]
& \scriptstyle a_1- 1 & \scriptstyle a_1 \\
\scriptstyle 1 & \scriptstyle 2& \none[\dots]& \none[\dots]
& \scriptstyle a_{2} \\
\none[\vdots] & \none[\vdots]
& \none[\vdots] \\
\scriptstyle 1 & \scriptstyle a_{k-1} \\
\scriptstyle a_k
\end{ytableau}
\end{equation}
where the number of squares $a_l$ in each of the $k$ rows obeys the following rules:
\begin{enumerate}
\item $a_{l+1}\leq a_{l}\;\;\;\forall l=1,...,k$.
\item The total number of squares in the tableau equals $k$: $\sum_{l=1}^k a_l=k$
\end{enumerate}
Then, the one-to-one correspondence between the class $\{r_l\}$ and the corresponding Young tableau is given by:
\begin{equation}
\label{1.1.53}
\begin{cases}
r_l=a_l-a_{l+1}\; , \;\;\;\forall \, l=1,...,k-1\\
r_k=a_k
\end{cases}
\end{equation}
\end{defi}
\noindent
For instance, $S_3$ can be associated with any of the three following Young tableaux:
\begin{equation}
\yng(3) \qquad \qquad \;\;\;\;\yng(2,1)\;\;\;\; \qquad \qquad \yng(1,1,1)
\end{equation}
that respectively correspond to three sets $\{r_l\}$: $\{3,0,0\}$, $\{1,1,0\}$ and $\{0,0,1\}$. Incidentally, this also constitutes one method for identifying the partitions of a fixed integer $n$ (see Cap. 5.2 in Ref.\cite{tung1985group}).
Now, the constraint \eqref{1.1.49} takes an interesting form, thanks to the following Corollary.
\begin{cor}
\label{cor:2.1}
The conjugation classes $\{r_i\}$ of $S_{3N}$ containing elements represented by involutory binary matrices obey
\begin{equation}
\label{constra}
r_l=0\;\;\;\;\; \forall l\geq 3
\end{equation}
\end{cor}
\begin{proof}
This is a consequence of Proposition \ref{prop:2.7}: the conjugation class represented by $A$ contains elements $a$ such that $a^2=e$, that is its cycle decomposition can be made only of cycles of order $1$ or $2$.
\end{proof}
\noindent
To compute the number $|\{r_i\}|$ of elements in a conjugation class $\{r_i\}$, we use the following known fact \cite{tung1985group}:
\begin{theo}
The number of elements in the conjugation class $\{r_l\}$ is
\begin{equation}
\label{1.1.51}
|\{r_l\}|=\frac{k!}{\prod_{i=1}^k i^{r_i}r_i!}
\end{equation}
\end{theo}
\noindent
which leads us the fundamental result of this Section:
\begin{cor}[Number of generalized time reversal operations]
\label{cor:1.1.1}
Setting $M=3N$,
the maximum number of generalized time reversal operations involving binary matrices, for a system of $N$ particles described by the Hamiltonian \eqref{hamii}, is given by:
\begin{equation}
\label{1.1.56}
\Tilde{\Delta}_{even}(M)=\sum_{r_2=0}^{M/2}\frac{M!\, 2^{M-2r_2}}{(M-2r_2)!\,r_2!}
\end{equation}
if $M$ is even and by
\begin{equation}
\label{1.1.57}
\Tilde{\Delta}_{odd}(M)=\sum_{r_2=0}^{(M-1)/2}\frac{M!\, 2^{M-2r_2}}{(M-2r_2)!\,r_2!}
\end{equation}
if $M$ is odd.
\end{cor}
\begin{proof}
The total number of time reversal operations that satisfy Corollary \ref{cor:2.1} is the sum of the number of elements in the conjugation classes, {\em i.e.}
\begin{equation}
\label{1.1.58}
\tilde{\Delta}\equiv \sum_{\{r_l\}}|\{r_l\}|
\end{equation}
under the condition \eqref{constra} which, together with \eqref{1.1.49}, yields:
\begin{equation}
\label{1.1.59}
r_1+2r_2=M
\end{equation}
Then, using \eqref{1.1.51} and \eqref{1.1.59} in \eqref{1.1.58} we obtain
\begin{equation}
\label{1.1.60}
\Delta_{even}(M)=\sum_{r_2=0}^{M/2}\frac{M!}{(M-2r_2)!r_2!\,2^{r_2}}
\end{equation}
for $M$ even, while
\begin{equation}
\label{1.1.61}
\Delta_{odd}(M)=\sum_{r_2=0}^{(M-1)/2}\frac{M!}{(M-2r_2)!r_2!\,2^{r_2}}
\end{equation}
for $M$ odd, where the sum stops at $(M-1)/2$, because of the constraint in Eq.\eqref{1.1.59}.
Now, consider for instance one conjugation class $\{r_l\}$, hence a single addend of the summation in these formulae. Each cycle in a decomposition like \eqref{cycle} is defined up to a minus sign. For example, take the matrix representation
\begin{equation}
\begin{pmatrix}
0&s_p&0\\
s_P&0&0\\
0&0&s_3
\end{pmatrix}
\end{equation}
associated to the time reversal operation that acts as a block diagonal on each particle subspace as
\begin{equation}
\mathcal{M}(x,y,z,p_x,p_y,p_z)=(s_Py,s_Px,s_3z,-s_Pp_y,-s_Pp_x,-s_3p_z)
\end{equation}
That corresponds to a permutation belonging to the conjugation class $\{r_l\}=\{1,1,0\}$ of $S_3$: it is evident as the $2\times2$ block that corresponds to the cycle of order $2$ carries a sign $s_P$, and similarly the cycle of order $1$ for $s_3$. In general, we can multiply by $\pm 1$ each cycle block in the representative matrix. This proves that we have to multiply each addend of the summations in \eqref{1.1.60} and \eqref{1.1.61} by a factor $2^{r_1+r_2}$, obtaining:
\begin{equation}
\label{1.1.62}
\tilde{\Delta}_{even}(M)=\sum_{r_2=0}^{M/2}\frac{M!\,2^{r_1+r_2}}{(M-2r_2)!r_2!\,2^{r_2}}
\end{equation}
\begin{equation}
\label{1.1.63}
\tilde{\Delta}_{odd}(M)=\sum_{r_2=0}^{(M-1)/2}\frac{M!\,2^{r_1+r_2}}{(M-2r_2)!r_2!\,2^{r_2}}
\end{equation}
Finally, recalling relation \eqref{1.1.59} we get the two formulae \eqref{1.1.56} and \eqref{1.1.57}.
\end{proof}
\noindent
To illustrate this result, let us apply it to the $6$-dimensional subspace of a system with $N=1$. Here, $M=3$ so we use \eqref{1.1.57} and we obtain:
\begin{equation}
\label{1.1.64}
\Tilde{\Delta}_{odd}(3)=\sum_{r_2=0}^{1}\frac{M!\, 2^{M-2r_2}}{(M-2r_2)!\,r_2!}=\frac{3!\, 2^{3}}{3!\,0!}+\frac{3!\, 2^{1}}{1!\,1!}=8+12=20
\end{equation}
Correctly, this result coincides with the one obtained in Ref.\cite{carbone2020necessary} for single particle subspaces of $N$ particle systems. This kind of operations swap different coordinates and momenta of a given particle. For example, they include the time reversal transformation
\begin{equation}
\label{nondiag}
({x}_1,...,{x}_j,{x}_{j+1},...,{x}_M,{p}_1,....,{p}_j,{p}_{j+1},...,{p}_M)\xrightarrow{\mathcal{M}_{nd}}({x}_1,...,{x}_{j+1},{x}_j,...,{x}_M,-{p}_1,....,-{p}_{j+1},-{p}_j,...,-{p}_M)
\end{equation}
that permutes coordinates $x_j$ and $x_{j+1}$ of a given particle. In fact, this operation corresponds to a matrix $A$ of form
\begin{equation}
\begin{pmatrix}
I_{j-1}&0&0\\
0&B&0\\
0&0&I_{M-j-1}
\end{pmatrix}
\end{equation}
where $I_k$ is the $k\times k$ identity matrix and $B$ is a $2\times 2$ block such as
\begin{equation}
B=
\begin{pmatrix}
0&1\\
1&0
\end{pmatrix}
\end{equation}
This matrix is the representation of a permutation that belongs to the conjugation class $\{M-2,1\}$, that is the permutations composed by $M-2$ cycles of order $1$ and one cycle of order $2$.
As last remark, the generalization of the concept of time reversal leads to a notable outcome on the rule of transformation of other classical observables, as for example angular momentum. Let us recall its definition:
\begin{defi}
For a classical Hamiltonian system, the total angular momentum is defined by
\begin{equation}
\label{2.62}
\boldsymbol L=\sum_{i=1}^{N}\boldsymbol x^i\times\boldsymbol p_i
\end{equation}
\end{defi}
\noindent
The canonical time reversal operation $(\boldsymbol X,\boldsymbol P) \mapsto (\boldsymbol X,-\boldsymbol P)$ trivially transforms $\boldsymbol L$ into $-\boldsymbol L$. The same holds for the $20$ time reversal operations that do not permute coordinates of different particles. Restricting to the single particle subspaces $(\boldsymbol x^i,\boldsymbol p_i)$, and denoting the transformation rule by:
\begin{equation}
(\boldsymbol x^i,\boldsymbol p_i)\to(R\boldsymbol x^i,-R\boldsymbol p_i)
\end{equation}
where Theorem \ref{theo:1.1.2} implies $R\in O(3)$, we obtain:
\begin{equation}
\boldsymbol L \mapsto \boldsymbol L'=-\sum_{i=1}^{N}R\boldsymbol x^i\times R\boldsymbol p_i=-\boldsymbol L
\end{equation}
where we used the fact that the cross product is unchanged by coherent rotation of both factors.
On the other hand, transformations with matrix representation like \eqref{1.1.30}, where $A$ does not separately act on each particle subspace, do not reverse $\boldsymbol L$. Consider, for example, the following generalized time reversal operation:
\begin{equation}
\label{1.1.39}
({x}_1,...,{x}_j,{x}_{j+1},...,{x}_{3N},{p}_1,...,{p}_j,{p}_{j+1},...,{p}_{3N})\xrightarrow{\mathcal{M}_{nd}}({x}_{3N},...,{x}_{j},{x}_{j+1},...,{x}_1,-{p}_{3N},...,-{p}_{j},-{p}_{j+1},...,-{p}_1)
\end{equation}
which swaps $x_1$ and $x_{3N}$ and coherently acts on momenta as prescribed by Theorem \ref{theo:1.1.2}. Then, consider the $z$ component of $\boldsymbol L$. By definition of cross product
we have:
\begin{equation}
L_z=x^1p_2-x^2p_1+...+x^{3N-2}p_{3N-1}-x^{3N-1}p_{3N-2}
\end{equation}
which becomes
\begin{equation}
L_z=-x^{3N}p_2+x^2p_{3N}+...-x^{3N-2}p_{3N-1}+x^{3N-1}p_{3N-2}\neq -L_z
\end{equation}
under the application of $\mathcal{M}_{nd}$. The same holds for $L_x$ and $L_y$. The conclusion is the following:
\begin{prop}
\label{prop:2.6}
There exist generalized time reversal operations that do not reverse the sign of the classical mechanics angular momentum.
\end{prop}
\noindent
Note that the time reversal operation $\mathcal{M}_{nd}$ works, in particular for a system of free equal mass particles, whose Hamiltonian is expressed by $H=\sum_{i=1}^{3N} p_i^2/2m$.
We now extend to Quantum Mechanics, the present reasoning. Indeed, in Ref.\cite{rondoquantu}, only $8$ time reversal operations had been identified, that classically take the form
\begin{equation}
\mathcal{T}(x,y,z,p_x,p_y,p_z,t)=(s_1x,s_2y,s_3z,-s_1p_x,-s_2p_y,-s_3p_z,-t)
\end{equation}
\subsection{The time reversal operator}
\label{sec:The time reversal operator}
Let us begin with spinless non-relativistic particles. The fundamental axiom is that the state of the system is represented by a wave function, that obeys the Schr\"odinger equation:
\begin{equation}
\label{eq:2.1}
i\frac{\partial\psi(x,y,z,t)}{\partial t}=H\psi(x,y,z,t)
\end{equation}
where our units imply $\hslash=1$.
% \begin{rem}
% \label{rem:1.2.4}
As in Classical Mechanics, we have an evolution equation that is invariant under time reversal. The corresponding Hilbert space is the quantum counterpart of the classical phase space. Then, we follow Wigner's approach to define time reversal transformations, cf.\ Refs.\cite{wigner1932operation,sachs}. In the following we will consider operators and their matrix representations; for the sake of simplicity,
%and to parallel the classical situation,
the same symbol will be used.
\begin{defi}[Spinless particles]
\label{prop:1.2.1}
An operator on the Hilbert space of the wave functions of an $N$ particle system is called a time reversal operator $\mathcal{T}$ if it obeys:
\begin{itemize}
\item $\mathcal{T}=UK$, with $U$ a unitary operator and $K$ the complex conjugation operator, \textit{i.e.} it is antilinear
\item It is an involution in the quantum sense, that is $\mathcal{T}^2=\pm I$, where $I$ is the identity operator
\item It is kinematically admissible, that is it preserve the canonical commutation relations $[x_n,p_m]=i\delta_{nm}$
\end{itemize}
\end{defi}
% \begin{proof}
% We start applying $\mathcal{T}$ to both side of the equation of motion, obtaining:
% \begin{equation}
% \label{1.77}
% \mathcal{T}i\mathcal{T}^{-1}\mathcal{T} {\frac {\partial}{\partial t}}\mathcal{T}^{-1} \mathcal{T}\left|\psi (t)\right\rangle =\mathcal{T}H\mathcal{T}^{-1}\mathcal{T}\left|\psi (t)\right\rangle
% \end{equation}
% where we inserted identity operators $\mathcal{T}^{-1}\mathcal{T}=I$ in the middle. By Remark \ref{rem:1.2.4}, $\mathcal{T}$ is the quantum counterpart of the antisymplectic classical operator $\mathcal{M}$, {\em i.e.} it does not act directly on time:
% \begin{equation}
% \label{1.78}
% \mathcal{T}i\mathcal{T}^{-1} {\frac {\partial}{\partial t}} \mathcal{T}\left|\psi (t)\right\rangle =\mathcal{T}H\mathcal{T}^{-1}\mathcal{T}\left|\psi (t)\right\rangle
% \end{equation}
% Now, we define
% $\ket{\psi'(t')}\equiv\mathcal{T}\ket{\psi(t)}$; this is a completely general expression since we are supposing that $\mathcal{T}$ acts on states. Hence, we obtain:
% \begin{equation}
% \label{1.79}
% \mathcal{T}i\mathcal{T}^{-1} {\frac {\partial}{\partial t}} \left|\psi' (t')\right\rangle =\mathcal{T}H\mathcal{T}^{-1}\left|\psi' (t')\right\rangle
% \end{equation}
% Now, to study time reversal invariance we consider the case $t'=-t$; then, we can manipulate \eqref{1.79} obtaining:
% \begin{equation}
% \label{1.80}
% -\mathcal{T}i\mathcal{T}^{-1} {\frac {\partial}{\partial t'}} \left|\psi' (t')\right\rangle =\mathcal{T}H\mathcal{T}^{-1}\left|\psi' (t')\right\rangle
% \end{equation}
% Imposing time reversal invariance of the dynamic is equivalent to impose that \eqref{1.80} is equal to the Scr\"odinger equation \eqref{eq:2.1} for $\ket{\psi'(t')}$: we have then two possibilities:
% \begin{equation}
% \label{1.81}
% \begin{cases}
% \mathcal{T}i\mathcal{T}^{-1}=-i\\
% \mathcal{T}H\mathcal{T}^{-1}=H
% \end{cases}
% \end{equation}
% or
% \begin{equation}
% \label{1.82}
% \begin{cases}
% \mathcal{T}i\mathcal{T}^{-1}=i\\
% \mathcal{T}H\mathcal{T}^{-1}=-H
% \end{cases}
% \end{equation}
% The case \eqref{1.82} is clearly problematic: in fact, the systems we are considering in the following always present the kinetic term $p^2/2m$ in the Hamiltonian, so we should transform $p$ in $\pm ip$ in order to verify \eqref{1.82}, leading to the appearance of imaginary momenta. In a completely generic setup, \eqref{1.82} means that for each state with energy $E$ we have a counterpart with energy $-E$ in the Hilbert space, that is an infinite negative spectrum, a clearly nonphysical situation.\\
% Then, considering the remaining case \eqref{1.81}, the first constraint implies that $\mathcal{T}$ must contain the complex conjugation operator $K$. Moreover, since the dynamics must remain the same by definition, it is necessary that the state $\ket{\psi'(t')}$ belongs to the Hilbert space associated to $H$ as $\ket{\psi(t)}$ does. In other words, we have the freedom to transform the state also by a unitary operator $U$ distinct from the complex conjugation. Using these two results we obtain the statement of the Theorem
% \begin{equation}
% \label{1.83}
% \mathcal{T}=UK
% \end{equation}
% where by definition $KiK^{-1}=-i$, $K^2=I$ and so $K=K^{-1}$.\\
% Coming to the second thesis, we start considering that in Quantum Mechanics the involutory condition writes $\mathcal{T}^2=\eta I$, with $\eta$ a complex number of modulus one, since any state is defined up to a phase by the axioms of Quantum Mechanics. Now, we find that:
% \begin{equation}
% \label{1.84}
% \mathcal{T}^2=UKUK=UU^*
% \end{equation}
% where the star stands for the complex conjugation. Thus, we use the fact that $U$ is unitary and so that $U^{-1}=U^{\dagger}=(U^T)^*$ and then that $U^*=(U^T)^{-1}$. Using the latter in \eqref{1.84} we obtain:
% \begin{equation}
% \label{1.85}
% \mathcal{T}^2=U(U^T)^{-1}=\eta I
% \end{equation}
% and then
% \begin{equation}
% \label{1.86}
% U=\eta U^T
% \end{equation}
% Thus, to prove the thesis is sufficient to transpose this equation: we have $U^T=\eta U$, that reinserted in \eqref{1.86} yields to $U=\eta^2 U$, that is $\eta=\pm 1$.
% \end{proof}
\noindent
To proceed analogously to the classical case, we exploit the following result \cite{anderson1994canonical}:
\begin{prop}
\label{prop:1.2.3}
The matrix representation of a time reversal operator $\mathcal{T}$ %=UK$
for a system of $N$ particles belongs to %$GL(6N,\mathbb{C})$ and, in particular, to
$Sp(6N, \mathbb{C})$.
\end{prop}
\noindent
For instance, in the case of a system with only two coordinates, the operators $x_1$ and $x_2$, and of a time reversal operator such that:
\begin{equation}
\begin{split}
\mathcal{T}x_1\mathcal{T}^{-1}=x_2\\
\mathcal{T}x_2\mathcal{T}^{-1}=x_1\\
\end{split}
\end{equation}
the matrix representation of $\mathcal{T}$ in terms of coordinates is given by:
\begin{equation}
\begin{pmatrix}
0&1\\
1&0
\end{pmatrix}
\end{equation}
The Stone-Von Neumann Theorem \cite{von1931uniqueness} now states that in coordinates representation, {\em i.e.} for a wave function defined as coordinate dependent, the operators $x^i$ and $p_i$ uniquely act as:
\begin{equation}
\label{1.89}
\begin{dcases}
x^i\psi(x_0)=x_0^i\psi(x_0)\\
p_i\psi(x_0)=-i\frac{\partial \psi}{\partial x^i}(x_0)
\end{dcases}
\end{equation}
\noindent
One can equivalently define the action of coordinate and momentum operators in momentum space, via Fourier transform.
\begin{prop}
\label{prop:2.3b}
The complex conjugation preserves the canonical commutation relation.
\end{prop}
\begin{proof}
Using Eq. \eqref{1.89}, we can obtain the rule of transformation of the coordinate and momentum operators:
\begin{equation}
\label{2.42}
K p_i K^{-1}=-KiK^{-1}\dfrac{\partial }{\partial x_i}=i \dfrac{\partial }{\partial x^i}=-p_i
\end{equation}
and
\begin{equation}
\label{2.43}
Kx^iK=x^i
\end{equation}
The proof of the canonical nature of $K$ is then trivial, since applying $K$ to both side of the canonical commutation relations
\begin{equation}
\label{1.100c}
\left[x^i,p_j\right]=i \delta^i_j
\end{equation}
one obtains
\begin{equation}
\label{1.100b}
\left[Kx^iK,Kp_jK\right]=KiK \delta^i_j
\end{equation}
that is equivalent to \eqref{1.100c} by definition of $K$.
Then, the $6N\times6N$ matrix representing $K$ on the $6N$-dimensional space of symbolic vectors $(\boldsymbol X,\boldsymbol P)$, is given by:
\begin{equation}
\label{krepre}
K=
\begin{pmatrix}
I&0\\
0&-I
\end{pmatrix}
\end{equation}
which is trivially antisymplectic, and acts
separately on coordinate and momentum operators
%, not bounded by \eqref{1.89}, and then the complex conjugation that acts on all numbers, also the matrix components.
\end{proof}
Counting the generalized time reversal operations is now reduced to finding alternatives to the matrix representation of $K$. Table \ref{tab:my_label} illustrates the parallel with Classical Mechanics.
\begin{table}[h]
\centering
\begin{tabular}{c|c|c}
& Classical Mechanics & Quantum Mechanics\\
& & \\
Time Reversal & $\mathcal{M}$ &$\mathcal{T}=UK$\\
& & \\
Commutation Relations & $\{x^i,p_j\}=\delta^i_j$ &$\left[x^i,p_j\right]=i \delta^i_j$\\
\end{tabular}
\caption{Comparative scheme}
\label{tab:my_label}
\end{table}
\noindent
Recall that a fundamental aspect of Classical mechanics is that $\mathbb{T}$ acts on the Hamiltonian structure as an antisymplectic operator, and then $\mathcal{M}$ is also antisymplectic in order to obtain a symplectic transformation. Now, Proposition \ref{prop:1.2.3} says that in Quantum Mechanics $\mathcal{T}$ is represented by a symplectic matrix. Furthermore, $K$ is represented by an antisymplectic matrix, that is it plays the role of $\mathbb{T}$.
% This is not a contradiction with Proposition \ref{prop:2.3}, because the complex conjugation is not a linear operation: it is anti-linear. It can be represented by a matrix, with its action defined by \eqref{2.42} and
% \eqref{2.43}.,
So, the situation is very similar to
the classical one, and $U$ is antisymplectic.
\begin{theo}
\label{theo:1.2.1}
Take the $3N$-dimensional vectors $\boldsymbol X$ and $\boldsymbol P$ of respectively the coordinate and momentum operators of a quantum system made of $N$ particles. Given a time reversal operator $\mathcal{T}=UK$, the matrix representation of $U$ on the $6N$-vectors $(\boldsymbol X,\boldsymbol P)$ takes the form
\begin{equation}
\label{1.102}
U=
\begin{pmatrix}
A&0\\
0&-A
\end{pmatrix}
\end{equation}
where $A$ is a symmetric or an antisymmetric matrix.
\end{theo}
\begin{proof}
As in Theorem \ref{theo:1.1.2}, $\mathcal{T}$ cannot swap coordinates and momenta of different particles, because of the minimal coupling term in the Hamiltonian. Then, the matrix representation of $U$ takes the form
\begin{equation}
U=
\begin{pmatrix}
A&0\\
0&B
\end{pmatrix}
\end{equation}
As in the proof of Theorem \ref{theo:1.1.2}, we obtain $A^\dagger B=-I$. Then, we can write:
\begin{equation}
U=
\begin{pmatrix}
A&0\\
0&-(A^\dagger)^{-1}
\end{pmatrix}
\end{equation}
where we used the fact that the antisymplectic condition for $Sp(6N,\mathbb{C})$ is expressed by:
\begin{equation}
U^\dagger \omega U=-\omega
\end{equation} and involves the dagger operation instead of the transpose. Now, $U$ is unitary by Definition \ref{prop:1.2.1}, and so we have $A=(A^\dagger)^{-1}$, which leads to Eq. \eqref{1.102}.
To prove that $A$ is symmetric or antisymmetric, use the involutory property $\mathcal{T}^{2}=UKUK=UU^*=\pm I$, which implies $AA^*=\pm I$. Moreover, $U$ is unitary, hence $AA^\dagger=I$. Then, for $AA^* = I$
we obtain $A^T=A$, {\em i.e.} $A$ symmetric, whilst for $AA^*=- I$ we have $A^T=-A$, and is $A$ antisymmetric.
\end{proof}
Note that $K$ is of the form \eqref{1.102}, and corresponds to the the canonical time reversal operation, that preserves coordinates and reverses momenta. %Obviously, since $K$ preserves the commutation relations by itself, we could have reduced from the beginning to study the form of $U$, that is unitary, linear and so, by Proposition \ref{prop:1.2.3}, symplectic.
%By the way, we chose to analyze the issue using \eqref{krepre} and antisymplectic matrix to trace a parallel with the classical case.
%Although this definition seems to achieve the former purpose, there is the issue that $A$ can be complex.
If we restrict to the part of $U$ acting on the coordinate (and then separately on momentum) operators, we find that a complex transformation of $\boldsymbol X$ and $\boldsymbol P$ cannot be used. Suppose that
\begin{equation}
\mathcal{T}\boldsymbol P\mathcal{T}^{-1}=P'=\mathfrak{Re}(\boldsymbol P')+\mathfrak{Im}(\boldsymbol P')i
\end{equation}
and consider a system of free particles with Hamiltonian $H=\boldsymbol P^2/2m$. Applying the time reversal operator, we have:
\begin{equation}
H'=\mathcal{T}H\mathcal{T}^{-1}=\frac{(\boldsymbol P')^2}{2m}=\frac{\mathfrak{Re}(\boldsymbol P')^2+2\mathfrak{Im}(\boldsymbol P')\mathfrak{Re}(\boldsymbol P')i-\mathfrak{Im}(\boldsymbol P')^2}{2m}
\end{equation}
where the Hamiltonian is preserved only if
$2\mathfrak{Im}(\boldsymbol P')\mathfrak{Re}(\boldsymbol P')=0$, since the $H$ is real in $\boldsymbol X$ and $\boldsymbol P$. % ({\em n.b.} we are working on matrix representation, so we have not to consider that $\boldsymbol P$ is complex as differential operator).
But if $\mathfrak{Im}(\boldsymbol P')\ne0$, as
assumed, that requires $\mathfrak{Re}(\boldsymbol P')=0$. But then $H'<0$ should equal $H>0$, which is absurd. This reasoning can be extended to the case of an external magnetic field. Because $H\propto \sum_{i=0}^N (\boldsymbol p_i-q\boldsymbol A(\boldsymbol x_i))^2$, there is again an addend proportional to $\boldsymbol P^2$ and the proof applies. Therefore, we may adopt the following:
\begin{assu}
The unitary operation $U$ associated to a time reversal operator $\mathcal{T}=UK$ acting on the coordinate (and momentum) operators is real.
\end{assu}
\noindent
This assumption immediately leads to:
\begin{cor}
If $A$ is real and symmetric, it belongs to $O(3N)$, as in Classical Mechanics.
\end{cor}
\begin{proof}
In the real case,
the constraints $AA^\dagger=I$ and $A=A^T$ imply $AA^T=I$, which means that $A\in O(3N)$.
\end{proof}
\noindent
We can then straightforwardly repeat the arguments developed for the classical case, starting from Proposition \ref{prop:2.7}.
In particular, we consider permutation matrices also in this quantum framework.
Therefore, the generalized time reversal operators can be counted using Corollary \ref{cor:1.1.1}. Moreover, let $A$ be real but antisymmetric. In this case, Theorem \ref{theo:1.2.1} requires $A^2=-I$.
Then, consider the following definition.
\begin{defi}
Let $\mathbb{Z}_4$ be a finite cyclic group with generator $a$ ($a^4$ is the group identity $e$), and with the real $2\times 2$ matrix representation given by:
\begin{equation}
\mathcal{R}_4(e)=I\quad\quad \mathcal{R}_4(a)=\begin{pmatrix}
0&-1\\
1&0
\end{pmatrix}
\quad\quad \mathcal{R}_4(a^2)=-I\quad\quad \mathcal{R}_4(a^3)=\begin{pmatrix}
0&1\\
-1&0
\end{pmatrix}
\end{equation}
%where the subscript $4$ distinguishes this representation from the representative
%of the non trivial element of $\mathbb{Z}_2$:
%\begin{equation}
% \mathcal{R}(a)=\begin{pmatrix}
% 0&1\\
% 1&0
% \end{pmatrix}
%\end{equation}
\end{defi}
\noindent
Then, the following holds.
\begin{prop}
\label{prop:1.2.6}
If the number of particles $N$ is odd, $A$ is not real and antisymmetric. If $N$ is even, the number of time reversal operators $\tilde{\Delta}$ (that is the number of possible $A$'s) equals
\begin{equation}
\label{1.108}
\tilde{\Delta}=\frac{M!}{(M/2)!} \, \quad \mbox{with } ~~ M = 3N
\end{equation}
\end{prop}
\begin{proof}
The condition $A^2=-I$ means that $A\in GL(3N,\mathbb(R))$ is a linear representation of the element $a$ or $a^3$ of the group $\mathbb{Z}_4$ over a $3N$-dimensional vector space. Moreover, $A^T=-A$ implies that it is antisymmetric. Then, $A$ must contain only $2\times 2$ blocks, like $\mathcal{R}_4(a)$ and $\mathcal{R}_4(a^3)$, with vanishing diagonal elements. So, for odd $N$, the $3N \times 3N$ matrix cannot be built using $2\times 2$ blocks.
In the case of $N$ even, it suffices to use Eq.\eqref{1.1.51} with $k=M$ and $r_i=0$, for $i\neq 2$. Indeed, Corollary \ref{cor:2.1} holds, and antisymmetry implies that the diagonal elements vanish, excluding cycles of order $1$. Moreover, it is trivial to make a one-to-one correspondence between $\mathcal{R}_4(a)$ and the representative of the non trivial element of $\mathbb{Z}_2$:
\begin{equation}
\mathcal{R}(a)=
\begin{pmatrix}
0&1\\
1&0
\end{pmatrix}
\end{equation}
Then, one must count the elements of a conjugation class built only with cycles of order $2$, which can be done using Eq.\eqref{1.1.51} and considering that the number of cycles of order $2$ over $M$ elements is $M/2$. The result is:
\begin{equation}
\Delta=\frac{M!}{2^{M/2}(M/2)!}
\end{equation}
Finally, unlike the case of Classical mechanics, in which $\mathcal{M}^2=I$, here we
also admit the opposite sign for the $2\times 2$ block, that corresponds to $\mathcal{R}_4(a^3)$. Multiplying $\Delta$ by $2^{M/2}$, we finally get Eq.\eqref{1.108}.
\end{proof}
\noindent
The difference between Classical and Quantum cases is easily revealed by the following example. The Hamiltonian of a system made of $2$ free quantum particles moving in $1$-dimension is given by $H=p_1^2/2m+p_2^2/2m$. Consider a time reversal operator $\mathcal{T}$ whose unitary part is represented on the vectors of momentum operators $\boldsymbol P=(p_1,p_2)$ by the matrix
\begin{equation}
Q=
\begin{pmatrix}
0&-1\\
1&0
\end{pmatrix}
\end{equation}
While admitted in quantum mechanics, it is not acceptable in classical mechanics, because $Q^2=-I$.
Having exhausted our treatment of spinless particles coupled with an external magnetic field, we now turn to the case of particles with spin $1/2$.
\subsection{Time reversal with spin}
\label{sec:Time reversal with spin}
%The topic of the present section is another example of how the canonical time reversal operation described in the Sec.\ref{sec:Introduction} influenced the conceptualization of time reversal for systems coupled with a magnetic field.
% Let us start from the classical definition of angular momentum:
% \begin{defi}
% For a classical Hamiltonian system, the total angular momentum is defined by
% \begin{equation}
% \label{2.62}
% \boldsymbol L=\sum_{i=1}^{N}\boldsymbol x^i\times\boldsymbol p_i
% \end{equation}
% \end{defi}
% \noindent
% The canonical time reversal operation $(\boldsymbol X,\boldsymbol P) \mapsto (\boldsymbol X,-\boldsymbol P)$ trivially transforms $\boldsymbol L$ into $-\boldsymbol L$. The same holds for the $20$ time reversal operations that do not permute coordinates of different particles. Restricting to the single particle subspaces $(\boldsymbol x^i,\boldsymbol p_i)$, and denoting the transformation rule by:
% \begin{equation}
% (\boldsymbol x^i,\boldsymbol p_i)\to(R\boldsymbol x^i,-R\boldsymbol p_i)
% \end{equation}
% where Theorem \ref{theo:1.1.2} implies $R\in O(3)$, we obtain:
% \begin{equation}
% \boldsymbol L \mapsto \boldsymbol L'=-\sum_{i=1}^{N}R\boldsymbol x^i\times R\boldsymbol p_i=-\boldsymbol L
% \end{equation}
% where we used the fact that the cross product is unchanged by coherent rotation of both factors.
% On the other hand, transformations with matrix representation like \eqref{1.1.30}, where $A$ does not separately act on each particle subspace, do not reverse $\boldsymbol L$. Consider, for example, the following generalized time reversal operation:
% \begin{equation}
% \label{1.1.39}
% ({x}_1,...,{x}_j,{x}_{j+1},...,{x}_{3N},{p}_1,...,{p}_j,{p}_{j+1},...,{p}_{3N})\xrightarrow{\mathcal{M}_{nd}}({x}_{3N},...,{x}_{j},{x}_{j+1},...,{x}_1,-{p}_{3N},...,-{p}_{j},-{p}_{j+1},...,-{p}_1)
% \end{equation}
% which swaps $x_1$ and $x_{3N}$ and coherently acts on momenta as prescribed by Theorem \ref{theo:1.1.2}. Then, consider the $z$ component of $\boldsymbol L$. By definition of cross product
% we have:
% \begin{equation}
% L_z=x^1p_2-x^2p_1+...+x^{3N-2}p_{3N-1}-x^{3N-1}p_{3N-2}
% \end{equation}
% which becomes
% \begin{equation}
% L_z=-x^{3N}p_2+x^2p_{3N}+...-x^{3N-2}p_{3N-1}+x^{3N-1}p_{3N-2}\neq -L_z
% \end{equation}
% under the application of $\mathcal{M}_{nd}$. The same holds for $L_x$ and $L_y$. The conclusion is the following:
% \begin{prop}
% \label{prop:2.6}
% There exist generalized time reversal operations that do not reverse the sign of the classical mechanics angular momentum.
% \end{prop}
% \noindent
% Note that the time reversal operation $\mathcal{M}_{nd}$ works, in particular for a system of free equal mass particles, whose Hamiltonian is expressed by $H=\sum_{i=1}^{3N} p_i^2/2m$.
For non-relativistic spin systems, the equation of motion is the Pauli equation:
\begin{equation}
\label{eq:2.40}
\left\{\sum_{j=1}^N{\frac {1}{2m_j}}[{\boldsymbol {\sigma }_j}\cdot (\boldsymbol {p}_j -q_j\boldsymbol{A}(x_j,y_j,z_j) )^{2} \right\}|\psi \rangle =i {\frac {\partial }{\partial t}}|\psi \rangle
\end{equation}
where $\ket{\psi}$ represents the total state of the system of $N$ particles and the spin operator $\boldsymbol\sigma_j$ of the $j$-th particle is expressed by
\begin{equation}
\label{spacespin}
\boldsymbol{\sigma}_j=I\otimes...\otimes\underbrace{\boldsymbol{\sigma}}_j\otimes...\otimes I
\end{equation}
where $\boldsymbol\sigma=(\sigma_x,\sigma_y,\sigma_z)$ contains the Pauli matrices as Cartesian components and $I$ is the identity operator. The Pauli vector identity:
\begin{equation}
\label{eq:2.41}
(\boldsymbol{\sigma}\cdot \boldsymbol{a} )({\boldsymbol{\sigma}}\cdot \boldsymbol{b} )=\boldsymbol{a} \cdot \boldsymbol{b} +i{\boldsymbol{\sigma }}\cdot(\boldsymbol{a} \times \boldsymbol{b})
\end{equation}
yields
\begin{equation}
\label{eq:2.42}
\sum_{j=1}^N{\frac {1}{2m_j}}\left[(\boldsymbol {p}_j -q_j\boldsymbol {A}(x_j,y_j,z_j) )^{2}-q_j\boldsymbol{\sigma}_j\cdot\boldsymbol{B}\right]|\psi \rangle =i{\frac {\partial }{\partial t}}|\psi \rangle
\end{equation}
where $\boldsymbol{B}=\nabla\times\boldsymbol{A}$.
For the time reversal operator, we may adopt the following form:
\begin{equation}
\mathcal{T}=(U_c\otimes U_s)K
\end{equation}
where $U_c$ denotes the part acting on the coordinate and momentum operators, as in Sec.\ref{sec:The time reversal operator}; while $U_s$ is the part acting on the spin operators. Both $U_c$ and $U_s$ must be unitary, in order to represent time reversal operators. Since $\boldsymbol\sigma_j$ separately acts on single particles spin operators, in the following we omit the subscript $j$, considering, without loss of generality, the effect of $\mathcal{T}$ on $\boldsymbol\sigma$.
In this framework, time reversal has been
traditionally defined as an operator $\mathcal{T}$ that preserves the commutation relations and,
% \vskip 10pt
% MA \`E UN'ASSUNZIONE? NON \`E UN FATTO? \textcolor{blue}{è un assunzione sul Sachs. Il discorso sopra su L serve a mostrare che la vera richiesta fondamentale è quella dell'assunzione 2.3 (che effettivamente è meglio sia definizione). In altre parole, ma mi pare lo avessi scritto (?), la 2.68 è immotivata perchè parte dal presupposto che (A) lo spin sia un momento angolare ergo trasformi come L, ma non ho p e x in \\sigma!! (B) anche fosse, già per L ho dei time reversal che non fanno andare L in -L.
% Questo passaggio è importante (con la I maiuscola) perchè è l'analogo di accettare che non debba per forza ritracciare la stessa traiettoria in MC e quindi che possa avere altre cose oltre (x,-p)}
% \vskip 10pt
analogously to classical mechanics, yields
\cite{sachs}:
\begin{equation}
\label{1.95}
\mathcal{T}\boldsymbol \sigma\mathcal{T}^{-1}=-\boldsymbol \sigma
\end{equation}
\noindent
which directly implies:
\begin{equation}
U_s=\sigma_y
\end{equation}
Unfortunately, this condition does not allow us to treat particles with spin \cite{rondoquantu}. On the other hand, Proposition \ref{prop:2.6} suggests we may relax some restriction. Therefore, we adopt the following, minimal, definition that does not include condition \eqref{1.95} and that actualizes Definition \ref{prop:1.2.1} to particles with spin 1/2. %First of all, assuming that spin transforms as a classical angular momentum, we showed in Proposition \ref{prop:2.6} that $\boldsymbol L$ does not always change sign under time reversal. Secondly, it is well-known that spin shares with angular momentum the same operator algebra, but that is not related to a real rotating motion of the particle in study, \textit{i.e.} we have no "position cross momentum" inside spin. Thus, we decide to seek for a generalized concept of time reversal starting from the following minimal necessary assumption:
%Then, as shown in Sec.\ref{sec:The time reversal operator}, there were found weaker constraints on the form of these transformations, deduced by the fact that, in reality, a time reversal operation is valid if antisymplectic (and unitary in QM), involutional and if it preserves the Hamiltonian structure ({\em n.b.} the canonical commutation relations in QM) and equations of motion. The situation of the canonical time reversal operation represents a more strong constraint that, following the literature, we named \textit{microreversibility}, but that is not necessary for the validity of statistical quantities as Onsager relations.\\
%The situation seems to be the same if we consider the spin space: in fact,
% Assumption \ref{assu:2.3} is not necessary. In the first place, the spin is not associated to a real rotation, and
%this analogy is due to the form of the commutation relation of the Pauli matrices, realization of the $\mathfrak{su}(2)$ algebra and equivalent to the one of the quantum angular momenta.
%In fact,
% the fundamental request is the following:
\begin{defi}[Particles with spin 1/2]
\label{assu:2.4}
An antilinear operator $\cal T$ verifying Definition \ref{prop:1.2.1} is called a time reversal operator for particles with spin, if it preserves the commutation relations of Pauli matrices:
%of the Lie algebra $\mathfrak{su}(2)$:
\begin{equation}
\label{1.114}
[\sigma_{j},\sigma_{k}]=i \tensor{\varepsilon} {_{jk}^{l}}\sigma_{l}
\end{equation}
where $j, k, l = x, y, z$ and $\tensor{\varepsilon} {_{jk}^{l}}$ is the Levi-Civita symbol.
\end{defi}
%This is the fundamental request, analogue to the one done for coordinate and momentum operators.
%To sum up, Assumption \ref{assu:2.3} is purely based on an analogy with a classical quantity, reasoning always a bit dangerous in Quantum Mechanics; moreover, the similarities with the angular momentum are already considered in Assumption \ref{assu:2.4}. This could be sufficient to neglect the constraint \eqref{1.95}.
%In addition, as we stated in Proposition \ref{prop:2.6}, already in the classical framework there are time reversal operations that do not reverse the sign of the classical angular momentum.
%In conclusion Assumption \ref{assu:2.3} is based on a classical, and even partially false, analogy: we consider this a sufficient motivation to relax this presupposition and to study time reversal in spin space from a wider point of view.\\
%First of all we start from Assumption \ref{assu:2.4}: by definition,
\noindent
Note that only $\sigma_y$ is complex and so
%by direct computation,
the complex conjugation operator $K$ preserves \eqref{1.114}. To proceed, let us recall two results of Lie Group Theory \cite{gilmore2012lie},
taking $\mathbb{R}^3$ as a Lie algebra generated by three vectors $(\boldsymbol e_1,\boldsymbol e_2,\boldsymbol e_3)$:
\begin{theo}
\label{theo:2.4}
There exists a linear invertible map $\Phi:\mathfrak{su}(2)\xrightarrow[]{}\mathbb{R}^3$ such that \begin{equation}
\Phi\left(\left[\frac{i}{2}\sigma_j,\frac{i}{2}\sigma_k\right]\right)=\Phi\left(\frac{i}{2}\sigma_j\right)\times\Phi\left(\frac{i}{2}\sigma_k\right)\, ; \quad \mbox{with } ~~
\Phi\left(\frac{i}{2}\sigma_j\right)=\boldsymbol e_j
\end{equation}
where $\times$ (the usual cross product) is the Lie product.
\end{theo}
\begin{theo}
\label{theo:1.2.2}
An element of the unitary group $U\in U(2)$ can be expressed as
\begin{equation}
\label{1.116}
U=e^{\frac{i}{2}\lambda^0 }\left(I\cos{\frac{\lambda}{2}}+i\frac{\lambda^j\sigma_j}{\lambda}\sin{\frac{\lambda}{2}}\right)=e^{\frac{i}{2}\lambda^0 } V
\end{equation}
where $\lambda^j \in \mathbb{R}$ for $j=0,1,2,3$,
$\lambda=|\boldsymbol\lambda|$, for $\boldsymbol\lambda=(\lambda^1,\lambda^2,\lambda^3)$, and $V\in SU(2)$.
\end{theo}
\noindent
Theorem \ref{theo:2.4} allows us to find a $U_s$ that preserves the commutation relations \eqref{1.114}, by finding a transformation that preserves a right handed triad in $\mathbb{R}^3$.
\begin{lemma}
\label{lemma1}
Linear transformations mapping a right handed basis in $\mathbb{R}^3$ into another belong to $SO(3)$.
\end{lemma}
\noindent
Another useful fact is that given $V\in SU(2)$ as in \eqref{1.116}, one has:
\begin{equation}
\label{complex}
\sigma_y U\sigma_y=U^{*}
\end{equation}
where $U^*$ is the complex conjugate of $U$. Then, thanks to the properties of the Pauli matrices, we obtain:
\begin{equation}
\begin{split}
\sigma_y \left(I\cos{\frac{\lambda}{2}}+i\frac{\lambda^j\sigma_j}{\lambda}\sin{\frac{\lambda}{2}}\right)\sigma_y&=I\cos{\frac{\lambda}{2}}+i\frac{\lambda^1\sigma_y\sigma_x\sigma_y+\lambda^2\sigma_y\sigma_y\sigma_y+\lambda^3\sigma_y\sigma_z\sigma_y}{\lambda}\sin{\frac{\lambda}{2}}=\\&=\left(I\cos{\frac{\lambda}{2}}-i\frac{\lambda^j\sigma_j}{\lambda}\sin{\frac{\lambda}{2}}\right)
\end{split}
\end{equation}
Another useful classical result of group theory is the following:
\begin{theo}
\label{iso}
The quotient group $SU(2)/\mathbb{Z}_2$ is isomorphic to the group $SO(3)$. The isomorphism is defined by:
\begin{equation}
U^\dagger \sigma_j U=[\Lambda(U)]_j^k\sigma_k
\end{equation}
where $\Lambda(U)$ is a $3\times 3$ special orthogonal matrix associated to $U\in SU(2)/\mathbb{Z}_2$ and we used Einstein summation rule.
\end{theo}
For example, the operator $U_d$ acting on the spin operators of a single particle as
\begin{equation}
\label{1.117b}
\begin{split}
U_d\sigma_xU_d^{-1}=&s_x\sigma_x\\
U_d\sigma_yU_d^{-1}=&s_y\sigma_y\\
U_d\sigma_zU_d^{-1}=&s_z\sigma_z
\end{split}
\end{equation}
can be associated with the matrix
\begin{equation}
\Lambda(U_d)=
\begin{pmatrix}
s_x&0&0\\
0&s_y&0\\
0&0&s_z
\end{pmatrix}
\end{equation}
and the whole equivalence class $\{U_d,-U_d\}$ is is represented by $\Lambda(U_d)$.
Let us now analyze the case in which $\Lambda(U)$ is a permutation and an involution, hence it is not a cyclic or counter-cyclic permutation of the basis vectors.
%We will show later as this group relation is very powerful.
%As we may ensure the involutory nature, surely we cannot accept it is a trivial result of the fact that $\Lambda(U)$ is also an homomorphism.
Proceeding analogously to the case of time reversal operations acting on coordinates, we obtain the following theorem concerning the spin space.
\begin{theo}
%[Instance of time reversal operators on spin space]
\label{theo:1.2.3}
Given an operator ${\cal T}=UK$, where $K$ is complex conjugation, a time reversal operator acting on the spin space is obtained if $U$ is unitary and obeys:
\begin{equation}
\label{1.119b}
\begin{split}
&U^{(1)}_{xy}=\theta(\sigma_z\mp iI)\\
&U^{(1)}_{yz}=\theta(\sigma_x\mp i I)\\
&U^{(2)}_{xz}=\theta(\sigma_x\pm\sigma_z)\\
&U_d=\sigma_j
\end{split}
\end{equation}
where $\theta\in\mathbb{C}$, $|\theta|=1/\sqrt{2}$, and $j=x,y,z$.
\end{theo}
\begin{proof}
First, we consider the diagonal transformations $U_d$ that do not permute the Pauli matrices:
\begin{equation}
\label{1.117}
\begin{split}
U_d\sigma_xU_d^{-1}=&s_x\sigma_x\\
U_d\sigma_yU_d^{-1}=&s_y\sigma_y\\
U_d\sigma_zU_d^{-1}=&s_z\sigma_z
\end{split}
\end{equation}
where $s_i=\pm1$, $i=x,y,z$, to ensure the involution nature of the operation. Moreover, since the triad must remain right handed we need that $s_xs_ys_z=1$. Then, apart from the trivial case $s_x=s_y=s_z=1$, there are three possibilities. First, the case $s_y=1$ and $s_x=s_z=-1$ that corresponds to the canonical situation, since Eqs.\eqref{1.117} become
\begin{equation}
\label{1.118}
\begin{split}
U_d\sigma_xU_d^{-1}=&-\sigma_x\\
U_d\sigma_yU_d^{-1}=&\sigma_y\\
U_d\sigma_zU_d^{-1}=&-\sigma_z
\end{split}
\end{equation}
whose solution is $U_d=\sigma_y$. The second case is $s_x=1$ and $s_y=s_z=-1$, which leads to
% \begin{equation}
% \label{1.119}
% \begin{split}
% U_d\sigma_xU_d^{-1}=&\sigma_x\\
% U_d\sigma_yU_d^{-1}=&-\sigma_y\\
% U_d\sigma_zU_d^{-1}=&-\sigma_z
% \end{split}
% \end{equation}
$U_d=\sigma_x$. The third case, $U_d=\sigma_z$, holds for the $s_z=1$ and $s_x=s_y=-1$.
This leads to $\mathcal{T}^2=U_d K U_d K= \pm I$ as requested.
Regarding $U_d=\sigma_x$ and $U_d=\sigma_z$ the action of $K$ is irrelevant since these matrices are real, hence:
\begin{equation}
\mathcal{T}^2=\sigma_x^2=\sigma_z^2 = I
\end{equation}
Lastly, the case $U_d=\sigma_y$, gives $\mathcal{T}^2=-\sigma_y^2= - I$.
Concerning transformations that permute the order of the basis elements, and so of the Pauli matrices, we have the following possibilities: if we want to permute $\sigma_x$ and $\sigma_y$ we take either:
\begin{equation}
\label{1.122}
\begin{split}
U^{(1)}_{xy}\sigma_x(U^{(1)}_{xy})^{-1}=&\mp\sigma_y\\
U^{(1)}_{xy}\sigma_y(U^{(1)}_{xy})^{-1}=&\pm\sigma_x\\
U^{(1)}_{xy}\sigma_z(U^{(1)}_{xy})^{-1}=&\sigma_z
\end{split}
\end{equation}
or
\begin{equation}
\label{1.121}
\begin{split}
U^{(2)}_{xy}\sigma_x(U^{(2)}_{xy})^{-1}=&\pm\sigma_y\\
U^{(2)}_{xy}\sigma_y(U^{(2)}_{xy})^{-1}=&\pm\sigma_x\\
U^{(2)}_{xy}\sigma_z(U^{(2)}_{xy})^{-1}=&-\sigma_z
\end{split}
\end{equation}
To permute $\sigma_y$ and $\sigma_z$ we take either:
\begin{equation}
\label{eq:spin19}
\begin{split}
U^{(1)}_{yz}\sigma_x(U^{(1)}_{yz})^{-1}=&\sigma_x\\
U^{(1)}_{yz}\sigma_y(U^{(1)}_{yz})^{-1}=&\mp\sigma_z\\
U^{(1)}_{yz}\sigma_z(U^{(1)}_{yz})^{-1}=&\pm\sigma_y
\end{split}
\end{equation}
or
\begin{equation}
\label{eq:spin20}
\begin{split}
U^{(2)}_{yz}\sigma_x(U^{(2)}_{yz})^{-1}=&-\sigma_x\\
U^{(2)}_{yz}\sigma_y(U^{(2)}_{yz})^{-1}=&\pm\sigma_z\\
U^{(2)}_{yz}\sigma_z(U^{(2)}_{yz})^{-1}=&\pm\sigma_y
\end{split}
\end{equation}
To permute $\sigma_x$ and $\sigma_z$ we take either:
\begin{equation}
\label{eq:spin23}
\begin{split}
U^{(1)}_{xz}\sigma_x(U^{(1)}_{xz})^{-1}=&\mp\sigma_z\\
U^{(1)}_{xz}\sigma_y(U^{(1)}_{xz})^{-1}=&\sigma_y\\
U^{(1)}_{xz}\sigma_z(U^{(1)}_{xz})^{-1}=&\pm\sigma_x
\end{split}
\end{equation}
or
\begin{equation}
\label{eq:spin24}
\begin{split}
U^{(2)}_{xz}\sigma_x(U^{(1)}_{xz})^{-1}=&\pm\sigma_z\\
U^{(2)}_{xz}\sigma_y(U^{(1)}_{xz})^{-1}=&-\sigma_y\\
U^{(2)}_{xz}\sigma_z(U^{(1)}_{xz})^{-1}=&\pm\sigma_x
\end{split}
\end{equation}
Here, we required that the triad remains right handed, so the representation on the associated vectors $\boldsymbol e_i$ (see Theorem \ref{theo:2.4}) is a matrix of $SO(3)$. For instance, Eqs.\eqref{1.122} are associated with a matrix
\begin{equation}
O=\begin{pmatrix}
0&\mp 1&0\\
\pm 1&0&0\\
0&0&1
\end{pmatrix}
\end{equation}
that is special orthogonal. In fact, it is the linear application of $SO(3)$ that maps $(\boldsymbol e_x,\boldsymbol e_y,\boldsymbol e_z)$ to $(\mp\boldsymbol e_y,\pm\boldsymbol e_x,\boldsymbol e_z)$.
Now, use Theorem \ref{theo:1.2.2}. Because $U_{xy}$ must be unitary, we express it as the following linear combination:
\begin{equation}
\label{1.123}
U_{xy}=\alpha\sigma_x+\beta\sigma_y+\gamma\sigma_z+\delta I\quad\quad \alpha,\beta,\gamma,\delta\in\mathbb{C}
\end{equation}
whose coefficients obey Eq. \eqref{1.116}.
Substituting \eqref{1.123} in the third equation of Eqs.\eqref{1.122}, we obtain the condition:
\begin{equation}
\label{1.124}
\alpha\sigma_x\sigma_z+\beta\sigma_y\sigma_z+\gamma I+\delta \sigma_z=\alpha\sigma_z\sigma_x+\beta\sigma_z\sigma_y+\gamma I+\delta \sigma_z
\end{equation}
and, recalling that the Pauli matrices satisfy the fundamental relation
\begin{equation}
\label{1.145}
\sigma_j\sigma_k=\delta_{jk}I+i\tensor{\varepsilon} {_{jk}^{l}}\sigma_l
\end{equation}
Eq\eqref{1.124} writes
\begin{equation}
\label{line}
-i\alpha\sigma_y+i\beta\sigma_x+\gamma I+\delta\sigma_z=i\alpha\sigma_y-i\beta\sigma_x+\gamma I+\delta\sigma_z
\end{equation}
Since the set $\{\sigma_j,I\}$ is a basis for the algebra $\mathfrak{u}(2)$, the coefficient of the linear combinations on both sides of Eq.\eqref{line} must coincide, which means $\alpha=\beta=0$, and $U_{xy}=\gamma\sigma_z+\delta I$. Substituting in the first of Eqs.\eqref{1.122} one gets:
\begin{equation}
\label{1.125}
\gamma \sigma_z\sigma_x+\delta \sigma_x=\mp\gamma \sigma_y\sigma_z\mp\delta \sigma_y
\end{equation}
that is
\begin{equation}
\label{1.126}
i\gamma \sigma_y+\delta \sigma_x=\mp i\gamma \sigma_x\mp\delta \sigma_y
\end{equation}
which implies $\delta=\mp i\gamma$. To ensure that the corresponding matrix
\begin{equation}
U^{(1)}_{xy}=\gamma\sigma_z\mp i\gamma I=\gamma(\sigma_z\mp iI)
\end{equation}
is unitary, we need $UU^\dagger=I$, that is:
\begin{equation}
\gamma^2(\sigma_z\mp iI)(\sigma_z\pm iI)=2\gamma^2 I=I
\end{equation}
which means: $|\gamma|=1/\sqrt{2}$. The last thing to check, for a time reversal operator,
is that it is an involution, {\em i.e.}\ that $\mathcal{T}^2=UKUK=\pm I$ holds. This is indeed the case:
\begin{equation}
U^{(1)}_{xy} K U^{(1)}_{xy} K=\gamma^2(\sigma_z\mp iI)(\sigma_z\pm i I)=I
\end{equation}
The same reasoning can be repeated for the remaining permutations, which leads to:
\begin{equation}
\begin{split}
U^{(2)}_{xy}=\theta(\sigma_x\pm\sigma_y)\\
U^{(1)}_{yz}=\theta(\sigma_x\mp i I)\\
U^{(2)}_{yz}=\theta(\sigma_y\pm\sigma_z)\\
U^{(1)}_{xz}=\theta(\sigma_y\pm i I)\\
U^{(2)}_{xz}=\theta(\sigma_x\pm\sigma_z)\\
\end{split}
\end{equation}
where $\theta$ is a complex number with modulus $|\theta|=1/\sqrt{2}$. One may easily check that $\mathcal{T}^2=\pm I$ in all cases:
\begin{equation}
\begin{split}
U^{(2)}_{xy} K U^{(2)}_{xy} K=\theta^2(\sigma_x\pm\sigma_y)(\sigma_x\mp\sigma_y)=\mp i\sigma_z\neq\pm I\\
U^{(1)}_{yz} K U^{(1)}_{yz} K=\theta^2(\sigma_x\mp i I)(\sigma_x\pm iI)=I\\
U^{(2)}_{yz} K U^{(2)}_{yz} K=\theta^2(\sigma_y\pm \sigma_z)(-\sigma_y\pm\sigma_z)=\pm i\sigma_x\neq\pm I\\
U^{(1)}_{xz} K U^{(1)}_{xz} K=\theta^2(\sigma_y\pm i I)(-\sigma_y\mp i I)=\mp 2i\sigma_y\neq\pm I\\
U^{(2)}_{xz} K U^{(2)}_{xz} K=\theta^2(\sigma_x\pm\sigma_z)(\sigma_x\pm\sigma_z)=I
\end{split}
\end{equation}
The Theorem is proven, taking $U^{(1)}_{xy}$, $U^{(1)}_{yz}$ and $U^{(2)}_{xz}$ as the non diagonal time reversal operators on spin space.
\end{proof}
Now, swapping a pair of coordinates requires at most $2$ matrices. The total, considering also $U_d$, makes $9$ possible matrices $U_s$. Because, each of them is defined up to a complex number of modulus $1$, it seems that infinitely many are allowed. But whenever $U_s$ is applied to an operator on the spin space, giving $U_s f(\sigma_j) U_s^{-1}$, that number is eliminated. Therefore, there are $9$ operators acting differently on Pauli matrices.
This exhausts our present investigation of generalized time reversal invariance, for quantum non-relativistic systems, with and without half integer spin. In the following,
we apply our results to the theory of the Onsager relations, looking for generalizations of the compatibility conditions found in Ref.\cite{carbone2020necessary}, between time reversal transformations and a generic magnetic field.
% But
% one may still consider time reversal operations on spin space, that permute the particle indexes. Our analysis can in principle be extended introducing an operator $\mathcal{T}$ such that
% \begin{equation}
% \mathcal{T}\boldsymbol\sigma_i\mathcal{T}^{-1}=\boldsymbol\sigma_j\;\;\;\;\;\; i\neq j
% \end{equation}
% analogously to Sec.\ref{sec:The time reversal operator}, where coordinates of different particles are permuted.
\subsection{Compatibility condition between time reversal operations and magnetic field}
\label{sec:compatib}
The reasoning developed by Kubo about correlation functions and the entailing transport coefficients \cite{kubo1957statistical}, requires the existence of a time reversal operation, that commutes with the Hamiltonian of the system of interest. Because the quantum observables are defined as hermitian operators, one cannot follow the classical procedure to derive correlator relations. Given two observables $\phi$ and $\psi$, the mean value $\langle\phi(0)\psi(t)\rangle$ is ill-defined, as the combined operator $\phi(0)\psi(t)$ is not guaranteed to be hermitian. Therefore, Kubo introduced the \textit{canonical} correlator in Response Theory as:
\begin{equation}
\label{eq:2.23}
\langle\phi(0);\psi(t)\rangle=\frac{1}{\beta}\int_0^\beta d\lambda\;\; \mathbf{Tr}[\rho e^{\lambda H}\phi(0)e^{-\lambda H}\psi(t)]
\end{equation}
where $\rho$ is the density operator, $H$ the hamiltonian, and {\bf Tr} the trace over the Hilbert. Kubo then proved that this is indeed
a real quantity. Note that non-relativistic systems with spin are allowed: it suffices to include the spin degrees of freedom within the Hilbert space. Another possible definition is to symmetrize the correlator as $\langle\phi(0),\psi(t)\rangle=\langle\phi(0)\psi(t)\rangle+\langle\psi(t)\phi(0)\rangle$, but Kubo showed that \ref{eq:2.23} is sufficiently general.
The sufficient condition for the validity of the Onsager relations can then be formulated as {\em e.g.}\ Ref.\cite{rondoquantu}:
\begin{prop}
\label{prop:2.3}
Consider a quantum mechanical particle system in a magnetic field $\bf B$, with Hamiltonian $H$ and density matrix $\rho=\rho(H)$. Let $\mathcal{T}=UK$ be a time reversal operator that commutes with $H$, and let $\phi$ and $\psi$ be two observables with
signatures $\eta_\phi$ and $\eta_\psi$ with respect to $\cal T$, {\em i.e}\
\begin{equation}
\label{eq:2.24}
\mathcal{T}\phi\mathcal{T}^{-1}=\eta_\phi\phi\;\;\;\;\;\;\mathcal{T}\psi\mathcal{T}^{-1}=\eta_\psi\psi
\end{equation}
Then, the equality:
\begin{equation}
\langle\phi(0);\psi(t)\rangle_{\bf B} =
\eta_\phi \eta_\psi \langle\phi(t);\psi(0)\rangle_{\bf B}
\label{PhiPsiB}
\end{equation}
holds.
\end{prop}
% \begin{proof}
% In the Heisemberg picture the integrand of \ref{eq:2.23} can be rewritten as $Tr[\rho e^{\lambda H}\phi(0)e^{-\lambda H}e^{iHt}\psi(0)e^{-iHt}]$, then it holds the following chain of equalities
% \begin{equation}
% \label{eq:2.26}
% \begin{split}
% \mathbf{Tr}[\rho e^{\lambda H}\phi(0)e^{-\lambda H}e^{iHt}\psi(0)e^{-iHt}]&=\mathbf{Tr}[\rho e^{\lambda H}\phi(0)e^{-\lambda H}e^{iHt}\mathcal{T}^{-1}\mathcal{T}\psi(0)\mathcal{T}^{-1}\mathcal{T}e^{-iHt}]\\
% &=\eta_\psi \mathbf{Tr}[\mathcal{T}\rho e^{\lambda H}\phi(0)\mathcal{T}^{-1}e^{-\lambda H}e^{-iHt}\psi(0)e^{iHt}]\\
% &=\eta_\psi \mathbf{Tr}[\rho e^{\lambda H}\mathcal{T}\phi(0)\mathcal{T}^{-1}e^{-\lambda H}\psi(-t)]\\
% &=\eta_\psi\eta_\phi \mathbf{Tr}[\rho e^{\lambda H}\phi(0)e^{-\lambda H}\psi(-t)]\\
% \end{split}
% \end{equation}
% where we used the hypothesis listed above, i.e. $\mathcal{T}H=H\mathcal{T}$, \ref{eq:2.23} and \ref{eq:2.24}. Moreover, as it is commonly speculated in the classical case, we assumed that the density operator, quantum analogue of the classical probability density, commutes with the time reversal operator. The proof of \ref{eq:2.25} is then achieved inserting the integrand in \ref{eq:2.23} and using the hypothesis of stationary.
% \end{proof}
\noindent
\begin{rem}
Assuming that the density matrix depends on $H$ is quite general; for example, the canonical ensemble satisfies this constraint. Moreover, this assumption is sufficient for deduce the results of Proposition \eqref{prop:2.6}; it is not necessary to clarify what kind of statistical ensemble (bosons, fermions, anyons etc.) we are considering.
\end{rem}
This is the core of the Onsager reciprocal relations in Quantum mechanics in the presence of a magnetic field. Therefore, we now investigate how particular magnetic fields affect the number of time reversal operations consistent with Eq.\eqref{PhiPsiB}.
Consider the following Hamiltonian for a system of spinless particles coupled with a potential vector $\mathbf{A}$:
\begin{equation}
\label{eq:2.27}
H=\sum_{i=1}^N\left[\frac{(\mathbf{p}_i-q_i\mathbf{A}(x_i,y_i,z_i))^2}{2m_i}\right]
%+U(\mathbf{X},\mathbf{P},\mathbf{C})
\end{equation}
where $N$ is the number of particles, $q_i$ and $m_i$ are the charge and the mass of the $i$-th particle, and
%and $U(\mathbf{X},\mathbf{P},\mathbf{C})$ is the force potential, that in general depends on coordinate and momentum operators, as well as on a set of parameters $\mathbf{C}$. Let us take $U(\mathbf{X},\mathbf{P},\mathbf{C})=0$.
%The first thing to treat is dealing with the operator nature of the momenta in coordinates representation. Going into details, the minimal coupling term
%contains the operator $\mathbf{p}=-i\nabla$ and so in full is equal to
\begin{equation}
\label{eq:2.28}
(\mathbf{p}-q\mathbf{A})^2=-\nabla^2+iq\nabla\cdot\mathbf{A}+iq\mathbf{A}\cdot\nabla+q^2\mathbf{A}^2
\end{equation}
The commonly used Coulomb gauge $\nabla\cdot\boldsymbol{A}=0$ that importantly effects on this expression eliminating the second addend. As well-known, the vector potential associated to a magnetic field is defined up to the gradient of a scalar function since $\boldsymbol B=\nabla \times \boldsymbol A$. Thus, we can define an equivalence class containing the vector potentials that originate the same magnetic field. In the following, we refer to $[\boldsymbol{A}(\boldsymbol{x})]_R$ to denote a representative of the class containing $\boldsymbol{A}(\boldsymbol{x})$.
After this clarifications, let us step back to the classical case: we now show the core theorem that regards the compatibility between a particular time reversal operation and a generic magnetic field (and so vector potential). We point out as in the following we are going to refer to $V(\mathbb{R}^{n})$ as the space of vector fields on $\mathbb{R}^{n}$.
\begin{theo}[Compatibility conditions for $\boldsymbol A$ in Classical Mechanics]
\label{theo:2.7}
Consider a system of $N$ particles of equal mass $m$ and charge $q$. Let $\mathcal{M}$ be a generalized time reversal operator acting on the coordinates as $({\cal M} A)(\boldsymbol X)=A({\cal M}_M \boldsymbol X)$, where $\mathcal{M}_M\in O(3N)$. Denote by $\boldsymbol{\mathcal{A}}(\boldsymbol X)=(\boldsymbol{A}(\boldsymbol{x}_1),...,\boldsymbol{A}(\boldsymbol{x}_N))$ the
$3N$ dimensional vectors of coordinates, and by $[\boldsymbol{\mathcal{A}}(\boldsymbol X)]_R=([\boldsymbol{A}(\boldsymbol{x}_1)]_R,...,[\boldsymbol{A}(\boldsymbol{x}_N)]_R)$ the corresponding equivalence class. Introduce the reversal operator $\mathcal{M}':V(\mathbb{R}^{3N})\xrightarrow{}V(\mathbb{R}^{3N})$ defined by:
\begin{equation}
\label{3.13bb}
(\mathcal{M}'\boldsymbol{\mathcal{A}})(\boldsymbol X)\equiv\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol X)
\end{equation}
The operator $\mathcal{M}$ yields TRI in the presence of a magnetic vector potential $\boldsymbol{A}$, if and only if
\begin{equation}
\label{3.13}
(\mathcal{M}'\boldsymbol{\mathcal{A}})(\boldsymbol X)=[-\boldsymbol{\mathcal{A}}(\boldsymbol X)]_R
\end{equation}
\end{theo}
\begin{proof}
Recall that TRI holds if there is a time reversal transformation that preserves the equations of motion, as well as the Hamiltonian up to a gauge transformation.
In our case, the equations of motion are invariant under a gauge transformation, as can be easily seen. Then, in the presence of the minimal coupling with a magnetic field, the condition $H(\mathcal{M}\Gamma)=H(\Gamma)$ is verified for a gauge transformation, for any $\Gamma$. Therefore, the Hamiltonian can be written as
\begin{equation}
\label{3.14}
H(\boldsymbol X,\boldsymbol P)=\frac{1}{2m}\sum_i[\boldsymbol{p}_i -q\boldsymbol{A}(\boldsymbol{x}_i) ]^{2}=\frac{1}{2m}[\boldsymbol{P} -q\boldsymbol{\mathcal{A}}(\boldsymbol{X})]^2 = \frac{1}{2m}[\boldsymbol{P}+ q\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})] \cdot [\boldsymbol{P}+ q\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})]
\end{equation}
where $\boldsymbol{P}$ and $\mathcal{A}$ are $3N$-dimensional vectors. By Theorem \ref{theo:1.1.2}, a time reversal operation for systems subject to a magnetic field must act separately on coordinates and momenta, namely the transformation must be expressed by:
\begin{equation}
\label{3.15}
(\boldsymbol X, \boldsymbol P)\xrightarrow[]{\mathcal{M}}(\mathcal{M}_M\boldsymbol X,-\mathcal{M}_M\boldsymbol P)
\end{equation}
with $\mathcal{M}_M\in O(3N)$. This yields
\begin{equation}
\label{3.16}
\begin{split}
H(\mathcal{M}_M\boldsymbol X,-\mathcal{M}_M\boldsymbol P)&=\frac{1}{2m}[-\mathcal{M}_M\boldsymbol{P} -q\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})]^2\\
&=\frac{1}{2m}[\boldsymbol{P}+ q\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})]^2
\end{split}
\end{equation}
where we used the fact that the scalar product is invariant under rotations such as $\mathcal{M}_M$.
%Now, as $\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})$ is the rule of transformation of a vector field of $\mathbb{R}^{3N}$ under a rotation, that is the transformation applies to the component and to the coordinate dependence.
Now, let $\cal G$ be the scalar function involved in the gauge choice, define $\nabla_{3N}\mathcal{G}(\boldsymbol X)=(\nabla G(\boldsymbol{x}_1),...,\nabla G(\boldsymbol{x}_N ))$, and let us introduce the equivalence class of vector potentials defined by:
\begin{equation}
\label{3.17}
[\boldsymbol{A}]=\{\boldsymbol{A}\;\;|\;\; \nabla\times \boldsymbol A=\boldsymbol B\}
\end{equation}
This is the set of vector potentials corresponding to a given magnetic field $\boldsymbol B$. One representative element of this class, denoted by a subscript $R$, is expressed by:
\begin{equation}
[\boldsymbol{\mathcal{A}}(\boldsymbol X)]_R=([\boldsymbol{A}(\boldsymbol{x}_1)]_R,...,[\boldsymbol{A}(\boldsymbol{x}_N)]_R)
\end{equation}
%At this point, we have to prove the double implication in the statement of the theorem:
Now, taking
$\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})=-(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))$, substitution shows that
\eqref{3.14} and \eqref{3.16} are equal, up to a gauge transformation. The transformation leaves the equations of motion unchanged.
If, on the other hand, \eqref{3.14} and \eqref{3.16} are equivalent for any value of $\boldsymbol P$ and $\boldsymbol X$, that is:
\begin{equation}
\label{3.19}
\begin{split}
&\hskip -50pt
\boldsymbol{P}^2+2q \boldsymbol{P}\cdot(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))+q^2(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))^2\\
&\hskip 50pt =\boldsymbol{P}^2-2q\boldsymbol{P}\cdot\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})+q^2(\boldsymbol{\mathcal{M}_M\mathcal{A}}(\mathcal{M}_M\boldsymbol{X}))^2 \, , \quad \forall \boldsymbol P
\end{split}
\end{equation}
we can choose $\boldsymbol P=0$ obtaining
\begin{equation}
q^2(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))^2=q^2(\boldsymbol{\mathcal{M}_M\mathcal{A}}(\mathcal{M}_M\boldsymbol{X}))^2
\end{equation}
Then, Eq.\eqref{3.19} reduces to
\begin{equation}
\label{3.20}
2q \boldsymbol{P}\cdot(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))=-2q\boldsymbol{P}\cdot\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})
\end{equation}
Therefore, $\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})=-(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))$ holds.
\end{proof}
\noindent
This result generalizes the compatibility conditions of Ref.\cite{carbone2020necessary}, which focused on the $20$ transformations that separately act on each particle subspace. The result of Ref.\cite{carbone2020necessary} can be interpreted as a corollary of Theorem \ref{theo:2.7}.
\begin{cor}
Take a block diagonal transformation $\mathcal{M}_M$ separately acting on each particle subspace, denoting by $\mathcal{M}_m$ one such $3\times3$ block. Then
\begin{equation}
\label{3.21}
\mathcal{M}'\boldsymbol{A}=\mathcal{M}_m\boldsymbol{A}(\mathcal{M}_m\boldsymbol X)=[-\boldsymbol{A}]_R
\end{equation}
\end{cor}
\begin{proof}
It trivially follows form the definition of $\boldsymbol{\mathcal{A}}$ given in Theorem \ref{theo:2.7}, and from Eq.\eqref{3.13}, where $\mathcal{M}_M$ is a block diagonal acting on a single $3$-dimensional particle subspace.
\end{proof}
\begin{prop}
\label{prop:1.1.7}
Take the Hamiltonian \eqref{eq:2.27} in which $(m_i,q_i) \ne (m_j,q_j)$ if $i \ne j$ are particle indices. Then, a time reversal operator yielding TRI is given by a block diagonal matrix $A$, whose blocks separately act on single particle subspaces.
\end{prop}
\begin{proof}
By {\em reductio ad absurdum}, suppose a valid non diagonal time reversal operation $\mathcal{M}_{nd}$ exists for the system described by the Hamiltonian \eqref{eq:2.27}. Without loss of generality, assume that the operation acts on the $M \times M$ coordinates (with $M=3N$) swapping those of particle $j$ with those of particle $j+1$:
\begin{equation}
\label{1.1.39b}
(\boldsymbol{x}_1,...,\boldsymbol{x}_j,\boldsymbol{x}_{j+1},...,\boldsymbol{x}_N,\boldsymbol{p}_1,...,\boldsymbol{p}_j,\boldsymbol{p}_{j+1},...,\boldsymbol{p}_N)
\xrightarrow{\mathcal{M}_{nd}}(\boldsymbol{x}_1,...,\boldsymbol{x}_{j+1},\boldsymbol{x}_j,...,\boldsymbol{x}_N,
-\boldsymbol{p}_1,...,-\boldsymbol{p}_{j+1},-\boldsymbol{p}_{j},...,-\boldsymbol{p}_N )
\end{equation}
By hypothesis, $H(\mathcal{M}_{nd}\Gamma)=H(\Gamma)$, because $\mathcal{M}_{nd}$ yields TRI. It suffices to check the contributions due to particles $j$ and $j+1$, because all other contributions are unchanged. Thus we can write:
\begin{eqnarray}
\label{1.1.40}
\sum_{i=1}^N\left[\frac{(\boldsymbol{p}_i-q_i\boldsymbol{A}(\boldsymbol{x}_i))^2}{2m_i}\right] &=& ...+\frac{(\boldsymbol{p}_j-q_j\boldsymbol{A}(\boldsymbol x_j))^2}{2m_j}+\frac{(\boldsymbol{p}_{j+1}-q_{j+1}\boldsymbol{A}(\boldsymbol x_{j+1}))^2}{2m_{j+1}}+...
\\
\label{1.1.41}
&&...+\frac{(\boldsymbol{p}_{j+1}+q_j\boldsymbol{A}(\boldsymbol x_{j+1}))^2}{2m_{j}}+\frac{(\boldsymbol{p}_{j}+q_{j+1}\boldsymbol{A}(\boldsymbol x_{j}))^2}{2m_{j+1}}+...
\end{eqnarray}
But $m_i\neq m_j$ or $q_i\neq q_j$ for $\forall i\neq j$ by hypothesis. Then, in general, one has $H(\mathcal{M}_{nd}\Gamma)\neq H(\Gamma)$, unless special cases are considered, because the coordinates and the momenta may take any value in $\mathbb{R}^3$. This is absurd. As the reasoning can be repeated varying $i$ and $j$ in \eqref{1.1.39b}, $\mathcal{M}_{nd}$ cannot mix the different single particles coordinates and momenta; it must be block diagonal, with each block acting on a single particle subspace.
\end{proof}
\noindent
This result may seem to be a drastic limitation on the range of TRI, but in reality it is not. Systems with particles of same charge and mass are the most widely studied, both theoretically and experimentally.
\vskip 5pt
In Ref.\cite{carbone2020necessary}, the compatibility conditions is expressed also in terms of the magnetic field, instead of the vector potential. Here, we present an alternative derivation of that result result. We point out as in the following we are going to refer to $V_p(\mathbb{R}^{n})$ as the space of pseudovector fields on $\mathbb{R}^{n}$.
\begin{theo}[Restricted compatibility conditions for $\boldsymbol B$ in Classical Mechanics]
\label{theo:3.1.2}
Consider a system of $N$ particles of equal mass $m$ and charge $q$. Let $\mathcal{M}$ be a generalized time reversal operator acting on the coordinates as $\mathcal{M}\boldsymbol X=(\mathcal {M}_m \boldsymbol x_1,...,\mathcal {M}_m \boldsymbol x_N)$, where $\mathcal{M}_m\in O(3)$. Introduce the pseudovector field rotation operator $\mathcal{M}':V_p(\mathbb{R}^{3})\xrightarrow{}V_p(\mathbb{R}^{3})$ defined by:
\begin{equation}
\label{3.13c}
(\mathcal{M}'\boldsymbol{B})(\boldsymbol x)=\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m\boldsymbol{B}(\mathcal{M}_m\boldsymbol x)
\end{equation}
The operator $\mathcal{M}$ yields TRI in the presence of a magnetic field $\boldsymbol{B}$, if and only if
\begin{equation}
\label{3.22}
(\mathcal{M}'\boldsymbol{B})(\boldsymbol x)=-\boldsymbol{B}(\boldsymbol x)
\end{equation}
\end{theo}
% \begin{theo}[Restricted compatibility conditions for $\boldsymbol B$ in Classical Mechanics]
% \label{theo:3.1.2}
% A generalized time reversal operation $\mathcal{M}$, separately acting on each particle subspace, preserves TRI in the presence of a magnetic field $\boldsymbol{B}$ ({\em i.e.}\ it preserves the equations of motion as well as the Hamiltonian, up to a gauge transformation) if and only if the associated transformation on the $3$-dimensional vector fields, $\mathcal{M}':V(\mathbb{R}^{3})\xrightarrow{}V(\mathbb{R}^{3})$, verifies:
% \begin{equation}
% \label{3.22}
% \mathcal{M}'\boldsymbol{B}=\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m\boldsymbol{B}(\mathcal{M}_m\boldsymbol x)=-\boldsymbol{B}(\boldsymbol x)
% \end{equation}
% with $\mathcal{M}_m$ the specific orthogonal matrix representation on the $3D$ coordinate point $\boldsymbol x\equiv(x,y,z)$. %We consider $U(\boldsymbol X,\boldsymbol P,\boldsymbol C)=0$.
% \end{theo}
\begin{proof}
We have to prove that \eqref{3.21} and \eqref{3.22} are equivalent. Let us introduce the notation:
\be
\nabla_{\boldsymbol x}
\equiv(\partial_x,\partial_y,\partial_z)
\ee
hence $\nabla_{{\cal M}_m \boldsymbol x}$ is the gradient with respect to the coordinates rotated by ${\cal M}_m$. Then, keeping in mind that $\boldsymbol B = \nabla\times \boldsymbol A$, the application of the transformation $\mathcal{M}'$ defined by the orthogonal matrix $\mathcal{M}_m$ yields \cite{milne1948vectorial}:
\begin{equation}
\label{3.23}
\begin{split}
\mathcal{M}'(\nabla_{\boldsymbol x} \times \boldsymbol A)&=(\mathcal{M}_m\nabla_{\mathcal{M}_m\boldsymbol x})\times(\mathcal{M}_m\boldsymbol A \circ \mathcal{M}_m\boldsymbol)\\
&=\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m(\nabla_{\mathcal{M}_m\boldsymbol x} \times \boldsymbol A \circ \mathcal{M}_m)\\
&=\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m\boldsymbol B \circ \mathcal{M}_m
\end{split}
\end{equation}
because an orthogonal matrix such as $\mathcal{M}_m$ yields:
\begin{equation}
\nabla_{\mathcal{M}_m\boldsymbol x} \times \boldsymbol A \circ \mathcal{M}_m = \nabla\times \boldsymbol A = {\boldsymbol B} \, ,
\quad
\mathcal{M}_m\nabla_{\mathcal{M}_m\boldsymbol x}=\nabla_{\boldsymbol x}
\end{equation}
%and so that the first line of \eqref{3.23} is
%\begin{equation}
%\label{3.25}
%(\mathcal{M}_m\nabla(\mathcal{M}_m\boldsymbol x))\times(\mathcal{M}_m\boldsymbol A(\mathcal{M}_m\boldsymbol x))=\nabla(\boldsymbol x)\times(\mathcal{M}_m\boldsymbol A(\mathcal{M}_m\boldsymbol x))
%\end{equation}
Now, assuming Eq.\eqref{3.21} holds, we have:
\begin{equation}
\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m\boldsymbol B \circ \mathcal{M}_m =\nabla_{\boldsymbol x}\times[-\boldsymbol{A} ]_R=-\boldsymbol B
\end{equation}
Vice versa, assuming Eq.\eqref{3.22} holds, we start from
\begin{equation}
\operatorname{det}(\mathcal{M}_m)\mathcal{M}_m \nabla_{\mathcal{M}_m\boldsymbol x}\times \boldsymbol A \circ \mathcal{M}_m =-\nabla_{\boldsymbol x} \times \boldsymbol A
\end{equation}
Moreover, using the first and second lines of \eqref{3.23}, we get
\begin{equation}
\nabla_{\boldsymbol x} \times (\mathcal{M}_m\boldsymbol A \circ \mathcal{M}_m) =-\nabla_{\boldsymbol x} \times \boldsymbol A
\end{equation}
and so, considering the fact that the rotor of a gradient is null, we obtain the thesis \eqref{3.21}, where gauge freedom is included.
\end{proof}
%Theorem \ref{theo:2.7} does not seem to directly apply in the presence of magnetic fields, except the $20$ block diagonal time reversal operations.
In the presence of magnetic fields, Theorem \ref{theo:2.7} seems to directly apply only in the case of the block diagonal time reversal operations.
Unlike Theorem \ref{theo:2.7},
Theorem \ref{theo:3.1.2} cannot be immediately generalised to the cases with a magnetic field, because in $\mathbb{R}^{3N}$ we miss the analogue of the relation $\boldsymbol A=\nabla\times\boldsymbol B$, which holds in $\mathbb{R}^3$. Operators similar to the curl can be defined in spaces other than $\mathbb{R}^3$, see for example Ref.\cite{conlon2008differentiable}, but we cannot claim that $\boldsymbol{\mathcal{A}}=\nabla \times \mathcal{B}$, for $\boldsymbol{\mathcal{A}} , \boldsymbol{\mathcal{B}}\in\mathbb{R}^{3N}$.
Finally coming to the quantum context, the following holds.
%In the following, we extend to Quantum Mechanics the compatibility conditions found in Classical Mechanics. We will see that spinless particles can be treated as fully analogous to classical particles.
\begin{prop}
\label{prop:3.1.2}
The statements of Theorems \ref{theo:2.7} and \ref{theo:3.1.2} extend to Quantum Mechanics on compact spaces, when $\boldsymbol X$ is identified with the position operator, cf.\ Theorem \ref{theo:1.2.1}.
\end{prop}
\begin{proof}
Let us start from Theorem \ref{theo:2.7}: regarding \eqref{3.14} and \eqref{3.15} the only differences are that $\boldsymbol P=-i\nabla$ and that the case $\mathcal{M}_M^2=-I$ is admitted. The proof can be repeated until \eqref{3.19}, since using Coulomb gauge and \eqref{eq:2.28} we do not have the $2$ in front of the mixed product. As a side note, the Coulomb gauge condition involves a scalar product too and so it invariant under rotation. \\
At this point, since we are using operators, we cannot impose $\boldsymbol P=0$ as done in Classical Mechanics. Nevertheless, \eqref{3.19} is an equivalence between operators and so it must holds for any wave function defined on a certain domain. If the domain is compact, we can consider the case of the constant without issues of normalization; the application to it of the differential operator $\boldsymbol P$ yields the constraint \begin{equation}
q^2(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))^2=q^2(\boldsymbol{\mathcal{M}_M\mathcal{A}}(\mathcal{M}_M\boldsymbol{X}))^2
\end{equation}
Finally, we obtain an expression similar to \eqref{3.20}:
\begin{equation}
(\boldsymbol{\mathcal{A}}(\boldsymbol{X})+\nabla_{3N}\mathcal{G}(\boldsymbol X))\cdot \boldsymbol{P}=-\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol{X})\cdot\boldsymbol{P}
\end{equation}
where we used the correct ordering of the momentum operators. Also in this case the only way to verify this operator equality is to have $\mathcal{M}_M\boldsymbol{\mathcal{A}}(\mathcal{M}_M\boldsymbol X)=[-\boldsymbol{\mathcal{A}}]_R$.\\
Coming to Theorem \ref{theo:3.1.2}, its extension is trivial since the proof involves only functions of the coordinate operator which acts in a multiplicative way in coordinate representation; in this way, we can proceed as if we were manipulating numbers and not operators, easily reproducing the demonstration.
\end{proof}
This is our main result for the spinless case. Two observations and a fundamental example are in order.
\begin{rem}
\normalfont
The compatibility condition on magnetic fields of Theorem \ref{theo:3.1.2} seems to be of limited applicability, compared to the one about vector potentials, since it is limited to transformations separately acting on single particle subspaces. Nevertheless, that is the most interesting case, since particles usually interact with each other. For instance, in the case of the central interaction potential:
\begin{equation}
\label{poten}
V=\sum_{i<j}v(\boldsymbol x_i-\boldsymbol x_j)
\end{equation}
with given pair potential $v(\boldsymbol x)=v(|\boldsymbol x|)$, treated in Ref.\cite{carbone2020necessary}, the time reversal operations must be block diagonal. It must separately and identically act on each particle subspace, otherwise $v(\boldsymbol x_i-\boldsymbol x_j)$ would not be preserved, in general. If the time reversal operation is such a block diagonal operator $Q$,
it can be combined with the operator $P_{ij}$ that swaps particle $i$ with particle $j$, obtaining a new involution $QP_{ij}$: $QP_{ij}QP_{ij}=I$. The same applies to the case of half integer spin particles, and it is not necessary to repeat the reasoning.
Obviously, an interaction term, such as \eqref{poten}, prevents the use of the time reversal operations of Proposition \ref{prop:1.2.6}, that cannot be expressed by $3\times 3$ block diagonal matrices. Thus, in the following we neglect the case with $Q^2=-I$.
%it is not necessary to study time reversal operations that act on spin operators as
%\begin{equation}
% \mathcal{T}\boldsymbol\sigma_i\mathcal{T}^{-1}=\Lambda\boldsymbol\sigma_j\;\;\;\;\;\; i\neq j
%\end{equation}
%with as usual $i,j$ indexes of particle subspaces and $\Lambda\in SO(3)$. This because we can apply the same composition procedure performed above in the context of coordinate operators.
\end{rem}
\begin{rem}
\label{rem:2.1}
\normalfont
The action of the operator $\mathcal{T}$ of Proposition \ref{prop:2.3} deeply differs from that of the traditional one, which includes the inversion of the magnetic field, $\mathcal{T}_B$, cf.\ Ref.\cite{toda1991statistical}.
Nevertheless, $\mathcal{T}$ suffices for the validity of the Onsager relations, because they arise from a phase space integration and, pairing differently the contributions to the correlator, the integral may still yield the same result. %For example, imagine in the classical case a contribution $F_{1\xrightarrow{}2}$ to the integral, with 1 and 2 the initial and the final conditions and $F$ denoting forward in time. The original Onsager theory relies on the existence of an equal contribution $B_{2\xrightarrow{}1}$, with $B$ denoting backward evolution. In the presence of a magnetic field, the trajectory corresponding to $B_{2\xrightarrow{}1}$ does not exist in the original system, it only does in another system with opposite field. Nevertheless, a contribution $B_{3\xrightarrow{}4}$, equal to $B_{2\xrightarrow{}1}$, may stem in the original system from different initial and final conditions, identified by some symmetry of the dynamics in the magnetic field.
Statistical mechanical relations are typically of this kind. Therefore, they do not require microreversibility. Weaker conditions suffice, such as those discussed here, that we may call \textit{statistical time reversibility}, cf.\ also Refs.\cite{Klages,ColR1,ColR2}.
This equally applies to the quantum case, if trajectories are replaced by the time evolution of states in the Hilbert space. In fact, the invariance of the Hamiltonian and of the equation of motion ensures that
$$
\ket{\psi'(t')}=\mathcal{T}\ket{\psi(t')}\, , \quad \mbox{with } ~~ t'=-t
$$
is a physical state if $\ket{\psi(t)}$ is. Roughly speaking, in the quantum case it is just a matter of rearranging the contributions to the trace in \eqref{eq:2.23}.
\end{rem}
Let us reconsider, in the light of the above result, the case of a constant magnetic field.
%Before Ref.\cite{rondoe} was published, a constant magnetic field was thought to completely break TRI, and to lead to violations of the Onsager relations. At this point we can close the circle, showing that
%The result is the following.
\begin{theo}
Take a system of particles interacting via
the central potential \eqref{poten}. Let this system be subjected to a constant external magnetic field, $\boldsymbol B=(B_1,B_2,B_3)$. The dynamics is invariant under infinitely many time reversal operations: those whose action on each single particle subspace is represented by:
\begin{equation}
\mathcal{M}_m=
\begin{pmatrix}
a&b&0\\
b&-a&0\\
0&0& 1
\end{pmatrix}
\label{teosd}
\end{equation}
where $a,b \in \mathbb{R}$, $a^2+b^2=1$ and $\operatorname{det}(\mathcal{M}_m)=-1$.
\end{theo}
\begin{proof}
The time evolution of the system does not depend on the coordinates frame, so we can choose the axis $z$ along the direction of $\boldsymbol B$, and write
$B=(0,0,1)$, up to a dimensional constant. Consider the time reversal transformations whose action on each single particle subspace is represented by Eq.\eqref{teosd}.
By definition, $\mathcal{M}_m\in O(3)$ and $\mathcal{M}_m^2=I$, therefore $\mathcal{M}_m$ represents the action on coordinates of a time reversal operation, as follows from Theorem \ref{theo:1.1.2}. Furthermore, the compatibility condition of Theorem \ref{theo:3.1.2} is trivially verified:
\begin{equation}
\operatorname{det}(\mathcal{M}_m) \mathcal{M}_m \boldsymbol B =(0,0,-1) = -\boldsymbol B
\end{equation}
Taking $a=\cos{\theta}$ and $b=\sin{\theta}$, with $\theta\in[0,2\pi)$, infinitely many possible choices are allowed for $\mathcal{M}_m$.
\end{proof}
This result seems to boldly contradict the traditional opinion that
%is a fundamental endpoint since it is the exact contrary of what was believed: not only the presence of
any magnetic field
%is not sufficient to
breaks TRI.
%in fact there are infinite valid time reversal operations.
With hindsight, it is not so surprising, when the existence of generalized time reversal symmetries has been ascertained.
%A posteriori, as often happens, it is quite obvious:
In particular, a constant magnetic field directed along along the $z$ axis preserves all the symmetries that differ by rotations of the $xy$ plane, if it preserves one of them.
%in this case, the key is to understand how to build a generalized time reversal operation making use of rotations, finally achieving this important result.\\
In the case of particles with spin, the Pauli Hamiltonian \eqref{eq:2.40} contains the minimal coupling term, for which
%thus, we surely have to verify
%the conditions found in this section, since
Proposition \ref{prop:3.1.2} applies.
%can be extended to quantum systems.
However, magnetic field and spins are also coupled, with the spins belonging to a different space. Then, a time reversal operation ought to take the form:
\begin{equation}
\mathcal{T}=(U_c\otimes U_s^1\otimes ...\otimes U_s^N)K
\label{UcUs1UsN}
\end{equation}
where $U_c$ acts on coordinate and momentum operators, while $U_s^j$ acts on the spin operator of particle $j$. The situation differs from that considered in Theorem \ref{theo:2.7}, because we now have:
\begin{equation}
\mathcal{T}H\mathcal{T}^{-1}\propto \boldsymbol B(U_c\boldsymbol x_j U_c^{-1})\cdot U_s^j K\boldsymbol \sigma_j K U_s^j
\end{equation}
where $H$ is the Hamiltonian and $\boldsymbol B$ the magnetic field. This way, the minimal coupling and the spin couplings with $\boldsymbol B$ are considered independently of each other:
%Or at least, it is what one could think; in reality, this is not the case:
a time reversal operation compatible with the minimal coupling leaves the spin field coupling unchanged.
\begin{rem}
The transformation $U_c$ acts on coordinates and not on $\boldsymbol B$ as a vector field, that is the components of $\boldsymbol B$ transform as \begin{equation}
U_c B_i U_c^{-1}\equiv B_i(U_c\boldsymbol x_i U_c^{-1})
\end{equation}
\end{rem}
% In conclusion, the main objective is expressing some conditions on $\boldsymbol B(U_c\boldsymbol x_i U_c^{-1})$ for each kind of time reversal operation on spin space found in Theorem \ref{theo:1.2.3}:
% \begin{theo}[Compatibility conditions for particles with spin]
% Let us consider a system of particles with spin coupled with a magnetic field \begin{equation}
% \boldsymbol B(\boldsymbol x)=(B_1(\boldsymbol x),B_2(\boldsymbol x),B_3(\boldsymbol x))
% \end{equation}
% First of all, the time reversal operator $U_c$ must act in a block diagonal way on the single particle subspace. Moreover, we can express, with regard to the notation used in Theorem \ref{theo:1.2.3} and in \eqref{1.119b}, the compatibility conditions between $U_s$, $U_c$ and $\boldsymbol B$:
% \begin{itemize}
% \item If one consider the so-called diagonal transformations of Pauli matrices $U_s=U_d=\sigma_i$, the counterpart $U_c$ and the magnetic field must be such that
% \begin{equation}
% \label{3.31}
% \sigma_x:
% \begin{cases}
% U_{c}B_1U_{c}=B_1\\
% U_{c}B_2U_{c}=B_2\\
% U_{c}B_3U_{c}=-B_3
% \end{cases}
% \;\;\sigma_y:
% \begin{cases}
% U_{c}B_1U_{c}=-B_1\\
% U_{c}B_2U_{c}=-B_2\\
% U_{c}B_3U_{c}=-B_3
% \end{cases}
% \;\;\sigma_z:
% \begin{cases}
% U_{c}B_1U_{c}=-B_1\\
% U_{c}B_2U_{c}=B_2\\
% U_{c}B_3U_{c}=B_3
% \end{cases}
% \end{equation}
% where $B_i=B_i(\boldsymbol x)$ for $i=1,2,3$.
% \item If we consider the three non-diagonal possibilities we must verify that
% \begin{equation}
% \label{3.32}
% U_{xy}^{(1)}:
% \begin{cases}
% U_cB_1U_c=\pm B_2\\
% U_cB_2U_c=\pm B_1\\
% U_cB_3U_c=B_3
% \end{cases}
% \;\; U_{yz}^{(1)}:
% \begin{cases}
% U_cB_1U_c=B_1\\
% U_cB_2U_c=\pm B_3\\
% U_cB_3U_c=\pm B_2
% \end{cases}
% \;\;U_{xz}^{(2)}:
% \begin{cases}
% U_cB_1U_c=\pm B_3\\
% U_cB_2U_c=B_2\\
% U_cB_3U_c=\pm B_1
% \end{cases}
% \end{equation}
% \end{itemize}
% \end{theo}
% \begin{proof}
% To start with, the coupling term in the Hamiltonian is
% \begin{equation}
% \label{3.33}
% \sum_i \boldsymbol B(\boldsymbol x_i)\cdot \boldsymbol\sigma_i
% \end{equation}
% where as usual $\boldsymbol x_i=(x_i,y_i,z_i)$ with $i$ the particle index. In the case in which the dependence of $\boldsymbol B$ on spacial coordinates is not completely general, as for example if $B(\boldsymbol x_i)=\boldsymbol B(x,y)$ does not depend on $z$, the transformations $U_c$ that contains permutation of the $z$ coordinates of different particles trivially preserves $\boldsymbol B$ and the conditions that we will find can be used.\\
% On the contrary, the case in which $U_c$ permutes coordinates which $\boldsymbol B$ depends on is not compatible with the time reversal operators on spin individuated in Theorem \ref{theo:1.2.3}: in fact, if $U_c$ mixes also the particle indexes but the dependence is not trivial as the previous case, there are $i$ and $j$ such that $x_i$ is mapped in $x_j$ by $U_c$ and we have
% \begin{equation}
% \boldsymbol B(x_i,...)\cdot \boldsymbol\sigma_i\xrightarrow{}\boldsymbol B(x_j,...)\cdot \boldsymbol\sigma_i\neq\boldsymbol B(x_i,...)\cdot \boldsymbol\sigma_i
% \end{equation}
% We remark that none of the operators individuated in Theorem \ref{theo:1.2.3} can swap particle indexes of spin operators. The conclusion is that in the most general case $U_c$ must be block diagonal on the single particle subspace, {\em i.e.} we reduce to study the $20$ time reversal operations for each particle.\\
% Coming to the mathematical constraint \eqref{3.31} and \eqref{3.32}, we show, for instance, the procedure used to derive them in the case $U_s=\sigma_x$; the other possibilities are conceptually obtained in the same way. Neglecting the particle index, one addend in the term \eqref{3.33} is transformed as
% \begin{equation}
% \begin{split}
% &\boldsymbol B(\boldsymbol x)\cdot \boldsymbol \sigma \longrightarrow\boldsymbol B(U_c\boldsymbol x_i U_c^{-1})\cdot U_s K\boldsymbol \sigma_i K U_s=\\
% &=B_1(U_c\boldsymbol x_i U_c^{-1})U_s \sigma_x U_s^{-1}-B_2(U_c\boldsymbol x_i U_c^{-1})U_s \sigma_y U_s^{-1}+B_3(U_c\boldsymbol x_i U_c^{-1})U_s \sigma_z U_s^{-1}\\
% &=B_1(U_c\boldsymbol x_i U_c^{-1})\sigma_x+B_2(U_c\boldsymbol x_i U_c^{-1}) \sigma_y-B_3(U_c\boldsymbol x_i U_c^{-1}) \sigma_z
% \end{split}
% \end{equation}
% Finally, the coupling term is preserved by the application if the original term is equal to the transformed one, that is if the coefficient of the Pauli matrices are equal; as usual, since we are comparing operators, we need to apply it to position eigenfunction in order to deal with numbers and to write the condition
% \begin{equation}
% \sigma_x:
% \begin{cases}
% U_{c}B_1U_{c}=B_1\\
% U_{c}B_2U_{c}=B_2\\
% U_{c}B_3U_{c}=-B_3
% \end{cases}
% \end{equation}
% As stressed before, this procedure can be repeated in a trivial way, so we omit the computation for the sake of brevity.
% \end{proof}
% This result is very important for various reasons. First of all, it make us understand why in classical textbooks like Ref.\cite{sachs}, but also in recent papers as Ref.\cite{rondoquantu}, the coupling with spin seemed to be very problematic in the context of TRI: in fact, if one reduces the analysis to the canonical point of view, that is $U_s=\sigma_y$ on spin space and $U_c$ the representation of the canonical time reversal operation, then the problem has no solution; indeed, it would be necessary that
% \begin{equation}
% \begin{cases}
% B_1(\boldsymbol x)=-B_1(\boldsymbol x)\\
% B_2(\boldsymbol x)=-B_2(\boldsymbol x)\\
% B_3(\boldsymbol x)=-B_3(\boldsymbol x)
% \end{cases}
% \end{equation}
% that is, obviously, absurd considering a non-null magnetic field. \\
\noindent
Note that statistical relations could be used to distinguish systems whose particles possess spin or from those which do not, if the extra conditions implied by the presence of spin reduced the number of suitable time reversal operators. In fact, for transformations like \eqref{UcUs1UsN}, and with block diagonal $U_c$ separately acting on single particles subspaces, this is not possible.
\begin{theo}
\label{theo:2.10}
The compatibility condition of Theorem \ref{theo:3.1.2} suffices for TRI to hold also in case of particles with spin that obey the Pauli equation \eqref{eq:2.40}.
\end{theo}
\begin{proof}
Write the Pauli hamiltonian as:
\begin{equation}
H=H_{mc}+H_{sc}
\end{equation}
where $H_{mc}$ refers to the minimal coupling, of the spinless case, and $H_{sc}=\sum_{j=1}^N \boldsymbol{\sigma}_j\cdot \boldsymbol B(\boldsymbol x_j)$ is the spin-field coupling, with $\boldsymbol\sigma=(\sigma_x,\sigma_y,\sigma_z)$. Given that we are analyzing transformations separately acting on single particle spaces, we study a single addend of the summation.
By hypothesis, there exists a time reversal operator $\mathcal{T}=(U_c\otimes U_s)K$ whose action on coordinates commutes with $H_{mc}$, \textit{i.e.} Eq.\eqref{3.22} holds or, equivalently,
\begin{equation}
\label{2.136}
\boldsymbol{B}(\mathcal{M}_m\boldsymbol x)=-\frac{1} {\operatorname{det}(\mathcal{M}_m)}\mathcal{M}_m^{-1}\boldsymbol{B}(\boldsymbol x)
\end{equation}
Because of Theorem \ref{theo:3.1.2},
$\mathcal{M}_m$ is the 3-dimensional orthogonal and involutory matrix representation of the action of the time reversal operator on a single particle position operator. Under the action of $U_c$, one has:
\begin{equation}
U_c(\boldsymbol\sigma\cdot B(\boldsymbol x))U_c^{-1}=\boldsymbol\sigma\cdot (-P\boldsymbol B(\boldsymbol x))=-P\boldsymbol \sigma\cdot \boldsymbol B(\boldsymbol x)
\end{equation}
where $P\boldsymbol\sigma=P^k_j\sigma_k$, and
\begin{equation}
P=\frac{1}{\operatorname{det}(\mathcal{M}_m)}\mathcal{M}_m^{-1}
\end{equation}
which is a special orthogonal matrix. By Theorem \ref{iso}, there exists a special unitary matrix $U$ such that
\begin{equation}
U \sigma_j U^{-1}=P^k_j\sigma_k
\end{equation}
Then, because $P^2=I$ by definition of time reversal operation, we have $U^2=\pm I$, since the isomorphism maps $\pm I\in SU(2)$ to $I\in SO(3)$. Moreover, letting $K$ be the complex conjugation operator, we have
\begin{equation}
\sigma_y K \sigma_j K\sigma_y=-\sigma_j
\end{equation}
where $j,k=1,2,3=x,y,z$. Now, take $U_s=U\sigma_y$. By definition, the time reversal operator acts on spin operators as
\begin{equation}
\mathcal{T}\sigma_j\mathcal{T}^{-1}=U\sigma_y K \sigma_j K\sigma_y U^{-1}=-P_j^k\sigma_k
\end{equation}
where we used the decomposition of the Hilbert space of the system as the direct product of the spin space and of the coordinate space. This allows us to separately change the coordinate system in each space, and to write:
\begin{equation}
\sum_{k=1}^N \boldsymbol{\sigma}_k\cdot \boldsymbol B(\boldsymbol x_k)=\sum_{k=1}^N
\Big[
\boldsymbol{\sigma}_x\otimes\boldsymbol B_1(\boldsymbol x_k)+\boldsymbol{\sigma}_y\otimes\boldsymbol B_2(\boldsymbol x_k)+\boldsymbol{\sigma}_z\otimes\boldsymbol B_3(\boldsymbol x_k)
\Big]
\end{equation}
As the choice of the $z$ axis in the spin space is not related to that of the $z$ axis in coordinate space, $U_s$ does not depend on the choice of $U_c$. This closes the circle: for any valid $U_c$, we can always find $U$, and so $U_s$, in order to keep $H_{sc}$ invariant. In particular, we obtain:
\begin{equation}
\mathcal{T}\boldsymbol\sigma\cdot B(\boldsymbol x)\mathcal{T}^{-1}=-P\boldsymbol \sigma \cdot -P \boldsymbol B(\boldsymbol x)=\boldsymbol \sigma\cdot\boldsymbol B(\boldsymbol x)
\end{equation}
\end{proof}
\begin{lemma}
\label{lem1}
The Hamiltonian is not invariant under the action of $U_s=\sigma_y$ if the magnetic field is constant. To be recover the original Hamiltonian, the magnetic field must be manually inverted.
\end{lemma}
\begin{proof}
The choice $U_s=\sigma_y$ corresponds to $U=I$, that is $P=I$ and $\mathcal{M}_m=\pm I$. This yields to two constraints on the magnetic field: $\boldsymbol B(\boldsymbol x)=-\boldsymbol B(\boldsymbol x)$ choosing the plus and $\boldsymbol B(-\boldsymbol x)=-\boldsymbol B(\boldsymbol x)$ with the minus, both absurd if $\boldsymbol B(\boldsymbol x)=\boldsymbol B$ is constant. But inverting $\boldsymbol B$ obviously restores the original Hamiltonian.
\end{proof}
\begin{lemma}
The operator $U_s=U\sigma_y$ is such that
\begin{equation}
U_s K U_s K=\pm I
\end{equation}
\end{lemma}
\begin{proof}
By direct computation
\begin{equation}
U_s K U_s K=U\sigma_y K U\sigma_y K= U \sigma_y U^{*}\sigma_y^{*} =- U \sigma_y U^{*}\sigma_y=-U^2
\end{equation}
where we used \eqref{complex}. This complete the proof.
\end{proof}
\noindent
In particular, taking $U=I$ in order to lie in the case of Lemma \ref{lem1}, we find the usual relation $\sigma_y K\sigma_y K=-I$.
%This result could appear disappointing because it limits the possibility of founding a difference between systems of particles with or without spin in the number of valid time reversal operators. In reality, Theorem \eqref{theo:2.10} is very
This means
%for the opposite reason: we started from the fundamental assumption of preserving the commutation relation of $\mathfrak{su}(2)$, without the unmotivated imposition $\mathcal{T}\boldsymbol\sigma \mathcal{T}^{-1}=-\boldsymbol\sigma$; we arrived to prove that,
that a class of time reversal operations, %\textit{i.e.}
the ones separately acting on the single particle subspaces, can \textit{always} be absorbed by an internal change of coordinates in spin space. Consequently, one finds that, \textit{contrary} to what is commonly believed, the presence of spin does not prevent
%at all for
time reversibility.
% On the second hand, the extra compatibility conditions can, in principle, reduce the number of suitable time reversal operator; if this difference were observed, it would possible to distinguish from statistical equalities, as the Onsager reciprocal ones, if the particles possess spin properties or not. This could be potentially fundamental in some situation in which a direct measure of the spin operator is not feasible. However, this discussion is not related to Classical Mechanics because it seems improbable that classical particles exhibit spin properties.
\subsection{Application of generalized TRI}
\label{sec:example}
Let us consider the calculation of Ref.\cite{bonella2017time}, concerning the diffusion tensor for a particle system in the presence of a constant magnetic field along the $z$ axis:
\begin{equation}
\label{3.54}
D_{\alpha\beta}\propto\langle v_i^\alpha(0) v_j^\beta (t)\rangle\;\;\;\; \forall i,j
\, ; \quad \alpha,\beta=x,y,z
\end{equation}
where $\alpha,\beta=x,y,z$ and $i,j$ are particle labels. In particular, Ref.\cite{rondom} proved that the
%\begin{prop}
%Considering a
correlator of velocities \eqref{3.54}
vanishes if there are two time reversal operations that map $p_j^\beta(t)$ respectively in $p_k^\gamma(-t)$ and $-p_k^\gamma(-t)$ (the case with $i=j$ and $\alpha=\beta$ simultaneously being excluded), while they act in the same way on $p_i^\alpha(t)$.
The proof simply observes that two expressions must be satisfied at once:
\begin{equation}
\begin{cases}
\langle v_i^\alpha(0) v_j^\beta (t)\rangle=(\pm 1)(+1)\langle v_i^\alpha(0) v_k^\gamma (-t)\rangle\\
\langle v_i^\alpha(0) v_j^\beta (t)\rangle=(\pm 1)(-1)\langle v_i^\alpha(0) v_k^\gamma(-t)\rangle
\end{cases}
\end{equation}
which only happens if the correlator vanishes.
The case $i=j$ and $\alpha=\beta$ is not included, since the autocorrelation $\langle v_i^\alpha(0) v_i^\alpha (t)\rangle$ is always mapped in another autocorrelation with plus sign. This example shows how
%We can lead back to this result all the cases showed in literature by Coretti et al.: they considered the cases $\alpha=\beta=\gamma$, $j=k$ and $\beta=\gamma$, $j=k$
%As a side note, we understand as the effect of TRI is very
practical the apparently abstract notion of TRI can be. In the present case, it allows a direct evaluation of transport coefficients.
For the classical case, we may now extend the result of \cite{bonella2017time} considering a more general form of magnetic field.
\begin{prop}
Let a system of coupled particles be subjected to an external magnetic field directed along the $z$-axis, $\boldsymbol B=B(x,y)\hat{\boldsymbol k}$. Assume $B(x,y)=B(x,-y)$ and $B(x,y)=B(y,x)$. Then, the diffusion tensor obeys:
\begin{equation}
D_{xy}=-D_{yx}
\end{equation}
\label{Prot29}
\end{prop}
\begin{proof}
Notice first that Eq.\eqref{3.22} is verified by a magnetic field like the one in this Proposition. For example, consider the time reversal operation on the single particle subspace defined as
\begin{equation}
\mathcal{M}(x,y,z,p_x,p_y,p_z)=(y,x,z,p_y,p_x,p_z)
\end{equation}
which implies
\begin{equation}
\mathcal{M}_m=
\begin{pmatrix}
0&1&0\\
1&0&0\\
0&0&1
\end{pmatrix}
\end{equation}
with the notation of Theorem \ref{theo:3.1.2}.
Equation \eqref{3.22} then holds for such a transformation and the magnetic field.
Applying the same to all the particle subspaces and computing the correlators we obtain
\begin{equation}
\langle v_i^x(0) v_j^y (t)\rangle=\langle v_i^y(0) v_j^x (-t)\rangle=-\langle v_i^y(0) v_j^x (t)\rangle
\end{equation}
that means $D_{xy}=-D_{yx}$.
\end{proof}
\noindent
Examples of magnetic fields that satisfy Proposition \ref{Prot29} are $B(x,y)=const$ and $B(x,y)=B(x^2+y^2)$, respectively the case of a constant magnetic field and of a magnetic field depending on the distance from the $z$ axis. The result cannot be obtained in term of the diagonal operations only, because the diagonal operations cannot disentangle the pairs of subscripts and superscripts $(i,x)$ and $(j,y)$.
%Ultimately, the found property of the diffusion tensor can be in principle easily investigated in an experiment and it surely deserves to be verified.
In this paper we extended the list of generalized time reversal transformations, to include operators that swap the coordinates of different particles. This is allowed by the fact that a point in a $6N$-dimensional space does not distinguish the nature of its single components. The formal definition of TRI does not prevent the use of this kind of operations. Their importance arises in particular when one studies systems of particles with the same mass and charge. These are important in the present context, since investigations of the Onsager relations rarely deal with more than two species of particles.
Moreover, we extended this treatment to the quantum mechanical framework, in the wake of the work of Ref.\cite{rondoquantu}, about the $8$ diagonal time reversal operations on the single particle subspace. In particular, we investigated the swap operation in a context of non-relativistic Quantum Mechanics.
We then defined generalized TRI for systems of particles with spin $1/2$, described by the Pauli equation. In previous works, this was considered out of reach \cite{rondoquantu}, because spin was believed to irremediably break TRI. On the contrary, we found combined sufficient conditions, concerning the generalized time reversal transformations and the form of the magnetic field, for the validity of TRI. That allow us to derive Onsager relations, as well as other relations requiring TRI, to hold in quantum systems coupled with an external magnetic field.
This is interesting not only because of the experimental and theoretical relevance of such statistical mechanical relations, but also for the method used: we took full advantage of the fact that relations such as Onsager's are statistical relations. This allows many different paths to the same result and, in particular, microreversibility results unnecessary for Onsager and Fluctuation Relations.
We then developed an application to the calculation of the diffusion tensor, which uses different time reversal symmetries to conclude that certain correlators identically vanish.
The investigation, however, is not over. As recalled in Ref.\cite{casa}, violations of Onsager relations, which would imply the existence non-dissipative currents, have never been observed. And indeed Ref.\cite{casa} proved that those relations hold in cases in which one would have expected they are violated. At the same time, or theory does not exhaust all possible cases.
%The second reflection is that it must be present a necessary condition almost always verified since it seems to be impossible to break Onsager relations, also for the just mentioned reason about dissipation.
%As a matter of fact, we used in the entire article the fact that Onsager relations are integral, then, leading the reasoning to the extremes, the equality \ref{eq:2.25} and its classical analogue could hold also for the particular functional form of the integrands.
%In conclusion, our idea is that the conditions under which it could happen are probably to be researched in some of the other hypothesis, for example the stationarity, that have to be made in the context of Linear Response theory.
Our results may also also pave the way to a new understanding of symmetries in quantum systems. The consequences of the generalized TRI have indeed rarely been fully investigated.
\vspace{1cm}\\
\small
\textbf{Conflict of interest}: The authors declare no conflict of interest.
\\
\textbf{Funding}: This research has been partially supported by Ministero dell’Istruzione, dell’Università e della Ricerca (MIUR) Grant No. E11G18000350001 “Dipartimenti di Eccellenza 2018-2022”.\\
% $ biblatex auxiliary file $
% $ biblatex bbl format version 3.1 $
% Do not modify the above lines!
%
% This is an auxiliary file used by the 'biblatex' package.
% This file may safely be deleted. It will be recreated by
% biber as required.
%
\begingroup
\makeatletter
\@ifundefined{ver@biblatex.sty}
{\@latex@error
{Missing 'biblatex' package}
{The bibliography requires the 'biblatex' package.}
\aftergroup\endinput}
{}
\endgroup
\refsection{0}
\datalist[entry]{none/global//global/global}
\entry{onsager1931reciprocal}{article}{}
\name{author}{1}{}{%
{{hash=d0bd03c548d04a59a960e8260c9474a6}{%
family={Onsager},
familyi={O\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{fullhash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{bibnamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authorbibnamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authornamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authorfullhash}{d0bd03c548d04a59a960e8260c9474a6}
\field{extraname}{1}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical review}
\field{number}{4}
\field{title}{Reciprocal relations in irreversible processes. I.}
\field{volume}{37}
\field{year}{1931}
\field{pages}{405}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/PhysRev.37.405
\endverb
\endentry
\entry{onsager1931reciprocal2}{article}{}
\name{author}{1}{}{%
{{hash=d0bd03c548d04a59a960e8260c9474a6}{%
family={Onsager},
familyi={O\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{fullhash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{bibnamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authorbibnamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authornamehash}{d0bd03c548d04a59a960e8260c9474a6}
\strng{authorfullhash}{d0bd03c548d04a59a960e8260c9474a6}
\field{extraname}{2}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical review}
\field{number}{12}
\field{title}{Reciprocal relations in irreversible processes. II.}
\field{volume}{38}
\field{year}{1931}
\field{pages}{2265}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/PhysRev.38.2265
\endverb
\endentry
\entry{callen1951irreversibility}{article}{}
\name{author}{2}{}{%
{{hash=fcbca29015ee1b510dd48a7df9eef49d}{%
family={Callen},
familyi={C\bibinitperiod},
given={Herbert\bibnamedelima B},
giveni={H\bibinitperiod\bibinitdelim B\bibinitperiod}}}%
{{hash=3e019c900ef2d944bede3277fa6efa48}{%
family={Welton},
familyi={W\bibinitperiod},
given={Theodore\bibnamedelima A},
giveni={T\bibinitperiod\bibinitdelim A\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{397dc76755edd22ba4b36c5c4dfb5487}
\strng{fullhash}{397dc76755edd22ba4b36c5c4dfb5487}
\strng{bibnamehash}{397dc76755edd22ba4b36c5c4dfb5487}
\strng{authorbibnamehash}{397dc76755edd22ba4b36c5c4dfb5487}
\strng{authornamehash}{397dc76755edd22ba4b36c5c4dfb5487}
\strng{authorfullhash}{397dc76755edd22ba4b36c5c4dfb5487}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review}
\field{number}{1}
\field{title}{Irreversibility and generalized noise}
\field{volume}{83}
\field{year}{1951}
\field{pages}{34}
\range{pages}{1}
\endentry
\entry{kubo1957statistical}{article}{}
\name{author}{1}{}{%
{{hash=a458ef503e5b3c9f91f4b64826611771}{%
family={Kubo},
familyi={K\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
}
\list{publisher}{1}{%
{The Physical Society of Japan}%
}
\strng{namehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{fullhash}{a458ef503e5b3c9f91f4b64826611771}
\strng{bibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorbibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authornamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorfullhash}{a458ef503e5b3c9f91f4b64826611771}
\field{extraname}{1}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Journal of the Physical Society of Japan}
\field{number}{6}
\field{title}{Statistical-mechanical theory of irreversible processes. I. General theory and simple applications to magnetic and conduction problems}
\field{volume}{12}
\field{year}{1957}
\field{pages}{570\bibrangedash 586}
\range{pages}{17}
\verb{doi}
\verb https://doi.org/10.1143/JPSJ.12.570
\endverb
\endentry
\entry{kubo1959some}{misc}{}
\name{author}{1}{}{%
{{hash=a458ef503e5b3c9f91f4b64826611771}{%
family={Kubo},
familyi={K\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Berlin: Springer) p}%
}
\strng{namehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{fullhash}{a458ef503e5b3c9f91f4b64826611771}
\strng{bibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorbibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authornamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorfullhash}{a458ef503e5b3c9f91f4b64826611771}
\field{extraname}{2}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Some Aspects of the Statistical-Mechanical Theory of Irreversible Processes (Lecture Notes in Theoretical Physics) ed WE Brittin and LG Dunham}
\field{year}{1959}
\endentry
\entry{kubo1966fluctuation}{article}{}
\name{author}{1}{}{%
{{hash=a458ef503e5b3c9f91f4b64826611771}{%
family={Kubo},
familyi={K\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
}
\list{publisher}{1}{%
{IOP Publishing}%
}
\strng{namehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{fullhash}{a458ef503e5b3c9f91f4b64826611771}
\strng{bibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorbibnamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authornamehash}{a458ef503e5b3c9f91f4b64826611771}
\strng{authorfullhash}{a458ef503e5b3c9f91f4b64826611771}
\field{extraname}{3}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Reports on progress in physics}
\field{number}{1}
\field{title}{The fluctuation-dissipation theorem}
\field{volume}{29}
\field{year}{1966}
\field{pages}{255}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1088/0034-4885/29/1/306
\endverb
\endentry
\entry{barbier:2018b}{article}{}
\name{author}{2}{}{%
{{hash=4f3cca50a72c2601468b289974ce942a}{%
family={Barbier},
familyi={B\bibinitperiod},
given={M},
giveni={M\bibinitperiod}}}%
{{hash=40d23c9b401ef5069b0ba77bd5f2c92c}{%
family={Gaspard},
familyi={G\bibinitperiod},
given={P},
giveni={P\bibinitperiod}}}%
}
\list{publisher}{1}{%
{{IOP} Publishing}%
}
\strng{namehash}{a7f6401ace9f420d99c218a3dea67c11}
\strng{fullhash}{a7f6401ace9f420d99c218a3dea67c11}
\strng{bibnamehash}{a7f6401ace9f420d99c218a3dea67c11}
\strng{authorbibnamehash}{a7f6401ace9f420d99c218a3dea67c11}
\strng{authornamehash}{a7f6401ace9f420d99c218a3dea67c11}
\strng{authorfullhash}{a7f6401ace9f420d99c218a3dea67c11}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{abstract}{Microreversibility rules the fluctuations of the currents flowing across open systems in nonequilibrium (or equilibrium) steady states. As a consequence, the statistical cumulants of the currents and their response coefficients at arbitrary orders in the deviations from equilibrium obey time-reversal symmetry relations, as established in quantum statistical mechanics. It is shown that some of these relations satisfy recurrences that can be simplified by means of the coefficients of Euler polynomials. This result allows us to systematically reduce the amount of independent quantities that need to be measured experimentally or computed theoretically in order to fully characterize the linear and nonlinear transport properties of general open systems. This reduction is shown to approach one half for quantities of arbitrarily high orders.}
\field{journaltitle}{Journal of Physics A: Mathematical and Theoretical}
\field{month}{7}
\field{number}{35}
\field{title}{Microreversibility, nonequilibrium current fluctuations, and response theory}
\field{volume}{51}
\field{year}{2018}
\field{pages}{355001}
\range{pages}{1}
\verb{doi}
\verb 10.1088/1751-8121/aad025
\endverb
\verb{urlraw}
\verb https://doi.org/10.1088%2F1751-8121%2Faad025
\endverb
\verb{url}
\verb https://doi.org/10.1088%2F1751-8121%2Faad025
\endverb
\endentry
\entry{Casimir}{article}{}
\name{author}{1}{}{%
{{hash=25657d16435d1747e444364246aded6f}{%
family={Casimir},
familyi={C\bibinitperiod},
given={H.\bibnamedelimi B.\bibnamedelimi G.},
giveni={H\bibinitperiod\bibinitdelim B\bibinitperiod\bibinitdelim G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{25657d16435d1747e444364246aded6f}
\strng{fullhash}{25657d16435d1747e444364246aded6f}
\strng{bibnamehash}{25657d16435d1747e444364246aded6f}
\strng{authorbibnamehash}{25657d16435d1747e444364246aded6f}
\strng{authornamehash}{25657d16435d1747e444364246aded6f}
\strng{authorfullhash}{25657d16435d1747e444364246aded6f}
\field{sortinit}{5}
\field{sortinithash}{20e9b4b0b173788c5dace24730f47d8c}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Reviews of Modern Physics}
\field{title}{On Onsager's Principle of Microscopic Reversibility}
\field{volume}{17}
\field{year}{1945}
\field{pages}{343}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/RevModPhys.17.343
\endverb
\endentry
\entry{landau1980statistical}{article}{}
\name{author}{2}{}{%
{{hash=d337245e299a90e4c87c62bbde6ccf56}{%
family={Landau},
familyi={L\bibinitperiod},
given={Lev\bibnamedelima Davidovich},
giveni={L\bibinitperiod\bibinitdelim D\bibinitperiod}}}%
{{hash=1e5737ddc6200d0118261e7bc577ab25}{%
family={Lifshitz},
familyi={L\bibinitperiod},
given={EM},
giveni={E\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Butterworth-Heinemann}%
}
\strng{namehash}{badb33d18900d06502a9cdd67eec7e54}
\strng{fullhash}{badb33d18900d06502a9cdd67eec7e54}
\strng{bibnamehash}{badb33d18900d06502a9cdd67eec7e54}
\strng{authorbibnamehash}{badb33d18900d06502a9cdd67eec7e54}
\strng{authornamehash}{badb33d18900d06502a9cdd67eec7e54}
\strng{authorfullhash}{badb33d18900d06502a9cdd67eec7e54}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Statistical Physics, Part 1: Volume 5}
\field{year}{1980}
\endentry
\entry{lax}{book}{}
\name{author}{1}{}{%
{{hash=d085abaaa21483d4f3b4d1dd98c620fd}{%
family={Lax},
familyi={L\bibinitperiod},
given={M.\bibnamedelimi J.},
giveni={M\bibinitperiod\bibinitdelim J\bibinitperiod}}}%
}
\list{publisher}{2}{%
{John Wiley}%
{Sons}%
}
\strng{namehash}{d085abaaa21483d4f3b4d1dd98c620fd}
\strng{fullhash}{d085abaaa21483d4f3b4d1dd98c620fd}
\strng{bibnamehash}{d085abaaa21483d4f3b4d1dd98c620fd}
\strng{authorbibnamehash}{d085abaaa21483d4f3b4d1dd98c620fd}
\strng{authornamehash}{d085abaaa21483d4f3b4d1dd98c620fd}
\strng{authorfullhash}{d085abaaa21483d4f3b4d1dd98c620fd}
\field{sortinit}{7}
\field{sortinithash}{108d0be1b1bee9773a1173443802c0a3}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Symmetry Principles in Solid State and Molecular Physics}
\field{year}{1972}
\endentry
\entry{sachs}{book}{}
\name{author}{1}{}{%
{{hash=74278f6c9fbd1e428e4ee2bafcafd36d}{%
family={Sachs},
familyi={S\bibinitperiod},
given={R.\bibnamedelimi G.},
giveni={R\bibinitperiod\bibinitdelim G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{University of Chicago Press}%
}
\strng{namehash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\strng{fullhash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\strng{bibnamehash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\strng{authorbibnamehash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\strng{authornamehash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\strng{authorfullhash}{74278f6c9fbd1e428e4ee2bafcafd36d}
\field{sortinit}{7}
\field{sortinithash}{108d0be1b1bee9773a1173443802c0a3}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{The physics of time reversal}
\field{year}{1987}
\endentry
\entry{de2013non}{book}{}
\name{author}{2}{}{%
{{hash=fa3d4f727edaace78d4ecb16389312d7}{%
family={De\bibnamedelima Groot},
familyi={D\bibinitperiod\bibinitdelim G\bibinitperiod},
given={Sybren\bibnamedelima Ruurds},
giveni={S\bibinitperiod\bibinitdelim R\bibinitperiod}}}%
{{hash=29ee052e4a82412b4fe6ce37b927559e}{%
family={Mazur},
familyi={M\bibinitperiod},
given={Peter},
giveni={P\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Courier Corporation}%
}
\strng{namehash}{7144860cc45ecfb348d19e0e565b7b91}
\strng{fullhash}{7144860cc45ecfb348d19e0e565b7b91}
\strng{bibnamehash}{7144860cc45ecfb348d19e0e565b7b91}
\strng{authorbibnamehash}{7144860cc45ecfb348d19e0e565b7b91}
\strng{authornamehash}{7144860cc45ecfb348d19e0e565b7b91}
\strng{authorfullhash}{7144860cc45ecfb348d19e0e565b7b91}
\field{sortinit}{7}
\field{sortinithash}{108d0be1b1bee9773a1173443802c0a3}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Non-equilibrium thermodynamics}
\field{year}{2013}
\endentry
\entry{toda1991statistical}{book}{}
\true{moreauthor}
\true{morelabelname}
\name{author}{4}{}{%
{{hash=3a0434af10feb11938569dd897018fda}{%
family={Toda},
familyi={T\bibinitperiod},
given={Morikazu},
giveni={M\bibinitperiod}}}%
{{hash=3db80a54dbc22f6e0f8ef62d81d5e4c4}{%
family={Kubo},
familyi={K\bibinitperiod},
given={Ryogo},
giveni={R\bibinitperiod}}}%
{{hash=feccb734402ec8955bf61b1e52a8ac2d}{%
family={Saitō},
familyi={S\bibinitperiod},
given={Nobuhiko},
giveni={N\bibinitperiod}}}%
{{hash=7b9b672197184cbd6a793fc4ffb2a30b}{%
family={Hashitsume},
familyi={H\bibinitperiod},
given={Natsuki},
giveni={N\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Springer Science \& Business Media}%
}
\strng{namehash}{77e7beef7c4887533c7ea5a468d5bf34}
\strng{fullhash}{5486e5c80004495dcc7c0b67fa7475bd}
\strng{bibnamehash}{5486e5c80004495dcc7c0b67fa7475bd}
\strng{authorbibnamehash}{5486e5c80004495dcc7c0b67fa7475bd}
\strng{authornamehash}{77e7beef7c4887533c7ea5a468d5bf34}
\strng{authorfullhash}{5486e5c80004495dcc7c0b67fa7475bd}
\field{sortinit}{7}
\field{sortinithash}{108d0be1b1bee9773a1173443802c0a3}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Statistical physics II: nonequilibrium statistical mechanics}
\field{volume}{2}
\field{year}{1991}
\endentry
\entry{robnik1986false}{article}{}
\name{author}{2}{}{%
{{hash=eb4534ce773d06343c38630624d52dad}{%
family={Robnik},
familyi={R\bibinitperiod},
given={Marko},
giveni={M\bibinitperiod}}}%
{{hash=d50a9e3fa0fbfb726741710acd0dd1f3}{%
family={Berry},
familyi={B\bibinitperiod},
given={Michael\bibnamedelima V},
giveni={M\bibinitperiod\bibinitdelim V\bibinitperiod}}}%
}
\list{publisher}{1}{%
{IOP Publishing}%
}
\strng{namehash}{d621fba9fca54f65d6f9b23144f4e6e2}
\strng{fullhash}{d621fba9fca54f65d6f9b23144f4e6e2}
\strng{bibnamehash}{d621fba9fca54f65d6f9b23144f4e6e2}
\strng{authorbibnamehash}{d621fba9fca54f65d6f9b23144f4e6e2}
\strng{authornamehash}{d621fba9fca54f65d6f9b23144f4e6e2}
\strng{authorfullhash}{d621fba9fca54f65d6f9b23144f4e6e2}
\field{sortinit}{8}
\field{sortinithash}{a231b008ebf0ecbe0b4d96dcc159445f}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Journal of Physics A: Mathematical and General}
\field{number}{5}
\field{title}{False time-reversal violation and energy level statistics: the role of anti-unitary symmetry}
\field{volume}{19}
\field{year}{1986}
\field{pages}{669}
\range{pages}{1}
\endentry
\entry{jacquod2012onsager}{article}{}
\name{author}{4}{}{%
{{hash=1a16322f4c9a803c796dc27074237103}{%
family={Jacquod},
familyi={J\bibinitperiod},
given={P.},
giveni={P\bibinitperiod}}}%
{{hash=750ea7bb31b61df2b5878ffc7fc386ec}{%
family={Whitney},
familyi={W\bibinitperiod},
given={R.\bibnamedelimi S.},
giveni={R\bibinitperiod\bibinitdelim S\bibinitperiod}}}%
{{hash=78b9d93e1ce1da1d165bfd6353ddf2c2}{%
family={Meair},
familyi={M\bibinitperiod},
given={J.},
giveni={J\bibinitperiod}}}%
{{hash=9508f25a74832b8ae2635b73382d65ee}{%
family={Büttiker},
familyi={B\bibinitperiod},
given={M.},
giveni={M\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{0a0f67f49a1a656d7428508cc7e30542}
\strng{fullhash}{6c3c856aae7ec09ce4119e5781411124}
\strng{bibnamehash}{6c3c856aae7ec09ce4119e5781411124}
\strng{authorbibnamehash}{6c3c856aae7ec09ce4119e5781411124}
\strng{authornamehash}{0a0f67f49a1a656d7428508cc7e30542}
\strng{authorfullhash}{6c3c856aae7ec09ce4119e5781411124}
\field{sortinit}{9}
\field{sortinithash}{0a5ebc79d83c96b6579069544c73c7d4}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review B}
\field{number}{15}
\field{title}{Onsager relations in coupled electric, thermoelectric, and spin transport: The tenfold way}
\field{volume}{86}
\field{year}{2012}
\field{pages}{155118}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/PhysRevB.86.155118
\endverb
\endentry
\entry{j2007statistical}{book}{}
\name{author}{2}{}{%
{{hash=ba7a1745cce9980667fd88c2ddfd510f}{%
family={J\bibnamedelima Evans},
familyi={J\bibinitperiod\bibinitdelim E\bibinitperiod},
given={Denis},
giveni={D\bibinitperiod}}}%
{{hash=11c046b9d4175e18bd4236ff5a97b688}{%
family={P\bibnamedelima Morriss},
familyi={P\bibinitperiod\bibinitdelim M\bibinitperiod},
given={Gary},
giveni={G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{ANU Press}%
}
\strng{namehash}{78fcff0634959c9cfc51779f861e8c43}
\strng{fullhash}{78fcff0634959c9cfc51779f861e8c43}
\strng{bibnamehash}{78fcff0634959c9cfc51779f861e8c43}
\strng{authorbibnamehash}{78fcff0634959c9cfc51779f861e8c43}
\strng{authornamehash}{78fcff0634959c9cfc51779f861e8c43}
\strng{authorfullhash}{78fcff0634959c9cfc51779f861e8c43}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Statistical mechanics of nonequilbrium liquids}
\field{year}{2007}
\endentry
\entry{rondoe}{article}{}
\name{author}{3}{}{%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
{{hash=f73f53362e53eb575dfb5ca8b9761b3f}{%
family={Ciccotti},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{IOP Publishing}%
}
\strng{namehash}{28a1415652e7e40b79468ef0adf12630}
\strng{fullhash}{28a1415652e7e40b79468ef0adf12630}
\strng{bibnamehash}{28a1415652e7e40b79468ef0adf12630}
\strng{authorbibnamehash}{28a1415652e7e40b79468ef0adf12630}
\strng{authornamehash}{28a1415652e7e40b79468ef0adf12630}
\strng{authorfullhash}{28a1415652e7e40b79468ef0adf12630}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{EPL (Europhysics Letters)}
\field{number}{6}
\field{title}{Time reversal symmetry in time-dependent correlation functions for systems in a constant magnetic field}
\field{volume}{108}
\field{year}{2015}
\field{pages}{60004}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1209/0295-5075/108/60004
\endverb
\endentry
\entry{rondom}{article}{}
\name{author}{4}{}{%
{{hash=7b6abee3731c19dda370cee646175993}{%
family={Coretti},
familyi={C\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
{{hash=f73f53362e53eb575dfb5ca8b9761b3f}{%
family={Ciccotti},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Taylor \& Francis}%
}
\strng{namehash}{cadb8d7d2901af24375b2baea1c2dece}
\strng{fullhash}{ec3ecbe8e37340eabc089dd8a8a1655c}
\strng{bibnamehash}{ec3ecbe8e37340eabc089dd8a8a1655c}
\strng{authorbibnamehash}{ec3ecbe8e37340eabc089dd8a8a1655c}
\strng{authornamehash}{cadb8d7d2901af24375b2baea1c2dece}
\strng{authorfullhash}{ec3ecbe8e37340eabc089dd8a8a1655c}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Molecular Physics}
\field{number}{21-22}
\field{title}{Time reversal and symmetries of time correlation functions}
\field{volume}{116}
\field{year}{2018}
\field{pages}{3097\bibrangedash 3103}
\range{pages}{7}
\verb{doi}
\verb https://doi.org/10.1080/00268976.2018.1464674
\endverb
\endentry
\entry{bonella2017time}{article}{}
\name{author}{4}{}{%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
{{hash=7b6abee3731c19dda370cee646175993}{%
family={Coretti},
familyi={C\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
{{hash=f73f53362e53eb575dfb5ca8b9761b3f}{%
family={Ciccotti},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{806e51a5109843b53f60fcd248aa8de6}
\strng{fullhash}{fac572e8fa40e752879545061913429c}
\strng{bibnamehash}{fac572e8fa40e752879545061913429c}
\strng{authorbibnamehash}{fac572e8fa40e752879545061913429c}
\strng{authornamehash}{806e51a5109843b53f60fcd248aa8de6}
\strng{authorfullhash}{fac572e8fa40e752879545061913429c}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review E}
\field{number}{1}
\field{title}{Time-reversal symmetry for systems in a constant external magnetic field}
\field{volume}{96}
\field{year}{2017}
\field{pages}{012160}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/PhysRevE.96.012160
\endverb
\endentry
\entry{rondoquantu}{article}{}
\name{author}{3}{}{%
{{hash=12a26c762dcfc045a1dd292265e43030}{%
family={De\bibnamedelima Gregorio},
familyi={D\bibinitperiod\bibinitdelim G\bibinitperiod},
given={P.},
giveni={P\bibinitperiod}}}%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Multidisciplinary Digital Publishing Institute}%
}
\strng{namehash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\strng{fullhash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\strng{bibnamehash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\strng{authorbibnamehash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\strng{authornamehash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\strng{authorfullhash}{9ab6c0118d7bfe363eb9cc5e1ab727e9}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Symmetry}
\field{number}{7}
\field{title}{Quantum Correlations under Time Reversal and Incomplete Parity Transformations in the Presence of a Constant Magnetic Field}
\field{volume}{9}
\field{year}{2017}
\field{pages}{120}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.3390/sym9070120
\endverb
\endentry
\entry{rondoFT1}{article}{}
\name{author}{3}{}{%
{{hash=7b6abee3731c19dda370cee646175993}{%
family={Coretti},
familyi={C\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{d04828505a8699aca46fae838d58001a}
\strng{fullhash}{d04828505a8699aca46fae838d58001a}
\strng{bibnamehash}{d04828505a8699aca46fae838d58001a}
\strng{authorbibnamehash}{d04828505a8699aca46fae838d58001a}
\strng{authornamehash}{d04828505a8699aca46fae838d58001a}
\strng{authorfullhash}{d04828505a8699aca46fae838d58001a}
\field{extraname}{1}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review E}
\field{number}{3}
\field{title}{Fluctuation relations for systems in a constant magnetic field}
\field{volume}{102}
\field{year}{2020}
\field{pages}{030101}
\range{pages}{1}
\endentry
\entry{rondoFT2}{article}{}
\name{author}{3}{}{%
{{hash=7b6abee3731c19dda370cee646175993}{%
family={Coretti},
familyi={C\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
{{hash=c17d63a8a812d35c6e3867ff45d1c225}{%
family={Bonella},
familyi={B\bibinitperiod},
given={S.},
giveni={S\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Multidisciplinary Digital Publishing Institute}%
}
\strng{namehash}{d04828505a8699aca46fae838d58001a}
\strng{fullhash}{d04828505a8699aca46fae838d58001a}
\strng{bibnamehash}{d04828505a8699aca46fae838d58001a}
\strng{authorbibnamehash}{d04828505a8699aca46fae838d58001a}
\strng{authornamehash}{d04828505a8699aca46fae838d58001a}
\strng{authorfullhash}{d04828505a8699aca46fae838d58001a}
\field{extraname}{2}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Entropy}
\field{number}{2}
\field{title}{Fluctuation relations for dissipative systems in constant external magnetic field: theory and molecular dynamics simulations}
\field{volume}{23}
\field{year}{2021}
\field{pages}{146}
\range{pages}{1}
\endentry
\entry{carbone2020necessary}{article}{}
\name{author}{2}{}{%
{{hash=0bcddae384ed8a62d0600c434373f5ce}{%
family={Carbone},
familyi={C\bibinitperiod},
given={D.},
giveni={D\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Multidisciplinary Digital Publishing Institute}%
}
\strng{namehash}{72869a41842b39ba22abc4a95981dfe6}
\strng{fullhash}{72869a41842b39ba22abc4a95981dfe6}
\strng{bibnamehash}{72869a41842b39ba22abc4a95981dfe6}
\strng{authorbibnamehash}{72869a41842b39ba22abc4a95981dfe6}
\strng{authornamehash}{72869a41842b39ba22abc4a95981dfe6}
\strng{authorfullhash}{72869a41842b39ba22abc4a95981dfe6}
\field{sortinit}{1}
\field{sortinithash}{4f6aaa89bab872aa0999fec09ff8e98a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Symmetry}
\field{number}{8}
\field{title}{Necessary and Sufficient Conditions for Time Reversal Symmetry in Presence of Magnetic Fields}
\field{volume}{12}
\field{year}{2020}
\field{pages}{1336}
\range{pages}{1}
\endentry
\entry{callen1948application}{article}{}
\name{author}{1}{}{%
{{hash=fcbca29015ee1b510dd48a7df9eef49d}{%
family={Callen},
familyi={C\bibinitperiod},
given={Herbert\bibnamedelima B},
giveni={H\bibinitperiod\bibinitdelim B\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{fcbca29015ee1b510dd48a7df9eef49d}
\strng{fullhash}{fcbca29015ee1b510dd48a7df9eef49d}
\strng{bibnamehash}{fcbca29015ee1b510dd48a7df9eef49d}
\strng{authorbibnamehash}{fcbca29015ee1b510dd48a7df9eef49d}
\strng{authornamehash}{fcbca29015ee1b510dd48a7df9eef49d}
\strng{authorfullhash}{fcbca29015ee1b510dd48a7df9eef49d}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review}
\field{number}{11}
\field{title}{The application of Onsager's reciprocal relations to thermoelectric, thermomagnetic, and galvanomagnetic effects}
\field{volume}{73}
\field{year}{1948}
\field{pages}{1349}
\range{pages}{1}
\endentry
\entry{miller1960thermodynamics}{article}{}
\name{author}{1}{}{%
{{hash=251f65f9346b28f5bdaa2002cb30ebf8}{%
family={Miller},
familyi={M\bibinitperiod},
given={Donald\bibnamedelima G},
giveni={D\bibinitperiod\bibinitdelim G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{ACS Publications}%
}
\strng{namehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{fullhash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{bibnamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authorbibnamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authornamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authorfullhash}{251f65f9346b28f5bdaa2002cb30ebf8}
\field{extraname}{1}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Chemical Reviews}
\field{number}{1}
\field{title}{Thermodynamics of Irreversible Processes. The Experimental Verification of the Onsager Reciprocal Relations.}
\field{volume}{60}
\field{year}{1960}
\field{pages}{15\bibrangedash 37}
\range{pages}{23}
\endentry
\entry{benenti2017fundamental}{article}{}
\name{author}{4}{}{%
{{hash=9c17dbdc8c60d7e626801b6c0d0978fc}{%
family={Benenti},
familyi={B\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=26d01de871b5bb135cd0a7652c46fbaa}{%
family={Casati},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=fd843c4774e2b9fe607f80fe64207e30}{%
family={Saito},
familyi={S\bibinitperiod},
given={K.},
giveni={K\bibinitperiod}}}%
{{hash=750ea7bb31b61df2b5878ffc7fc386ec}{%
family={Whitney},
familyi={W\bibinitperiod},
given={R.\bibnamedelimi S.},
giveni={R\bibinitperiod\bibinitdelim S\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Elsevier}%
}
\strng{namehash}{1bfe7e0432c1cd2edfa3db028d54a972}
\strng{fullhash}{94019ecbd17fac12b0fb2fa71c0c081c}
\strng{bibnamehash}{94019ecbd17fac12b0fb2fa71c0c081c}
\strng{authorbibnamehash}{94019ecbd17fac12b0fb2fa71c0c081c}
\strng{authornamehash}{1bfe7e0432c1cd2edfa3db028d54a972}
\strng{authorfullhash}{94019ecbd17fac12b0fb2fa71c0c081c}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physics Reports}
\field{title}{Fundamental aspects of steady-state conversion of heat to work at the nanoscale}
\field{volume}{694}
\field{year}{2017}
\field{pages}{1\bibrangedash 124}
\range{pages}{124}
\endentry
\entry{benenti2020power}{article}{}
\name{author}{3}{}{%
{{hash=9c17dbdc8c60d7e626801b6c0d0978fc}{%
family={Benenti},
familyi={B\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=26d01de871b5bb135cd0a7652c46fbaa}{%
family={Casati},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=3702f1c146e64a6c91fc47b81ea59fc7}{%
family={Wang},
familyi={W\bibinitperiod},
given={J.},
giveni={J\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{c6aaad0405f9079b79f533780be5dc3a}
\strng{fullhash}{c6aaad0405f9079b79f533780be5dc3a}
\strng{bibnamehash}{c6aaad0405f9079b79f533780be5dc3a}
\strng{authorbibnamehash}{c6aaad0405f9079b79f533780be5dc3a}
\strng{authornamehash}{c6aaad0405f9079b79f533780be5dc3a}
\strng{authorfullhash}{c6aaad0405f9079b79f533780be5dc3a}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review E}
\field{number}{4}
\field{title}{Power, efficiency, and fluctuations in steady-state heat engines}
\field{volume}{102}
\field{year}{2020}
\field{pages}{040103}
\range{pages}{1}
\endentry
\entry{casa}{article}{}
\name{author}{4}{}{%
{{hash=9da0bb83b4fbfd38e4e4b5fad2f3e0b2}{%
family={Luo},
familyi={L\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
{{hash=9c17dbdc8c60d7e626801b6c0d0978fc}{%
family={Benenti},
familyi={B\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=26d01de871b5bb135cd0a7652c46fbaa}{%
family={Casati},
familyi={C\bibinitperiod},
given={G.},
giveni={G\bibinitperiod}}}%
{{hash=3702f1c146e64a6c91fc47b81ea59fc7}{%
family={Wang},
familyi={W\bibinitperiod},
given={J.},
giveni={J\bibinitperiod}}}%
}
\list{publisher}{1}{%
{American Physical Society}%
}
\strng{namehash}{879ebdf5753605c6041738264476ed18}
\strng{fullhash}{4f3997bdc844556be93c19db8cd6ccc9}
\strng{bibnamehash}{4f3997bdc844556be93c19db8cd6ccc9}
\strng{authorbibnamehash}{4f3997bdc844556be93c19db8cd6ccc9}
\strng{authornamehash}{879ebdf5753605c6041738264476ed18}
\strng{authorfullhash}{4f3997bdc844556be93c19db8cd6ccc9}
\field{sortinit}{2}
\field{sortinithash}{8b555b3791beccb63322c22f3320aa9a}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{issue}{2}
\field{journaltitle}{Phys. Rev. Research}
\field{month}{4}
\field{title}{Onsager reciprocal relations with broken time-reversal symmetry}
\field{volume}{2}
\field{year}{2020}
\field{pages}{022009}
\range{pages}{1}
\verb{doi}
\verb https://doi.org/10.1103/PhysRevResearch.2.022009
\endverb
\endentry
\entry{evans1993probability}{article}{}
\name{author}{3}{}{%
{{hash=daa3bd00ef066c0db475467614677bce}{%
family={Evans},
familyi={E\bibinitperiod},
given={Denis\bibnamedelima J},
giveni={D\bibinitperiod\bibinitdelim J\bibinitperiod}}}%
{{hash=6961c3e59aeba341562858205762ad5e}{%
family={Cohen},
familyi={C\bibinitperiod},
given={Ezechiel\bibnamedelimb Godert\bibnamedelima David},
giveni={E\bibinitperiod\bibinitdelim G\bibinitperiod\bibinitdelim D\bibinitperiod}}}%
{{hash=11e86db7daa418d391aa13749fe4602b}{%
family={Morriss},
familyi={M\bibinitperiod},
given={Gary\bibnamedelima P},
giveni={G\bibinitperiod\bibinitdelim P\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{9a82804b56f4aeb2253d3d904abb1d8f}
\strng{fullhash}{9a82804b56f4aeb2253d3d904abb1d8f}
\strng{bibnamehash}{9a82804b56f4aeb2253d3d904abb1d8f}
\strng{authorbibnamehash}{9a82804b56f4aeb2253d3d904abb1d8f}
\strng{authornamehash}{9a82804b56f4aeb2253d3d904abb1d8f}
\strng{authorfullhash}{9a82804b56f4aeb2253d3d904abb1d8f}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical review letters}
\field{number}{15}
\field{title}{Probability of second law violations in shearing steady states}
\field{volume}{71}
\field{year}{1993}
\field{pages}{2401}
\range{pages}{1}
\endentry
\entry{miller1973onsager}{incollection}{}
\name{author}{1}{}{%
{{hash=251f65f9346b28f5bdaa2002cb30ebf8}{%
family={Miller},
familyi={M\bibinitperiod},
given={Donald\bibnamedelima G},
giveni={D\bibinitperiod\bibinitdelim G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Springer}%
}
\strng{namehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{fullhash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{bibnamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authorbibnamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authornamehash}{251f65f9346b28f5bdaa2002cb30ebf8}
\strng{authorfullhash}{251f65f9346b28f5bdaa2002cb30ebf8}
\field{extraname}{2}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{booktitle}{Foundations of Continuum Thermodynamics}
\field{title}{The Onsager relations; experimental evidence}
\field{year}{1973}
\field{pages}{185\bibrangedash 214}
\range{pages}{30}
\endentry
\entry{letey1969movement}{article}{}
\name{author}{2}{}{%
{{hash=c539e5ec540ddafb21d1d6cad2f43701}{%
family={Letey},
familyi={L\bibinitperiod},
given={John},
giveni={J\bibinitperiod}}}%
{{hash=c25d6237b2f6df1fdc18e3036465d147}{%
family={Kemper},
familyi={K\bibinitperiod},
given={WD},
giveni={W\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Wiley Online Library}%
}
\strng{namehash}{e9a34323d12eeebaa16c6b2a564692f4}
\strng{fullhash}{e9a34323d12eeebaa16c6b2a564692f4}
\strng{bibnamehash}{e9a34323d12eeebaa16c6b2a564692f4}
\strng{authorbibnamehash}{e9a34323d12eeebaa16c6b2a564692f4}
\strng{authornamehash}{e9a34323d12eeebaa16c6b2a564692f4}
\strng{authorfullhash}{e9a34323d12eeebaa16c6b2a564692f4}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Soil Science Society of America Journal}
\field{number}{1}
\field{title}{Movement of water and salt through a clay-water system: experimental verification of Onsager reciprocal relation}
\field{volume}{33}
\field{year}{1969}
\field{pages}{25\bibrangedash 29}
\range{pages}{5}
\endentry
\entry{monroe2006onsager}{article}{}
\name{author}{2}{}{%
{{hash=86a4e450e705fe23f89bf3541effd4ee}{%
family={Monroe},
familyi={M\bibinitperiod},
given={Charles\bibnamedelima W},
giveni={C\bibinitperiod\bibinitdelim W\bibinitperiod}}}%
{{hash=887e883f31937e46bbf9eb74a2db896f}{%
family={Newman},
familyi={N\bibinitperiod},
given={John},
giveni={J\bibinitperiod}}}%
}
\list{publisher}{1}{%
{ACS Publications}%
}
\strng{namehash}{550cc76e96fc7eac09c07528889b662d}
\strng{fullhash}{550cc76e96fc7eac09c07528889b662d}
\strng{bibnamehash}{550cc76e96fc7eac09c07528889b662d}
\strng{authorbibnamehash}{550cc76e96fc7eac09c07528889b662d}
\strng{authornamehash}{550cc76e96fc7eac09c07528889b662d}
\strng{authorfullhash}{550cc76e96fc7eac09c07528889b662d}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Industrial \& engineering chemistry research}
\field{number}{15}
\field{title}{Onsager Reciprocal Relations for Stefan- Maxwell Diffusion}
\field{volume}{45}
\field{year}{2006}
\field{pages}{5361\bibrangedash 5367}
\range{pages}{7}
\endentry
\entry{gaeta1994phonons}{article}{}
\name{author}{4}{}{%
{{hash=58efc532704b35a78e539750aec71105}{%
family={Gaeta},
familyi={G\bibinitperiod},
given={FS},
giveni={F\bibinitperiod}}}%
{{hash=337c1b11214b0618efa042c58aed9bff}{%
family={Albanese},
familyi={A\bibinitperiod},
given={C},
giveni={C\bibinitperiod}}}%
{{hash=dba9347149dd4216d3b07b86e6ba6a60}{%
family={Mita},
familyi={M\bibinitperiod},
given={DG},
giveni={D\bibinitperiod}}}%
{{hash=7cf501422b20b43c18b2b3ea5fceb79e}{%
family={Peluso},
familyi={P\bibinitperiod},
given={F},
giveni={F\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{5c8a8db7736ef8929d7fba0357c3eb79}
\strng{fullhash}{9d911f73b55f7e559bc32167e5302e0d}
\strng{bibnamehash}{9d911f73b55f7e559bc32167e5302e0d}
\strng{authorbibnamehash}{9d911f73b55f7e559bc32167e5302e0d}
\strng{authornamehash}{5c8a8db7736ef8929d7fba0357c3eb79}
\strng{authorfullhash}{9d911f73b55f7e559bc32167e5302e0d}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review E}
\field{number}{1}
\field{title}{Phonons in liquids, Onsager’s reciprocal relations, and the heats of transport}
\field{volume}{49}
\field{year}{1994}
\field{pages}{433}
\range{pages}{1}
\endentry
\entry{miyatani1981experimental}{article}{}
\name{author}{1}{}{%
{{hash=012c1b083326e301d453483f9b43b17e}{%
family={Miyatani},
familyi={M\bibinitperiod},
given={S},
giveni={S\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Elsevier}%
}
\strng{namehash}{012c1b083326e301d453483f9b43b17e}
\strng{fullhash}{012c1b083326e301d453483f9b43b17e}
\strng{bibnamehash}{012c1b083326e301d453483f9b43b17e}
\strng{authorbibnamehash}{012c1b083326e301d453483f9b43b17e}
\strng{authornamehash}{012c1b083326e301d453483f9b43b17e}
\strng{authorfullhash}{012c1b083326e301d453483f9b43b17e}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Solid State Communications}
\field{number}{4}
\field{title}{Experimental evidence for the Onsager reciprocal relation in mixed conduction}
\field{volume}{38}
\field{year}{1981}
\field{pages}{257\bibrangedash 259}
\range{pages}{3}
\endentry
\entry{rowley1978dufour}{article}{}
\name{author}{2}{}{%
{{hash=53b90e8b59bca9c1916da67f1464164f}{%
family={Rowley},
familyi={R\bibinitperiod},
given={Richard\bibnamedelima L},
giveni={R\bibinitperiod\bibinitdelim L\bibinitperiod}}}%
{{hash=a0ab7e4957913066566109a6b9c81a68}{%
family={Horne},
familyi={H\bibinitperiod},
given={Frederick\bibnamedelima H},
giveni={F\bibinitperiod\bibinitdelim H\bibinitperiod}}}%
}
\list{publisher}{1}{%
{American Institute of Physics}%
}
\strng{namehash}{35ba171f837afeb047551ccd53903fc2}
\strng{fullhash}{35ba171f837afeb047551ccd53903fc2}
\strng{bibnamehash}{35ba171f837afeb047551ccd53903fc2}
\strng{authorbibnamehash}{35ba171f837afeb047551ccd53903fc2}
\strng{authornamehash}{35ba171f837afeb047551ccd53903fc2}
\strng{authorfullhash}{35ba171f837afeb047551ccd53903fc2}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{The Journal of Chemical Physics}
\field{number}{1}
\field{title}{The Dufour effect. II. Experimental confirmation of the Onsager heat--mass reciprocal relation for a binary liquid mixture}
\field{volume}{68}
\field{year}{1978}
\field{pages}{325\bibrangedash 326}
\range{pages}{2}
\endentry
\entry{matthews2014experimental}{article}{}
\name{author}{5}{}{%
{{hash=bfae20ee430d9266f32b92f7120a5cb7}{%
family={Matthews},
familyi={M\bibinitperiod},
given={J},
giveni={J\bibinitperiod}}}%
{{hash=79c6b479154554aa435189556224a68a}{%
family={Battista},
familyi={B\bibinitperiod},
given={Francesca},
giveni={F\bibinitperiod}}}%
{{hash=29bf70e2644edaa8bd15f172082bc09a}{%
family={Sánchez},
familyi={S\bibinitperiod},
given={David},
giveni={D\bibinitperiod}}}%
{{hash=b32be5f0a9af555416ea31fbd02751c2}{%
family={Samuelsson},
familyi={S\bibinitperiod},
given={Peter},
giveni={P\bibinitperiod}}}%
{{hash=010f4003e3faab5ad675994b36870e33}{%
family={Linke},
familyi={L\bibinitperiod},
given={Heiner},
giveni={H\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{282b65ec2b82ec10246a27dbefaa3f60}
\strng{fullhash}{b20da5edb6fa963728290e6d4f6f732a}
\strng{bibnamehash}{b20da5edb6fa963728290e6d4f6f732a}
\strng{authorbibnamehash}{b20da5edb6fa963728290e6d4f6f732a}
\strng{authornamehash}{282b65ec2b82ec10246a27dbefaa3f60}
\strng{authorfullhash}{b20da5edb6fa963728290e6d4f6f732a}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical Review B}
\field{number}{16}
\field{title}{Experimental verification of reciprocity relations in quantum thermoelectric transport}
\field{volume}{90}
\field{year}{2014}
\field{pages}{165428}
\range{pages}{1}
\endentry
\entry{kimura2007room}{article}{}
\name{author}{5}{}{%
{{hash=bb541e65905b787395bf77dc2ea601e0}{%
family={Kimura},
familyi={K\bibinitperiod},
given={Takashi},
giveni={T\bibinitperiod}}}%
{{hash=79a3c96959fec04c684a0fe22dfe14a2}{%
family={Otani},
familyi={O\bibinitperiod},
given={Y},
giveni={Y\bibinitperiod}}}%
{{hash=ca8df0535bc96db6c2cb72a627f3b5ed}{%
family={Sato},
familyi={S\bibinitperiod},
given={T},
giveni={T\bibinitperiod}}}%
{{hash=f87163df89d207b686b75a6a76b6ffbe}{%
family={Takahashi},
familyi={T\bibinitperiod},
given={S},
giveni={S\bibinitperiod}}}%
{{hash=baabef28db698021fc3047e91d289f68}{%
family={Maekawa},
familyi={M\bibinitperiod},
given={S},
giveni={S\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{388be0b0c34c0e9d1ccf32299c7b9b38}
\strng{fullhash}{ba2f6bf16932cc399d245f249d165d3c}
\strng{bibnamehash}{ba2f6bf16932cc399d245f249d165d3c}
\strng{authorbibnamehash}{ba2f6bf16932cc399d245f249d165d3c}
\strng{authornamehash}{388be0b0c34c0e9d1ccf32299c7b9b38}
\strng{authorfullhash}{ba2f6bf16932cc399d245f249d165d3c}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical review letters}
\field{number}{15}
\field{title}{Room-temperature reversible spin Hall effect}
\field{volume}{98}
\field{year}{2007}
\field{pages}{156601}
\range{pages}{1}
\endentry
\entry{sola2019experimental}{article}{}
\name{author}{5}{}{%
{{hash=e7da0ecc7ea8cef471f4a00cbf415605}{%
family={Sola},
familyi={S\bibinitperiod},
given={Alessandro},
giveni={A\bibinitperiod}}}%
{{hash=e10bc8873652fb8c218919cd93028e36}{%
family={Basso},
familyi={B\bibinitperiod},
given={Vittorio},
giveni={V\bibinitperiod}}}%
{{hash=ff397336f67fe28b6dc8fa0189b2f3f9}{%
family={Kuepferling},
familyi={K\bibinitperiod},
given={Michaela},
giveni={M\bibinitperiod}}}%
{{hash=46e823c141ddb639ed18597d12598b0f}{%
family={Dubs},
familyi={D\bibinitperiod},
given={Carsten},
giveni={C\bibinitperiod}}}%
{{hash=e01afbd9d786fd2f7e0de1e5f6be67ed}{%
family={Pasquale},
familyi={P\bibinitperiod},
given={Massimo},
giveni={M\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Nature Publishing Group}%
}
\strng{namehash}{fe99ad086d96d90abc94014c9d2dc2f3}
\strng{fullhash}{da5234ad72dc000e276827bbd5f220a1}
\strng{bibnamehash}{da5234ad72dc000e276827bbd5f220a1}
\strng{authorbibnamehash}{da5234ad72dc000e276827bbd5f220a1}
\strng{authornamehash}{fe99ad086d96d90abc94014c9d2dc2f3}
\strng{authorfullhash}{da5234ad72dc000e276827bbd5f220a1}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Scientific reports}
\field{number}{1}
\field{title}{Experimental proof of the reciprocal relation between spin Peltier and spin Seebeck effects in a bulk YIG/Pt bilayer}
\field{volume}{9}
\field{year}{2019}
\field{pages}{1\bibrangedash 8}
\range{pages}{8}
\endentry
\entry{marklund2007dynamics}{article}{}
\name{author}{2}{}{%
{{hash=ad18ef6e86add2530be38235aecfc9df}{%
family={Marklund},
familyi={M\bibinitperiod},
given={Mattias},
giveni={M\bibinitperiod}}}%
{{hash=0151ec19b8734e8d9782fd64fde6d31e}{%
family={Brodin},
familyi={B\bibinitperiod},
given={Gert},
giveni={G\bibinitperiod}}}%
}
\list{publisher}{1}{%
{APS}%
}
\strng{namehash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\strng{fullhash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\strng{bibnamehash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\strng{authorbibnamehash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\strng{authornamehash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\strng{authorfullhash}{27ff19c757e1ddbbfaefcf23f2263d7a}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physical review letters}
\field{number}{2}
\field{title}{Dynamics of Spin-1 2 quantum plasmas}
\field{volume}{98}
\field{year}{2007}
\field{pages}{025001}
\range{pages}{1}
\endentry
\entry{anderson1994canonical}{article}{}
\name{author}{1}{}{%
{{hash=076c274796c1620dbfad5894924750da}{%
family={Anderson},
familyi={A\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Elsevier}%
}
\strng{namehash}{076c274796c1620dbfad5894924750da}
\strng{fullhash}{076c274796c1620dbfad5894924750da}
\strng{bibnamehash}{076c274796c1620dbfad5894924750da}
\strng{authorbibnamehash}{076c274796c1620dbfad5894924750da}
\strng{authornamehash}{076c274796c1620dbfad5894924750da}
\strng{authorfullhash}{076c274796c1620dbfad5894924750da}
\field{sortinit}{3}
\field{sortinithash}{ad6fe7482ffbd7b9f99c9e8b5dccd3d7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Annals of Physics}
\field{number}{2}
\field{title}{Canonical transformations in quantum mechanics}
\field{volume}{232}
\field{year}{1994}
\field{pages}{292\bibrangedash 331}
\range{pages}{40}
\endentry
\entry{abraham1978foundations}{book}{}
\name{author}{8}{}{%
{{hash=ee85bb5038fcaee524ed247a0c225794}{%
family={Abraham},
familyi={A\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
{{hash=2b57da7e5996941c3af7d50b780e992c}{%
family={Ralph},
familyi={R\bibinitperiod},
given={A.},
giveni={A\bibinitperiod}}}%
{{hash=a109e64d7dd72d3d3700a1a0216efd4b}{%
family={Marsden},
familyi={M\bibinitperiod},
given={J.\bibnamedelimi E.},
giveni={J\bibinitperiod\bibinitdelim E\bibinitperiod}}}%
{{hash=3e1fd327b9181c55e076f39c1229bbb1}{%
family={Jerrold},
familyi={J\bibinitperiod},
given={M.E.},
giveni={M\bibinitperiod}}}%
{{hash=e21395f04db22d6a5ba2b12da12c3e1d}{%
family={Marsden},
familyi={M\bibinitperiod},
given={U.J.E.},
giveni={U\bibinitperiod}}}%
{{hash=ffd1e2f9dff6ffd62128428fbeedddc7}{%
family={Ratiu},
familyi={R\bibinitperiod},
given={T.S.},
giveni={T\bibinitperiod}}}%
{{hash=ffd1e2f9dff6ffd62128428fbeedddc7}{%
family={Ratiu},
familyi={R\bibinitperiod},
given={T.S.},
giveni={T\bibinitperiod}}}%
{{hash=bfdf3aef38391502e163af0e2c7d5d33}{%
family={Cushman},
familyi={C\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Basic Books}%
}
\strng{namehash}{0fe65b39f29ac8cb7a37a258d642bdfe}
\strng{fullhash}{6ae820de3d64fbbf67c9db3b7a6e5f9f}
\strng{bibnamehash}{6ae820de3d64fbbf67c9db3b7a6e5f9f}
\strng{authorbibnamehash}{6ae820de3d64fbbf67c9db3b7a6e5f9f}
\strng{authornamehash}{0fe65b39f29ac8cb7a37a258d642bdfe}
\strng{authorfullhash}{6ae820de3d64fbbf67c9db3b7a6e5f9f}
\field{sortinit}{4}
\field{sortinithash}{9381316451d1b9788675a07e972a12a7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Foundations Of Mechanics}
\field{year}{1978}
\endentry
\entry{tung1985group}{book}{}
\name{author}{1}{}{%
{{hash=d3561b550e179b0ff9e4c3be1b9d6f60}{%
family={Tung},
familyi={T\bibinitperiod},
given={W.K.},
giveni={W\bibinitperiod}}}%
}
\list{publisher}{1}{%
{World Scientific Publishing Company}%
}
\strng{namehash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\strng{fullhash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\strng{bibnamehash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\strng{authorbibnamehash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\strng{authornamehash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\strng{authorfullhash}{d3561b550e179b0ff9e4c3be1b9d6f60}
\field{sortinit}{4}
\field{sortinithash}{9381316451d1b9788675a07e972a12a7}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Group theory in physics: an introduction to symmetry principles, group representations, and special functions in classical and quantum physics}
\field{year}{1985}
\endentry
\entry{wigner1932operation}{article}{}
\name{author}{1}{}{%
{{hash=0f65f64e6050b36184863dbf3ec857c1}{%
family={Wigner},
familyi={W\bibinitperiod},
given={E.},
giveni={E\bibinitperiod}}}%
}
\strng{namehash}{0f65f64e6050b36184863dbf3ec857c1}
\strng{fullhash}{0f65f64e6050b36184863dbf3ec857c1}
\strng{bibnamehash}{0f65f64e6050b36184863dbf3ec857c1}
\strng{authorbibnamehash}{0f65f64e6050b36184863dbf3ec857c1}
\strng{authornamehash}{0f65f64e6050b36184863dbf3ec857c1}
\strng{authorfullhash}{0f65f64e6050b36184863dbf3ec857c1}
\field{sortinit}{5}
\field{sortinithash}{20e9b4b0b173788c5dace24730f47d8c}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-Physikalische Klasse}
\field{title}{Über die Operation der Zeitumkehr in der Quantenmechanik}
\field{volume}{1932}
\field{year}{1932}
\field{pages}{546\bibrangedash 559}
\range{pages}{14}
\verb{doi}
\verb https://doi.org/10.1007/978-3-662-02781-3_15
\endverb
\endentry
\entry{von1931uniqueness}{article}{}
\name{author}{1}{}{%
{{hash=7a75c208bc8e03ef11be2cc8099037b4}{%
family={Von\bibnamedelima Neumann},
familyi={V\bibinitperiod\bibinitdelim N\bibinitperiod},
given={J.},
giveni={J\bibinitperiod}}}%
}
\strng{namehash}{7a75c208bc8e03ef11be2cc8099037b4}
\strng{fullhash}{7a75c208bc8e03ef11be2cc8099037b4}
\strng{bibnamehash}{7a75c208bc8e03ef11be2cc8099037b4}
\strng{authorbibnamehash}{7a75c208bc8e03ef11be2cc8099037b4}
\strng{authornamehash}{7a75c208bc8e03ef11be2cc8099037b4}
\strng{authorfullhash}{7a75c208bc8e03ef11be2cc8099037b4}
\field{sortinit}{5}
\field{sortinithash}{20e9b4b0b173788c5dace24730f47d8c}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Math. Ann}
\field{title}{On the uniqueness of the Schrödinger operators}
\field{volume}{104}
\field{year}{1931}
\field{pages}{570\bibrangedash 578}
\range{pages}{9}
\endentry
\entry{gilmore2012lie}{book}{}
\name{author}{1}{}{%
{{hash=e6e7891235a13a4f1e04f8d3ed334879}{%
family={Gilmore},
familyi={G\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Courier Corporation}%
}
\strng{namehash}{e6e7891235a13a4f1e04f8d3ed334879}
\strng{fullhash}{e6e7891235a13a4f1e04f8d3ed334879}
\strng{bibnamehash}{e6e7891235a13a4f1e04f8d3ed334879}
\strng{authorbibnamehash}{e6e7891235a13a4f1e04f8d3ed334879}
\strng{authornamehash}{e6e7891235a13a4f1e04f8d3ed334879}
\strng{authorfullhash}{e6e7891235a13a4f1e04f8d3ed334879}
\field{sortinit}{5}
\field{sortinithash}{20e9b4b0b173788c5dace24730f47d8c}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Lie groups, Lie algebras, and some of their applications}
\field{year}{2012}
\endentry
\entry{milne1948vectorial}{book}{}
\name{author}{1}{}{%
{{hash=79b873ef5025a6f41b791e01aad1f55f}{%
family={Milne},
familyi={M\bibinitperiod},
given={E.\bibnamedelimi A.},
giveni={E\bibinitperiod\bibinitdelim A\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Interscience}%
}
\strng{namehash}{79b873ef5025a6f41b791e01aad1f55f}
\strng{fullhash}{79b873ef5025a6f41b791e01aad1f55f}
\strng{bibnamehash}{79b873ef5025a6f41b791e01aad1f55f}
\strng{authorbibnamehash}{79b873ef5025a6f41b791e01aad1f55f}
\strng{authornamehash}{79b873ef5025a6f41b791e01aad1f55f}
\strng{authorfullhash}{79b873ef5025a6f41b791e01aad1f55f}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Vectorial mechanics}
\field{year}{1948}
\endentry
\entry{conlon2008differentiable}{book}{}
\name{author}{1}{}{%
{{hash=1f4026c8b1b5d479002a31b2b8909e43}{%
family={Conlon},
familyi={C\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Springer Science \& Business Media}%
}
\strng{namehash}{1f4026c8b1b5d479002a31b2b8909e43}
\strng{fullhash}{1f4026c8b1b5d479002a31b2b8909e43}
\strng{bibnamehash}{1f4026c8b1b5d479002a31b2b8909e43}
\strng{authorbibnamehash}{1f4026c8b1b5d479002a31b2b8909e43}
\strng{authornamehash}{1f4026c8b1b5d479002a31b2b8909e43}
\strng{authorfullhash}{1f4026c8b1b5d479002a31b2b8909e43}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{title}{Differentiable manifolds}
\field{year}{2008}
\endentry
\entry{Klages}{article}{}
\name{author}{4}{}{%
{{hash=264c5e154963ca7511adf890344ab64f}{%
family={Colangeli},
familyi={C\bibinitperiod},
given={M.},
giveni={M\bibinitperiod}}}%
{{hash=4991ebf7b63f49dafe10a52f223e362b}{%
family={Klages},
familyi={K\bibinitperiod},
given={R.},
giveni={R\bibinitperiod}}}%
{{hash=12a26c762dcfc045a1dd292265e43030}{%
family={De\bibnamedelima Gregorio},
familyi={D\bibinitperiod\bibinitdelim G\bibinitperiod},
given={P.},
giveni={P\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{IOP Publishing}%
}
\strng{namehash}{3bbcc3cf34f681c38fd508a8f3c05259}
\strng{fullhash}{4fde6df4276bf8bfcde5d247b85f53db}
\strng{bibnamehash}{4fde6df4276bf8bfcde5d247b85f53db}
\strng{authorbibnamehash}{4fde6df4276bf8bfcde5d247b85f53db}
\strng{authornamehash}{3bbcc3cf34f681c38fd508a8f3c05259}
\strng{authorfullhash}{4fde6df4276bf8bfcde5d247b85f53db}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Journal of Statistical Mechanics: Theory and Experiment}
\field{number}{04}
\field{title}{Steady state fluctuation relations and time reversibility for non-smooth chaotic maps}
\field{volume}{2011}
\field{year}{2011}
\field{pages}{P04021}
\range{pages}{-1}
\endentry
\entry{ColR1}{article}{}
\name{author}{2}{}{%
{{hash=264c5e154963ca7511adf890344ab64f}{%
family={Colangeli},
familyi={C\bibinitperiod},
given={M.},
giveni={M\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Elsevier}%
}
\strng{namehash}{327e2e20c7567f65a882a2cec2e8120d}
\strng{fullhash}{327e2e20c7567f65a882a2cec2e8120d}
\strng{bibnamehash}{327e2e20c7567f65a882a2cec2e8120d}
\strng{authorbibnamehash}{327e2e20c7567f65a882a2cec2e8120d}
\strng{authornamehash}{327e2e20c7567f65a882a2cec2e8120d}
\strng{authorfullhash}{327e2e20c7567f65a882a2cec2e8120d}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Physica D: Nonlinear Phenomena}
\field{number}{6}
\field{title}{Equilibrium, fluctuation relations and transport for irreversible deterministic dynamics}
\field{volume}{241}
\field{year}{2012}
\field{pages}{681\bibrangedash 691}
\range{pages}{11}
\endentry
\entry{ColR2}{article}{}
\name{author}{3}{}{%
{{hash=376fd38e009ceff4c20f236cb9d3f1c5}{%
family={Adamo},
familyi={A\bibinitperiod},
given={P.\bibnamedelimi A.},
giveni={P\bibinitperiod\bibinitdelim A\bibinitperiod}}}%
{{hash=264c5e154963ca7511adf890344ab64f}{%
family={Colangeli},
familyi={C\bibinitperiod},
given={M.},
giveni={M\bibinitperiod}}}%
{{hash=d996838298539f621b8fd9d0c6686dd2}{%
family={Rondoni},
familyi={R\bibinitperiod},
given={L.},
giveni={L\bibinitperiod}}}%
}
\list{publisher}{1}{%
{Elsevier}%
}
\strng{namehash}{e9780ff9ea244b3ff54cecab24bdffa2}
\strng{fullhash}{e9780ff9ea244b3ff54cecab24bdffa2}
\strng{bibnamehash}{e9780ff9ea244b3ff54cecab24bdffa2}
\strng{authorbibnamehash}{e9780ff9ea244b3ff54cecab24bdffa2}
\strng{authornamehash}{e9780ff9ea244b3ff54cecab24bdffa2}
\strng{authorfullhash}{e9780ff9ea244b3ff54cecab24bdffa2}
\field{sortinit}{6}
\field{sortinithash}{b33bc299efb3c36abec520a4c896a66d}
\field{labelnamesource}{author}
\field{labeltitlesource}{title}
\field{journaltitle}{Chaos, Solitons \& Fractals}
\field{title}{Role of ergodicity in the transient Fluctuation Relation and a new relation for a dissipative non-chaotic map}
\field{volume}{83}
\field{year}{2016}
\field{pages}{54\bibrangedash 66}
\range{pages}{13}
\endentry
\enddatalist
\endrefsection
\endinput