A binary model of long period radio transients and white dwarf pulsars
Extracted source text · arXiv 2507.15352
\documentclass{nature}
% \documentclass{article}
% GPM J1839-10 values
\newcommand{\gpmlpt}{GPM\,J1839\ensuremath{-}10} % Name
\newcommand{\gpmlptT}{$T = 31482.4 \pm 0.2$ s} % Orbital period T
\newcommand{\gpmlptPa}{$P_A = 1318.1957 \pm 0.0002$ s} % Beat period P_a
\newcommand{\gpmlptPb}{$P_B = 1265.2197 \pm 0.0002$ s} % Spin period P_b
\newcommand{\gpmlpti}{$i = 100.1 \pm 0.6^\circ $} % Orbital inclination i about y axis
\newcommand{\gpmlptalpha}{$\alpha = 52.1 \pm 0.4^\circ $} % Magnetic moment inclination alpha
\newcommand{\gpmlptphiz}{$\phi_0 = 179.29 \pm 0.04^\circ $} % Viewing angle azimuth about z axis, with 0 at x axis
\newcommand{\gpmlptzeta}{$\zeta = 61 \pm 2^\circ $} % Viewing angle zenith w.r.t z axis
\newcommand{\gpmlptWspin}{$W_\text{spin} = 65 \pm 3^\circ $} % Beam opening angle
\newcommand{\gpmlptWorb}{$W_\text{orb} = 70 \pm 1^\circ $} % Modulation width
\newcommand{\jlpt}{J1912\ensuremath{-}44}
\newcommand{\jlptT}{$T = 14525.5588 \pm 0.0001$ s}
\newcommand{\jlptP}{$P = 319.3490 \pm 0.0001$ s}
\newcommand{\jlpti}{$i \in [53, 65]^\circ$}
\newcommand{\jlptalpha}{$\alpha \in [ 42, 70]^\circ$}
\newcommand{\jlptphiz}{$\phi_0 \in [-48, -17]^\circ$}
\newcommand{\jlptzeta}{$\zeta \in [ 39, 73]^\circ$}
\newcommand{\jlptWspin}{$W_\text{spin} \in [20, 60]^\circ$}
\newcommand{\jlptWorb}{$W_\text{orb} \in [67, 90]^\circ$}
%% make sure you have the nature.cls and naturemag.bst files where
%% LaTeX can find them
\usepackage{aas_macros}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{hyperref}
\usepackage{lineno}
\usepackage{float}
% \usepackage{cleveref}
% \usepackage[numbers]{natbib}
% \usepackage{nameref}
\usepackage{booktabs}
\usepackage{enumitem}
% Biblatex seems to be incompatible with the nature.cls
%\usepackage[backend=biber,style=nature]{biblatex}
%\addbibresource{refs.bib}
\hypersetup{colorlinks,citecolor=blue,linkcolor=blue,urlcolor=blue}
%% Template for a preprint Letter or Article for submission
%% to the journal Nature.
%% Written by Peter Czoschke, 26 February 2004
%%
% To fix a compilation error
\DeclareUnicodeCharacter{2212}{-}
\bibliographystyle{naturemag}
%author color commenting
\def \nr#1{{\color{cyan}NR: #1}}
\def \csh#1{{\color{magenta}CsH: #1}}
\def \smc#1{{\color{green}SMc: #1}}
\title{A unified model for long-period radio transients and white dwarf binary pulsars}
%A unified model for long-period radio transients and M-dwarf/WD binaries
%% Notice placement of commas and superscripts and use of &
%% in the author list
\author{
Csan\'{a}d Horv\'{a}th$^{1}$,
Nanda Rea$^{2,3}$,
Natasha Hurley-Walker$^{1}$,
Samuel J. McSweeney$^{1}$,
Richard A. Perley$^{4}$,
Emil Lenc$^{5}$
}
%Articles
%%% This block stops it ignoring figures, as coded in the bst file
\usepackage{graphicx}
\makeatletter
\let\saved@includegraphics\includegraphics
\AtBeginDocument{\let\includegraphics\saved@includegraphics}
\renewenvironment{figure}{\@float{figure}}{\end@float}
\makeatother
%%% End figure block
\begin{document}
% Other
\newcommand{\lum}{erg\,s\ensuremath{^{-1}}}
\newcommand{\flux}{erg\,cm\ensuremath{^{-2}}\,s\ensuremath{^{-1}}}
\def\arcsec{\mbox{$^{\prime\prime}$}}
\def\nh{\hbox{$N_{\rm H}$}}
\maketitle
\begin{affiliations}
\item International Centre for Radio Astronomy Research, Curtin University, Kent St, Bentley WA 6102, Australia
\item Institute of Space Sciences (ICE, CSIC), Campus UAB, Carrer de Can Magrans s/n, 08193, Barcelona, Spain
\item Institut d’Estudis Espacials de Catalunya (IEEC), Esteve Terradas 1, RDIT Building, 08860, Castelldefels, Spain
\item National Radio Astronomy Observatory, Socorro, NM 87801, USA
\item Australia Telescope National Facility, CSIRO, Space \& Astronomy, Epping, New South Wales, Australia
\end{affiliations}
\begin{abstract}
Long-period radio transients (LPTs) represent a recently uncovered class of Galactic radio sources exhibiting minute-to-hour periodicities and highly polarised pulses of second-to-minute duration. Their phenomenology does not fit exactly in any other class, although it might resemble that of radio magnetars or of white dwarf (WD) radio emitting binary systems. Notably, two LPTs with confirmed multi-wavelength counterparts have been identified as synchronised white dwarf–M dwarf binaries (polars). Meanwhile, systems such as AR\,Scorpii and \jlpt{} exhibit short-period pulsations in a hr-tight orbit, with polarized radio emission proposed to be generated by the interaction of the WD magnetosphere with the low-mass companion wind.
Here we demonstrate that both LPTs and WD binary pulsars can be explained within a single geometric model in which radio emission is triggered when the magnetic pole of a rotating white dwarf intersects its companion’s wind in the binary orbital plane. We use a 36-year timing baseline to infer the orbital period and binary geometry solely from radio data of \gpmlpt{}, the longest-active LPT known. The model naturally predicts its intermittent emission and double-pulse structure. Crucially, we show that the beat period between the spin and the orbit matches the observed pulse substructure and polarisation signatures, providing strong support for the model. Applying this model to the WD binary pulsar \jlpt\, it can also reproduce the system emission and geometry. Our results place \gpmlpt{}, and other LPTs in general and radio emitting WD binaries, at different stages of a continuum between intermediate and synchronised polars, suggesting a unified population of magnetic WD binaries driving coherent radio emission.
%as the utlimate nature of most LPTs.
%Recently several long-period radio transients have been discovered, with strongly polarised coherent radio pulses appearing on timescales between tens to thousands of seconds \cite{2022NatAs...6..828C,2022Natur.601..526H}.....
\end{abstract}
%\nr{The "story" is more or less there... It misses some more quantitative info on the fitted model and the results (Csanad?), and more quantitative derivation for the radio emission model: i.e. how much Mdot wind do we need from the M-star to have such radio emission? This info might be found in the equations of the recent Christo Venter model for AR Sco (for Nanda to look into this still) \cite{2024MNRAS.532.4408D}. Link the figures to the main text, and check references too.}
%\section*{Main}
%\section*{Introductory comparison between WD radio pulsars and LPTs}
White dwarf binary systems are increasingly recognised as hosts of exotic pulsar-like emission, challenging long-standing assumptions that coherent radio pulses are exclusive to neutron star systems\cite{2017NatAs...1E..29B, pelisoli2023}. The discovery of long-period transients (LPTs), characterised by long periodicities (minutes to hours)
%compared to the bulk of the pulsar population
, and strong polarisation, further deepens the mystery of how and where coherent radio emission originates \cite{2022Natur.601..526H, Hurleywalker2023}. These intriguing systems are characterized by periodic radio pulses with high polarization (both linear and circular) and a high surface brightness pointing to a coherent radio emission process. However, the luminosities of the bright radio pulses in the $\sim$10 systems discovered thus far, exceed their rotational power, leaving uncertain the energy source. A strong magnetic field has been suggested to overcome this energy problem, suggesting at first the possibility of them being radio magnetars having perhaps had large fall-back at birth slowing down their periods \cite{2022ApJ...934..184R, genccali2022evolution}.
Two LPTs, ILT\,J1101$+$5521 \cite{de2025sporadic} and GLEAM$-$X\,J0704$-$37 \cite{hurley20242, rodrigues2025}, have recently been associated with a WD--M dwarf binary system via the detection of the orbital motion in the optical spectrum of the M companion star, at a period coincident with the radio pulse period. These systems might be hosting a WD with a spin synchronized with the orbital motion, hence belonging to the class of polars, despite not being in an accretion phase. In contrast, the AR Scorpii (Ar\,Sco) and J1912--44 radio emitting binary WDs are not synchronised with their companions; coherent radio pulses are observed at beat periods resulting from the interaction between the white dwarf spin and the orbital motion. The radio emission in these systems has been intriguing all along, and scenarios involving the dipolar losses of the white dwarf rotating field and the interaction with the M--dwarf companion wind have been proposed \cite{potter_and_buckley_2018,qu2025magnetic, geng2016model}
%\csh{Were you thinking of other references here, Nanda?}. NR: Perfect!
The phenomenological similarities between these sources invite a unified physical framework. We propose a geometric emission model in which radiation is generated when a WD's magnetic pole sweeps across its companion star, injecting relativistic particles from the M dwarf's wind into the WD's magnetosphere \cite{2021NatAs...5..648S, 2024MNRAS.527.3826P}. These particles are accelerated along magnetic field lines, producing a collimated radio beam aligned with the magnetic moment of the WD. The model incorporates orbital inclination, magnetic obliquity, and beam geometry, and predicts observable intensity variations as a function of the spin and orbital phase. Further intensity variations observed in the long-term radio outbursts of LPTs might be ascribed to the intrinsic variability of the M-dwarf wind.
Our model assumes a circular orbit with inclination angle $i$ between the orbital axis and the WD spin axis. The magnetic moment is tilted by angle $\alpha$ from the spin axis, and maximum radio emission is generated in the direction of the magnetic moment. Emission is modulated in two ways: (1) by the angle $\beta_{\text{MD}}$ between the magnetic pole and the M dwarf, governing the availability of particles via an assumed Gaussian function; and (2) by the angle $\beta$ between the magnetic pole and the Earth, determining visibility to observers.
This model, although simplified, provides predictive power: emission is expected when both $\beta_{\text{MD}}$ and $\beta$ are small. The resulting phase-space of emission naturally leads to double-peaked pulse groups and intermittent visibility consistent with observations.
GPM\,J1839$-$10 is the longest lived LPT known, exhibiting a 22-minute period with sporadic pulses observed across 36 years of archival radio data \cite{Hurleywalker2023}. In a coordinated campaign using MeerKAT, ASKAP, and VLA in 2024, we identified a higher-order periodicity: pulses cluster in double groups occurring every $\sim$8.75\,hr. We interpret this as the orbital period \gpmlptT{}, which we are able to measure to such low uncertainty by taking advantage of the 36 year baseline.
Folding the data over this 8.75\,hr modulation period and the previously reported 22-minute period ($P_A$)\cite{Hurleywalker2023}, as well as the derived beat period ($P_B$), reveals in \autoref{fig:period_comparison} that the pulse substructure aligns between pulses with \gpmlptPb{}, rather than \gpmlptPa{}.
The substructure timescale of $\sim$50 s is consistent throughout the archival data, first discovered by autocorrelation between pulses \cite{Hurleywalker2023}, and folding on $P_B$ reveals further correlation and vertical alignment of the brightest peaks. This suggests that $P_B$ corresponds to the true spin period of the white dwarf (while $P_A$ is a spin-orbit beat period) inviting its interpretation as the cross-section of the radio beam.
The phase difference between the left and right pulse groups cannot be attributed to the difference in light travel time between orbital phases because that would be on the order of seconds for an orbit of this size, not minutes. Instead, we interpret the groups as arising from the intersections of the M dwarf's highly inclined orbit with the cone traced out by the WD's magnetic pole. This explains both the intermittency and the phase delay between pulse groups. Our geometric model reproduces the observed dynamic pulse profile as shown in \autoref{fig:period_b_predicted_flux} when fitted using a Markov chain Monte Carlo method, with best-fit parameters indicating a near-perpendicular orbital inclination \gpmlpti{}. The modelled pulse groups are the regions where the $\beta_\text{MD}$ contours cut into the central column of the $\beta$ colour map --- where the Earth, MD, and WD pole are sufficiently aligned. No good fit exists for $P_A$ as the spin period.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{mega_plot_4.pdf}
\caption{Dynamic pulse profile and polarisation position angles of \gpmlpt{}. Profiles are folded vertically on the orbital period \gpmlptT{} and horizontally on \gpmlptPa{} (left) and \gpmlptPb{} (right). Above, the mean flux is calculated over two boxes enclosing the pulse groups: $\lbrace-0.18 < \text{orbital phase} < 0.18\rbrace$ for period A, $(\lbrace-0.18 < \text{orbital phase} < -0.06\rbrace \cap \lbrace\text{spin phase} < 0\rbrace) \cup (\lbrace0.06 < \text{orbital phase} < 0.18\rbrace \cap \lbrace\text{spin phase} > 0\rbrace)$ for period B. \label{fig:period_comparison}}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{mcmc_best_fit_prediction.pdf}
\caption{Modelled dynamic pulse profile of \gpmlpt{}. At top left, the flux density predicted by the model $I_\text{pred}$ using the best-fit parameters found using MCMC is overlaid on the real pulse profiles. At top right, the colourmap is the LOS-beam angle $\beta$ as a function of spin-orbit phase, and the contours are the beam-MD angle $\beta_\text{MD}$. $I_\text{pred}$ is greatest where $\beta$ and $\beta_\text{MD}$ are the smallest. The bottom panel is a full orbit recorded by ASKAP and the associated model prediction. The vertical lines are spaced by the spin period. \label{fig:period_b_predicted_flux}}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The emission is consistently highly linearly polarised with occasional 90$^\circ$ jumps in polarisation angle (PA) to orthogonal polarisation modes (OPMs)\cite{Hurleywalker2023}. However, \autoref{fig:period_comparison} reveals that the two pulse groups which appear at different orbital and spin phases are dominated by different OPMs, with a similar amount of variation in each group ($\sigma_\text{PA} \approx {\sim}10^\circ$). The PAs of individual pulses, however, may evolve more gradually and systematically across spin phase \cite{doi:10.1126/sciadv.adp6351} than can be seen in the point clusters in the figure.
It can further be seen that there is a small downturn in the PAs at around spin phase $-0.05$ (the trailing edge of the first group), as well as a similar upturn in the PAs at spin phase $0.05$ (the leading edge of the second group).
Despite the possibility that pulses arriving between the two groups could potentially connect the PAs smoothly across intermediate spin/orbital phases, the flatness of the PAs of the two groups themselves, as well as the occasional $90^\circ$ jump within each group, lead us to prefer an OPM interpretation over a rotating vector model one\cite{1969ApL.....3..225R}.
Although not well understood, OPMs in pulsars are thought to be magnetospheric in origin, possibly due to birefringent properties of the co-rotating plasma\cite{1997A&A...327..155G}. We advance a similarly broad interpretation here: the OPMs are dominated by the average magnetic properties of the plasma through which the WD beam propagates along our line of sight.
The consistency with which each pulse group exhibits the same dominant OPM support our geometric model, with the line of sight intersecting different physical sections of the intrabinary magnetic field during the two pulse groups.
Both the single pulse PA variations and the occasional OPM jump even within a pulse group may be due to turbulence in the intrabinary wind.
Other spectral characteristics (dispersion measure, rotation measure, spectral shape) remain constant across the orbit, suggesting that variability arises primarily from geometric effects rather than intrinsic source changes on a short timescale, or propagation effects.
On longer timescales (a few orbits) variability is primarily driven by MD wind activity.
% \csh{Or M-dwarf activity, right? Variation in the wind can explain variation in the brightness}
% \nr{yes, I would say so...}
% \csh{I also wonder if it's worth commenting on the apparent periodicity of the substructure?}
% \nr{I would not in the main text, since there is little we can saa about their interpretations, and it is certainly not geometry but rather connected to propagation or plasma physics I would say... what others say?}
% \smc{Agree with NR. The original discovery paper, however, did report autocorrelations of lightcurves for various individual pulses, and found variations on similar timescales, so just a sentence to link this substructure to the phenomenon reported in the discovery paper would be good.}
% \csh{I shoehorned the autocorrelations into the "Folding the data over this 8.75..." paragraph.}
%\nr{- Here we need to mention the exact derived properties of the GPM1839 system, such as beam angle, and all what can be derived exactly in the model with its error. (Csanad)}
To describe the system geometry, we place the WD at the origin with $\hat{z}$ as the WD spin axis, and the orbital axis inclined about $\hat{y}$. Fitting the model for \gpmlpt{}, we find an orbital inclination \gpmlpti{} and magnetic obliquity \gpmlptalpha{} with respect to $\hat{z}$. Our line of sight has azimuth \gpmlptphiz{} measured from $\hat{x}$ and zenith \gpmlptzeta{} measured from $\hat{z}$. The radio beam has an opening angle of \gpmlptWspin{} and becomes active when the MD is within \gpmlptWorb{} of the WD magnetic moment $\hat{\mu}$.
In \autoref{fig:scale_diagram} we show the derived beam angle and orbit of the system together with critical radii to understand the emission state of the binary. The derived orbital period and WD spin period imply that: 1) the whole system is inside the WD light cylinder, 2) the companion star, if assumed to be an M-dwarf (as for the other two LPTs), is not filling its Roche-Lobe, and 3) depending on the assumed M-dwarf mass loss rate ($\sim10^{-14}$ M$_\odot$ yr$^{-1}$ \cite{2005ApJ...628L.143W}), the WD Alfv\'en radius might be inside the corotation radius, implying that during stellar flares the system may undergo accretion episodes (possibly showing transient X-ray emission as observed in the LPT ASKAP\,J1832-0911 \cite{2024arXiv241116606W}), while being in propeller most of the time \cite{Lyutikov2020, 2024MNRAS.527.3826P}. Furthermore, the system geometry clearly shows how the rotating WD magnetosphere is naturally engulfed by the MD wind, possibly being the seed particles that are accelerated by the sweeping magnetic field lines generating the pulsed synchrotron radio emission \cite{Geng2016, Katz2017, takata2017model}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{scale_diagram_end_me.png}
\caption{Diagram of the binary system, in a moving reference frame centred on the WD.
At left is a not-to-scale diagram of the geometric parameters. In green is WD with its magnetic moment vector. In red, the MD with its orbital path. The blue vector points towards Earth.
At right is a to-scale projection of \gpmlpt{} on the y-z plane, with orbital phases marked. The cone traced by $\hat{\mu}$ crosses the y-z plane at the green full and dashed lines.
The shaded red, orange, and cyan regions cover the ranges of $R_\text{MD}$, $R_{rl}$, and $R_c$ respectively, given the MD mass is $M_\text{MD} \in [0.14, 0.5] M_\odot$ and the WD mass is $M_\text{WD} \in [0.6, 1.2] M_\odot$. The green sector and solid red line are $W_\text{spin}$ and $W_\text{orb}$ respectively. The small and large black dashed circles are the Alfvén radii for the minimum and maximum WD masses respectively, assuming a stellar mass loss rate of $\dot{M}_\text{MD}$ = $10^{-14}$ M$_\odot$ yr$^{-1}$ \cite{2005ApJ...628L.143W}. \label{fig:scale_diagram}}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\nr{- discuss here the same model applied to J1912 and what you derive from there. (Csanad)}
To test the generality of the model, we applied it to 7.5 hours of MeerKAT observations of the WD binary pulsar \jlpt{}. Folding the data over the 5.3 minute spin and 4.03 hour orbital periods \cite{pelisoli2023} forms the vertically aligned dynamic pulse profile in \autoref{fig:j1912}. We used a prior of $i = 59 \pm 6^\circ$ based on the inclination angle constraints as measured in the optical band \cite{2024MNRAS.527.3826P}. Given the prior, we find the following constraints on the geometry: \jlpti{}, \jlptalpha{}, \jlptphiz{}, \jlptzeta{}, \jlptWspin{}, and \jlptWorb{}. These results agree broadly with the earlier proposed geometry \cite{2024MNRAS.527.3826P}, but with only one pulse group the geometrical parameters are less constrained than for \gpmlpt{}.
A critical component of the model is that the direction of the radio beam is locked to the WD rotating frame in the direction of the magnetic moment.
If this is not the case it becomes difficult to explain the paired pulse group morphology, and it becomes unclear why the pulse substructure would be related to the spin phase. This is difficult to reconcile with models which place the emission site at or near the companion like the electron cyclotron maser emission (ECME) model \cite{2025ApJ...981...34Q} proposed for ILT\,J1101\ensuremath{+}5521 or the bow shock model \cite{geng2016model} proposed for AR\,Sco.
%\nr{Comment: are you sure? }\csh{pretty sure...}
One scenario which allows for an emission region locked to the WD involves relativistic electrons accelerating from the companion towards the WD along the closed magnetic field lines when the companion star's surface is heated by magnetic interaction \cite{buckley2017polarimetric, takata2017model}. The electrons emit synchrotron radiation as they reach magnetic mirror points on their way to the WD pole \cite{takata2017model}. This idea has been shown to reproduce the dynamic pulse profile of Ar\,Sco with spin and orbit modulated components similar to our model \cite{potter_and_buckley_2018}.
In any case, the emission mechanism must differ substantially from the common radio pulsar curvature radiation process, a testable prediction of which is the beam opening angle confined by the last open field lines at the pole. Assuming that the emission site is within the orbital radius ($r_{em} \leq A \leq 2.55 R_\odot$) places a generous limit on the half-opening angle of $\rho \leq 15^\circ$, inconsistent with the best-fit half-opening angle $W_\text{spin} = 65 \pm 3 ^\circ$.
%This suggests that the closed field lines play a significant role regardless of the exact mechanism.
Further separating \gpmlpt{} from neutron stars is the lack of PA sweep typically associated with the rotating vector model, although this is not a unifying feature of WD pulsars \cite{du2019constraining}.
%\csh{Something about the Alfven and corotation radii and evolution \cite{2021NatAs...5..648S} (Nanda?)}
%\nr{Yes. I am trying to figure out the evolution part. I am starting to get convinced that they are not different evolutionary states but rather different systems with different geometries and will not evolve one into the other... let's discuss this on Monday in zoom before I write something wrong here :)}
% \nr{For this last sentence maybe if it is not too much complicated and too much work we might want to try to model the data of the "interpulse"-LPT. If we show that this can also be geometrical I think the result is even stronger. What do you think. Is it feasible?}
% \csh{- I have done some modelling of Ar\,Sco, but the results aren't as impressive as the others. Ar\,Sco is just so messy. See screenshot at end of doc. Should I include this? }
% \nr{Mmmm... maybe it is not crucial at that point. We have 1912, it might be enough for our aims...then if asked by the referee it can be added with all the due caveats... no?}
%\nr{- Here comment on why one some are bright others are faint? why some have non-coherent luminosities and others coherent? Show that given our model and others simulations on their radio emission, this is expected (Nanda).}
Future work will include applying the model to the growing number of LPTs, which will not only test it, but serve to constrain the distribution of their geometries. Of particular interest is the 6.45 h LPT ASKAP\,J183950.5\ensuremath{-}075635.0 \cite{2025NatAs...9..393L} which has timing variation and pulse profile evolution phenomenologically similar to \gpmlpt{} but with an interpulse, extending the geometric parameter space. Also of interest would be J1755\ensuremath{-}2527, recently confirmed to be a 1.16 hour intermittent LPT \cite{sam2025}, the emission-state-switching 54 minute LPT ASKAP\,J1935+2148 \cite{caleb2024emission} and CHIME\,J0630\ensuremath{+}25 \cite{dong2024discoverynearby421stransient}, whose 421 s period makes it an interesting counterpart to \jlpt{}.
Our results demonstrate that LPTs like \gpmlpt{} can be understood within the same emission geometry framework as radio emitting WD binary pulsars such as \jlpt{} and AR Sco.
The diversity in the observed periods, pulse morphologies, and polarisation states arises from variations in orbital synchronisation and viewing geometry.
This implies that coherent radio emission from magnetic WD binaries is more common and varied than previously thought, potentially forming a new class of compact binary radio emitters.
% \csh{This places \gpmlpt{} at the long-period end of the variety of evolutionary stages of WD binaries, bridging synchronised and asynchronous, accreting and non-accreting, and radio bright and quiet systems. - I don't think this sentence is necessary.}
%\nr{%The common scenario between LPTs and the two radio emitting white dwarf binaries has now been demonstrated by this model.
However, the possible evolutionary connection between these systems remains debated, and we cannot exclude them being at different evolutionary stages but with common properties and geometries allowing them to emit bright polarized radio emission. With enough samples, we expect their evolutionary stages to emerge, and the tools developed in this work will enable insight into binary WD and magnetic field evolution in a way difficult to select from optical surveys alone.
%}\csh{Edited. First line I think was a repeat of "our results demonstrate that...", so I rearranged the ending.}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\include{supplementary}
% Use this up until the time for final submission, then do the final compilation and download the bbl file, and comment out this line, put bibitems below
\bibliography{refs.bib}
%\begin{thebibliography}{10}
%\expandafter\ifx\csname url\endcsname\relax
% \def\url#1{\texttt{#1}}\fi
%\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
%\providecommand{\bibinfo}[2]{#2}
%\providecommand{\eprint}[2][]{\url{#2}}
%\bibitem{2022NatAs...6..828C}
%\bibinfo{author}{{Caleb}, M.} \emph{et~al.}
%\newblock \bibinfo{title}{{Discovery of a radio-emitting neutron star with an
% ultra-long spin period of 76 s}}.
%\newblock \emph{\bibinfo{journal}{Nature Astronomy}}
% \textbf{\bibinfo{volume}{6}}, \bibinfo{pages}{828--836}
% (\bibinfo{year}{2022}).
%\end{thebibliography}
\section*{Methods}
\label{sec:methods}
\subsection{Data}
To obtain a contiguous 36-hour track of \gpmlpt{}, we employed three telescopes: the Australian Square Kilometre Array Pathfinder (ASKAP; \cite{2021PASA...38....9H}), the MeerKAT telescope \cite{2016mks..confE...1J} in South Africa, and the Karl G. Jansky Very Large Array (VLA \cite{2011ApJ...739L...1P}) in the United States of America. To maximise S/N for this steep-spectrum source, we used the lowest-frequency bands available for each telescope. Details are shown in \autoref{tab:obs}.
\begin{table} %%%%%%% Csanad: The table fits :-)
\footnotesize
\centering
\begin{tabular}{rrccccc}
\hline
Start & End & Telescope & Frequency (MHz) & Observation ID & Proposal/Project ID \\
\hline
2024-08-02 \vspace{-10pt} \\
09:30 & 18:30 & ASKAP & 800--1087 & SB64328 & AS113 \\
15:25 & 19:54 & MeerKAT & 544--1088 & 1722610713 & SCI-20230907-NH-01 \\
& 2024-08-03 \vspace{-10pt} \\
21:24 & 01:31 & MeerKAT & 544--1088 & 1722632983 & SCI-20230907-NH-01 \\
2024-08-03 \vspace{-10pt} \\
01:10 & 08:55 & VLA & 1000--2000 & 46270340 & 24A-493 \\
09:00 & 17:00 & ASKAP & 800--1087 & SB64345 & AS113 \\
15:36 & 21:14 & MeerKAT & 544--1088 & 1722697769 & SCI-20230907-NH-01 \\
& 2024-08-04 \vspace{-10pt} \\
22:12 & 01:33 & MeerKAT & 544--1088 & 1722722073 & SCI-20230907-NH-01 \\
\hline
\end{tabular}
\caption{Observing details for the campaign tracking \gpmlpt{}, in chronological order. Dates and times are on a UTC scale. ``Start'' refers to the first sample on target and ``end'' refers to the last (i.e. calibration scans are omitted).\label{tab:obs}}
\end{table}
For ASKAP and MeerKAT we used the standard observatory flagging and calibration.
The VLA data were flagged for obvious RFI and calibrated in AIPS.
%using well-established methods to account for instrumental antenna gains and atmospheric phase irregularities.
The flux density scale was based on the values determined for 3C286 (J1331+3030) \cite{perley2017accurate}. The EVPA values are based on the modified values for 3C286, as reported in EVLA Memo 219 \cite{perley2022evla}.
We imaged all data using \textsc{WSClean}\cite{2014MNRAS.444..606O}, masking \gpmlpt{} itself, forming a deep model of the sky for each observation. After subtracting this from the visibilities, we phase-rotated to \gpmlpt{}. Using \textsc{DSTools}\cite{2024zndo..13626183P} we averaged the baseline data, using the standard conventions to convert from instrumental to celestial Stokes, producing a dynamic spectrum for each observation.
% This is in another section
%We applied a parallactic angle correction to the VLA and MeerKAT data (the design of the rotating ASKAP feeds obviates this step).
% \subsubsection{J1912-44} \label{subsubsec:j1912}
\jlpt{} was observed at L-band by MeerKAT under project code DDT-20220620-PW-01 on 2022-06-26, and originally published by \cite{pelisoli2023}. It was tracked for 16 30-min scans at 208.984-kHz/2-s correlator resolution. We used the provided observatory calibration and re-imaged the data using \textsc{WSClean}, masking \jlpt{} to form a deep model of the sky. After subtracting this model, we averaged the baselines to form dynamic spectra of the pulses, then across the frequency axis to produce light curves. (Dispersion is negligible across this band.)
\subsection{Light curve generation}
The flux densities $S_\nu$ of each frequency channel of the dynamic spectra were scaled to the 1 GHz flux density $S_{1\text{GHz}}$ predicted by the spectral fit previously published \cite{Hurleywalker2023}. The spectral model is of the form
\begin{equation}\label{eqn:spectrum}
S_\nu = S_{1\text{GHz}} \left( \frac{\nu}{1\text{GHz}} \right)^\alpha \exp q \left( \log \frac{\nu}{1\text{GHz}} \right)^2
\end{equation}
with $\alpha = -3.17\pm0.06$ and $q = -0.56\pm0.03$.
The shape of the spectrum was not found to vary within uncertainty as either a function of orbital or spin phase. The dynamic spectra were then de-dispersed by $\text{DM} = 273.5\pm2.5$ pc cm$^{-3}$ and averaged over frequency to produce the light curves. The stokes $Q$ and $U$ components were corrected for rotation measure by $\text{RM} = 531.83\pm0.14$ rad m$^2$, which was also not found to vary.
Barycentric correction was applied to the time axis using the DE430 ephemeris for the Solar System barycentre, and the MeerKAT PA was corrected for the parallactic angle.
% and the PAs of MeerKAT, ASKAP, and VLA were shifted by -62.5\textdegree{}, -73.4\textdegree{}, and 33\textdegree{} respectively to align them. We are confident that the PA shifts are due to instrumental differences and do not represent real changes in the PA because the observation times with the different telescopes were interlaced with each other, yet the PA shifts were consistent.
\subsection{Timing analysis}\label{subsec:timing_analysis}
The \gpmlpt{} spin period \gpmlptPa{} was calculated from the previously reported \gpmlptPb{} as the spin-orbit beat period using
\begin{equation}\label{eqn:beat}
P_B = \frac{T}{T/P_A + 1}.
\end{equation}
To measure the orbital period, the sparseness of the data and the nature of the periodicity makes standard techniques like Fourier analysis and phase dispersion measure unreliable.
\autoref{fig:abacus} shows all pulses detected of \gpmlpt{} folded on the orbital period, including those found in historical data, with the oldest detection in 1988. See \cite{Hurleywalker2023} for details of the historical detections. The y axis is the number of orbits since the most recent detection on UTC 09/11/2024 06:17:30, and the x axis is the residual, with orbital phase 0 chosen to be between the two pulse groups. The orbital envelope width is measured from the leftmost pulse edge to the rightmost pulse edge. If we assume that the historical pulses arrived in the same range of orbital phase as the recent pulses, then for a convincing timing solution all pulses should lie in a central $\sim$12000 s column on \autoref{fig:abacus}.
Pulses were determined to be the regions of the light curves for which flux $>$ 10 mJy and flux $>$ 6 $\times$ RMS. The RMS of lightcurves was calculated only for the regions of the lightcurves for which flux $<$ 10 mJy to avoid including the pulses in the noise measurement.
Trial periods were searched in steps of 0.01 s, and for each period the envelope width is shown in \autoref{fig:abacus}. Envelope widths greater than half of the period are meaningless since the measurement depends on the choice of phase centre. There is only one minimum envelope width less than half the period at \gpmlptT{}. the period uncertainty is the width of the plateau at the minimum in the zoomed cutout. The uncertainty bars are the residual uncertainties given the uncertainty in period. All trial periods outside of the uncertainty range result in some of the historical pulses outside of the envelope of the most recent observations.
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{timing_combined_2.pdf}
\caption{Left: Envelope width for trial orbital periods. Right: Detected pulses folded on the orbital period. The y axis is the orbital period number since the most recent detection (negative means backward in time). The widths of the blue rectangles are the pulse widths. The blue lines are the uncertainties in residual given the uncertainty in period. The envelope width is measured from the leftmost pulse edge to the rightmost pulse edge.\label{fig:abacus}}
\end{figure}
\subsection{Geometric model}
% \autoref{fig:geometric_diagram} is a diagram of the binary system model.
The unit vector of the WD magnetic moment in Cartesian coordinates is
\begin{equation}\label{eqn:beam_posvec}
\hat{\mu} = (\cos\phi\sin\alpha, ~\sin\phi\sin\alpha, ~\cos\alpha)
\end{equation}
The unit vector from the origin to the MD is
\begin{equation}\label{eqn:md_posvec}
\hat{r}_\text{MD} = (\cos{i}\cos\phi_\text{orb}, ~\sin\phi_\text{orb}, ~-\sin{i}\cos\phi_\text{orb})
\end{equation}
The unit vector from the origin to the Earth is
\begin{equation}\label{eqn:earth_posvec}
\hat{r}_\text{E} = (\sin\zeta\cos\phi_0, ~\sin\zeta\sin\phi_0, ~\cos\zeta)
\end{equation}
The angle $\beta_\text{MD}$ between the MD and the WD beam is then
\begin{equation}\begin{aligned}\label{eqn:bead_ang}
\hat{\mu}\cdot\hat{r}_\text{MD} &= \cos\beta_\text{MD} \\
\beta_\text{MD}(\phi_\text{orb},\phi) &= \arccos\left(\cos\phi\sin\alpha\cos{i}\cos\phi_\text{orb} + \sin\phi\sin\alpha\sin\phi_\text{orb} - \cos\alpha\sin{i}\cos\phi_\text{orb}\right)
\end{aligned}\end{equation}
and the angle $\beta$ between the Earth and the WD beam is
\begin{equation}\begin{aligned}\label{eqn:earth_beam_ang}
\hat{r}_\text{E}\cdot\hat{\mu} &= \cos\beta \\
\beta(\phi) &= \arccos\left(\sin\zeta\sin\alpha\cos(\phi_0-\phi) + \cos\zeta\cos\alpha\right)
\end{aligned}\end{equation}
We assume that the radio beam has a Gaussian cross-section. This shape of course cannot reproduce the intricate structure, but given the variability a simple shape was decided to be the best approximation. We set the Gaussian's standard deviation to be 1/5$^\text{th}$ of the full beam width $W_\text{spin}$. Then, the fractional intensity observed on Earth is
\begin{equation}\label{eqn:beam}
f_\text{spin}(\beta) = \exp\frac{-\beta^2}{2(W_\text{spin}/5)^2}
\end{equation}
where $W_\text{spin}$ is the opening angle in radians.
We also assume that the interaction with the MD modulates the intensity with a Gaussian profile with a maximum at $\beta_\text{MD} = 0$ when the beam is pointed towards the MD. The orbital modulation function is
\begin{equation}\label{eqn:modulation}
f_\text{orb}(\beta_\text{MD}) = \exp\frac{-\beta_\text{MD}^2}{2(W_\text{orb}/5)^2}
\end{equation}
The predicted intensity $I_\text{pred}$ observed on Earth is the product $f_\text{spin}(\beta) f_\text{orb}(\beta_\text{MD})$, but to account for either pole producing emission we add the same product offset by $\pi$ radians. Thus,
\begin{equation}\label{eqn:flux}
I_\text{pred}(\beta,\beta_\text{MD}) = C [f_\text{spin}(\beta) f_\text{orb}(\beta_\text{MD}) + f_\text{spin}(\beta-\pi) f_\text{orb}(\beta_\text{MD}-\pi)]
\end{equation}
where $C$ is a coefficient with units of Jy.
\subsection{Model fitting}
We used the ensemble Markov chain Monte Carlo (MCMC) sampler \verb|emcee| \cite{2013PASP..125..306F}. We define the likelihood $P_\text{fit}$ that the model fits the data as
\begin{equation}{\label{eqn:pfit}}
\ln P_\text{fit} = -\frac{1}{2} \sum_{k=1}^N \left[ \left(\frac{I_\text{obs}^{(k)} - I_\text{pred}^{(k)}}{\sigma_k}\right)^2 + \ln(2\pi\sigma_k^2) \right]
\end{equation}
where $N$ is the number of data points, $I_\text{obs}^{(k)}$ and $I_\text{pred}^{(k)}$ are the observed and predicted intensities for the $k^\text{th}$ data point respectively, and $\sigma_k$ is the $k^\text{th}$ uncertainty.
The angles $\phi_\text{orb}^{(k)}$ and $\phi^{(k)}$ were calculated for each data point using the timing residuals $\tau$ for the spin and orbital periods
\begin{equation}\label{eqn:calc_res}
\tau(t,p) = t - t_0 - \left\lfloor\frac{t-t_0}{p}+\frac{1}{2}\right\rfloor p
\end{equation}
where $p$ is the respective period and $t_0$ is the reference time. Then,
\begin{equation}\label{eqn:calc_alpha}
\phi_\text{orb}^{(k)} = \frac{2\pi}{T}\tau(t_k,T) + \phi_\text{orb}^{(0)}
\end{equation}
\begin{equation}\label{eqn:calc_beta}
\phi^{(k)} = \frac{2\pi}{P}\tau(t_k,P) + \phi^{(0)}
\end{equation}
where $\phi_\text{orb}^{(0)}$ and $\phi^{(0)}$ are the $\phi_\text{orb}$ and $\phi$ angles at the reference time $t_0$.
The beam and modulation widths $W_\text{spin}$ and $W_\text{orb}$ were not fit using MCMC because it tended to find unreasonable local $P_\text{fit}$ maxima (such as $W_\text{spin} = W_\text{orb} = 0$) because the real pulse profiles are so non-Gaussian. Rather than an explicit description of the beam profile, the Gaussian functions may be thought of as a weighting for how well the predicted pulse region on the spin-orbit dynamic pulse profile agrees with the region where pulses were observed. Instead of fitting the widths, we calculate the minimum $W_\text{spin}$ and $W_\text{orb}$ such that all observed pulses would be observable given the other parameters:
\begin{equation}\label{eqn:beam_width_calc}
W_\text{spin} = \max_{k\in J} \left\lbrace \beta (\phi^{(k)}) \right\rbrace
\end{equation}
\begin{equation}\label{eqn:mod_width_calc}
W_\text{orb} = \min_{k\in J} \left\lbrace \beta_\text{MD}(\phi_\text{orb}^{(k)},\phi^{(k)}) \right\rbrace
\end{equation}
where $J$ is the set of data indices which contain pulses according to the conditions in \nameref{subsec:timing_analysis}.
The coefficient $C$ was also not fit using MCMC because it can be simply calculated analytically by solving for the roots of the derivative of \autoref{eqn:pfit} which gives
\begin{equation}
C = \frac{\sum_{k=1}^N {I_\text{pred*}^{(k)} I_\text{obs}^{(k)}}/{\sigma_k^2}}{\sum_{k=1}^N \left({I_\text{pred*}^{(k)}}/{\sigma_k} \right)^2}
\end{equation}
where $I_\text{pred*} = I_\text{pred}/C$. This way, the model parameters are reduced to 6 $(\phi_\text{orb}^{(0)}, \phi^{(0)}, i, \alpha, \phi_0, \zeta)$, and the parameter space is quite stable as shown by \autoref{fig:mcmc_corner_plot}.
\autoref{fig:mcmc_corner_plot} is a corner plot for the fitting of \gpmlpt{} folded on $P_B$. There is a linear relationship between $i$ and $\alpha$, but their range is limited by the other parameters. This degeneracy becomes significant if there is only one pulse group, such as for \jlpt{}. If we choose instead $P_A$ as the spin period, no good fit can be found because it is impossible for this model to predict a dynamic pulse profile where two pulse groups are on top of each other. We also tried other beat periods, and only $P_B$ produces a good fit. This makes us confident that $P_B$ is the spin period.
It should be noted that reflections along $\hat{x}$ and/or $\hat{z}$ lead to identical dynamic pulse profiles, so those transformations can be applied to get more pleasing angles.
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{o1s1_bgs_mga_mcmc.png}
\caption{MCMC corner plot for \gpmlpt{}. The axis units are radians. \label{fig:mcmc_corner_plot}}
\end{figure}
\begin{table}
\begin{center}
\begin{tabular}{ccc}
\hline
& Variable & Value \\
\hline
Priors & $P $ & 1265.2197 $\pm$ 0.0002 s \\
& $T $ & 31482.4 $\pm$ 0.2 s \\
& $M_\text{WD} $ & [0.600, 1.200] M$_\odot$ \\
& $M_\text{MD} $ & [0.140, 0.500] M$_\odot$ \\
\hline
Fitted & $i $ & 100.1 $\pm$ 0.6 $^\circ $ \\
& $\alpha $ & 52.1 $\pm$ 0.4 $^\circ $ \\
& $\phi_0 $ & 179.29 $\pm$ 0.04 $^\circ $ \\
& $\zeta $ & 61 $\pm$ 2 $^\circ $ \\
& $W_\text{spin}$ & 65 $\pm$ 3 $^\circ $ \\
& $W_\text{orb} $ & 70 $\pm$ 1 $^\circ $ \\
\hline
Calculated & $A $ & [1.934, 2.551] R$_\odot$ \\
& $R_\text{MD} $ & [0.207, 0.574] R$_\odot$ \\
& $R_c $ & [0.212, 0.267] R$_\odot$ \\
& $R_{rl} $ & [0.513, 0.784] R$_\odot$ \\
& $R_{lc} $ & 86.29970 $\pm$ 0.00002 R$_\odot$ \\
& $R_{A} $ & [0.187, 0.828] R$_\odot$ \\
\hline
\end{tabular}
\end{center}
\caption{Table of results for \gpmlpt{}. Top are priors for the MCMC fit and calculations. Middle are the results of the MCMC fit, with the uncertainties being the $1\sigma$ credible intervals. Bottom are results which do not depend on the fitted geometry.}
\label{tab:fit_results}
\end{table}
\begin{table}
\begin{center}
\begin{tabular}{ccc}
\hline
& Variable & Value \\
\hline
Priors & $P $ & 319.3490 $\pm$ 0.0001 s \\
& $T $ & 14525.5588 $\pm$ 0.0001 s \\
& $i $ & [ 53, 65] $^\circ$ \\
\hline
Fitted & $\alpha $ & [ 42, 70] $^\circ$ \\
& $\phi_0 $ & [-48, -17] $^\circ$ \\
& $\zeta $ & [ 39, 73] $^\circ$ \\
& $W_\text{spin}$ & [ 20, 60] $^\circ$ \\
& $W_\text{orb} $ & [ 67, 90] $^\circ$ \\
\hline
\end{tabular}
\end{center}
\caption{Table of results for \jlpt{}. Top are priors for the MCMC fit and calculations. Middle are the results of the MCMC fit, with the uncertainties being the $1\sigma$ credible intervals.}
\label{tab:j1912_results}
\end{table}
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{J1912-44_best_fit.pdf}
\caption{Modelled dynamic pulse profile for \jlpt{}.\label{fig:j1912}}
\end{figure}
\subsection{Alternative geometries}
The three panels of \autoref{fig:alternatives} are alternative interpretations of the geometry which we investigated but ultimately rejected.
\begin{enumerate}[label=\alph*.]
\item What if the emission is not in the direction of the WD magnetic moment, and the observed flux density is purely a function of $\beta_\text{MD}$, the angle between the WD magnetic moment and the MD? While the predicted pulse group pattern does resemble the observations, no better fit is possible than pictured. This means the emission must be beamed in (or close to) the direction of the WD pole.
\item What if the two pulse groups are from opposite poles of the WD? The two pulse groups must be separated by $\sim$0.5 spin phase, so this interpretation was discarded.
\item What if the emission is 180\textdegree{} bi-directional, regardless of which WD pole sweeps the MD? In this case, we should see emission around orbital phases 0.0 and 0.5, which requires doubling the orbital period to $\sim$18 hours. Although the predicted placement of the pulse groups is correct, there is no qualitative difference between the observed pulses at opposite orbital phases in either total flux density, polarised flux density, or position angle. We therefore favour the $\sim$9 hour orbit.
\end{enumerate}
\begin{figure}[H]
\centering
\includegraphics[width=\linewidth]{alternatives.pdf}
\caption{Best fits for alternative configurations. White indicates no predicted emission, and red or blue indicate emission from different sources. a) The emission is isotropic. b) The two pulse groups originate from opposite WD poles, indicated by red and blue. c) The emission is 180\textdegree{} bi-directional, regardless of which pole sweeps across the MD. In this case, the orbital period is doubled. Red and blue indicates whether the WD is behind or in front of the MD. \label{fig:alternatives}}
\end{figure}
\subsection{Other calculations}
Using a range of MD masses of $M_\text{MD} \in [0.14, 0.5] M_\odot$ and a range of WD masses of $M_\text{WD} \in [0.6, 1.2] M_\odot$, the orbital separation $A$ assuming a circular orbit is
\begin{equation}\label{eqn:orbital_radius}
A = \left(\frac{G (M_\text{MD} + M_\text{WD}) T^2}{4\pi^2}\right)^{\frac{1}{3}} \in [1.93, 2.55] R_\odot
\end{equation}
Approximating the Roche lobe as a sphere, its radius $R_{rl}$ is
\begin{equation}\label{eqn:roche_lobe}
R_{rl} = A (0.46224) \left( \frac{q}{1+q} \right)^{\frac{1}{3}} \in [0.513, 0.784] R_\odot
\end{equation}
where $q = M_\text{MD} / M_\text{WD}$.
Assuming an MD stellar radius $R_\text{MD}$ of
\begin{equation}\label{eqn:stellar_radius}
R_\text{MD} = \left(\frac{M_\text{MD}}{M_\odot}\right)^{0.8} R_\odot \in [0.207, 0.574] R_\odot
\end{equation}
the Roche lobe cannot be filled, therefore the system is not accreting. This is consistent with the lack of x-ray emission detected \cite{Hurleywalker2023}.
The light cylinder radius $R_{lc}$ is the radius at which the angular velocity of the WD matches the speed of light. We find
\begin{equation}\label{eqn:light_cylinder}
R_{lc} = \frac{c P}{2\pi} = 86.29970 \pm 0.00001 R_\odot
\end{equation}
The corotation radius $R_c$ is the radius at which the orbital period is equal to the spin period. Using \autoref{eqn:orbital_radius} we get
\begin{equation}\label{eqn:corotation}
R_c = \left(\frac{G (M_\text{MD} + M_\text{WD}) P^2}{4\pi^2}\right)^{\frac{1}{3}} \in [0.212, 0.267] R_\odot
\end{equation}
To calculate the Alfvén radius $R_A$ we assume $\dot{M}_\text{WD}$ = $10^{-14}$ M$_\odot$ yr$^{-1}$ for a solar-like star \cite{2005ApJ...628L.143W}. Then,
\begin{equation}\label{eqn:alfven}
R_A = \left( \frac{2R^6 B_p^2}{2\dot{M}_\text{MD}\sqrt{G M_\text{WD}}} \right)^{2/7} \in [0.187, 0.828] R_\odot
\end{equation}
The distance $A_\text{WD}$ of the WD from the centre of mass around which the system rotates is
\begin{equation}\label{eqn:dist_from_cm}
A_\text{WD} = \frac{A M_\text{MD}}{M_\text{MD} + M_\text{WD}} \in [0.246, 1.003] R_\odot
\end{equation}
The maximum difference in light travel time $\Delta t$ from the WD between opposite orbital phases is
\begin{equation}\label{eqn:light_travel_time}
\Delta t = 2 \frac{A_\text{WD}}{c} \in [1.15, 4.68] s
\end{equation}
where $c$ is the speed of light. The maximum possible time delay in our model is much smaller because we do not view the orbital plane edge-on. We did not correct the timing residuals for this delay because it is of the scale of our time resolution and insignificant compared with the intrinsic arrival time variability of the source.
In the neutron star pulsar case the emission site is bounded by the last open field lines of the dipolar magnetic field of the WD at the emission site at distance $r_{em}$ from the WD. Assuming that the emission site is within the orbital radius, $r_{em} \leq A \leq 2.55 R_\odot$, the angle $\theta$ between $\vec{\mu}$ and the emission site boundary is
\begin{equation}
\theta = \arcsin\sqrt{\frac{r_{em}}{R_{lc}}} \leq 9.898 ^\circ
\end{equation}
For a dipolar magnetic field in a spherical coordinate system,
\begin{equation}\label{eq:bvec}
\vec{B} \propto 2 \cos \theta \hat{r} + \sin\theta \hat{\theta}
\end{equation}
If the radiation at the emission site is generated tangential to the magnetic field line, then the angle $\rho$ of $\vec{B}$ with respect to $\vec{\mu}$ at the emission site boundary is the half-opening angle of the beam. We calculate the maximum half-opening angle using \autoref{eq:bvec} by
\begin{equation}
\tan(\rho - \theta) = \frac{\sin\theta}{2\cos\theta} = \frac{1}{2}\tan\theta
\end{equation}
to get $\rho \leq 14.9^\circ$.
\section*{Acknowledgements}
%We thank the anonymous referees for their comments, ....
%grant supports
This research is supported by an Australian Government Research Training Program (RTP) Scholarship.
N.H.-W. is the recipient of an Australian Research Council Future Fellowship (project number FT190100231).
N.R. is supported by the European Research Council (ERC) via the Consolidator Grant “MAGNESIA” (No. 817661) and the Proof of Concept ``DeepSpacePulse" (No. 101189496), by the Catalan grant SGR2021-01269
(PI: Graber/Rea), the Spanish grant ID2023-153099NA-I00 (PI: Coti Zelati), and by the program Unidad de Excelencia Maria de Maeztu CEX2020-001058-M.
% Nanda definitely needs to add her grant here.
This scientific work uses data obtained from Inyarrimanha Ilgari Bundara, the CSIRO Murchison Radio-astronomy Observatory. We acknowledge the Wajarri Yamaji People as the Traditional Owners and native title holders of the Observatory site. CSIRO’s ASKAP radio telescope is part of the Australia Telescope National Facility (https://ror.org/05qajvd42). Operation of ASKAP is funded by the Australian Government with support from the National Collaborative Research Infrastructure Strategy. ASKAP uses the resources of the Pawsey Supercomputing Research Centre. Establishment of ASKAP, Inyarrimanha Ilgari Bundara, the CSIRO Murchison Radio-astronomy Observatory and the Pawsey Supercomputing Research Centre are initiatives of the Australian Government, with support from the Government of Western Australia and the Science and Industry Endowment Fund.
% MeerKAT
The MeerKAT telescope is operated by the South African Radio Astronomy Observatory, which is a facility of the National Research Foundation, an agency of the Department of Science and Innovation.
%VLA
The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.
We acknowledge Ben Stappers, Ewan Barr, and Yunpeng Men for contributing to the MeerKAT proposal. We acknowledge Scott Hyman, Tracy Clarke, and Simona Giacintucci for contributing to the VLA proposal. We thank Ingrid Pelisoli and Simone Scaringi for useful discussions on white dwarf evolution.
\section*{Author contributions}
% CsH...
Cs.H performed the orbital timing, spin/beat period analysis, polarisation analysis, developed the geometric model, and contributed to the discussion in the manuscript.
% NR ...
N.R contributed to the formulation of the model and its physical interpretation, and contributed to the discussion in the manuscript.
% NH-W...
N.H.-W led the proposals for the radio data toward \gpmlpt{}, imaged and produced dynamic spectra from the radio data used in this work, and contributed to the formulation of the model and discussion in the manuscript.
% SM...
S.M contributed to the formulation of the model, and contributed to the discussion in the manuscript.
% RP...
R.P generated the observing scripts and calibrated the data for the VLA observations.
% EL
E.L processed the ASKAP data and contributed to the polarisation analysis.
\section*{Competing interests}
The authors declare no competing interests.
\section*{Data availability}
Data supporting the findings of this study are available from the corresponding author upon request.
\section*{Code availability}
The code used for the geometric model and MCMC fitting is available at\\\url{https://github.com/CsanadHorvath/GeometricLPT.git}.\\
An interactive web visualisation is available at \\\url{https://www.desmos.com/3d/6bisspujwe}.
% Separate methods bibliography -- when you're done writing the paper, compile the references for this section, then paste them from the bbl file into the below
%\begin{thebibliography}{10}
%\makeatletter
%\addtocounter{\@listctr}{23}
%\makeatother
%\expandafter\ifx\csname url\endcsname\relax
% \def\url#1{\texttt{#1}}\fi
%\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
%\providecommand{\bibinfo}[2]{#2}
%\providecommand{\eprint}[2][]{\url{#2}}
%\bibitem{2013PASA...30....7T}
%\bibinfo{author}{{Tingay}, S.~J.} \emph{et~al.}
%\newblock \bibinfo{title}{{The Murchison Widefield Array: The Square Kilometre
% Array Precursor at Low Radio Frequencies}}.
%\newblock \emph{\bibinfo{journal}{Publ. Astron. Soc. Aust.}}
% \textbf{\bibinfo{volume}{30}}, \bibinfo{pages}{7} (\bibinfo{year}{2013}).
%
%\bibitem{2018PASA...35...33W}
%\bibinfo{author}{{Wayth}, R.~B.} \emph{et~al.}
%\newblock \bibinfo{title}{{The Phase II Murchison Widefield Array: Design
% overview}}.
%\newblock \emph{\bibinfo{journal}{Publ. Astron. Soc. Aust.}}
% \textbf{\bibinfo{volume}{35}}, \bibinfo{pages}{33} (\bibinfo{year}{2018}).
%\end{thebibliography}
\end{document}
\begin{thebibliography}{10}
\expandafter\ifx\csname url\endcsname\relax
\def\url#1{\texttt{#1}}\fi
\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
\providecommand{\bibinfo}[2]{#2}
\providecommand{\eprint}[2][]{\url{#2}}
\bibitem{2017NatAs...1E..29B}
\bibinfo{author}{{Buckley}, D.~A.~H.}, \bibinfo{author}{{Meintjes}, P.~J.}, \bibinfo{author}{{Potter}, S.~B.}, \bibinfo{author}{{Marsh}, T.~R.} \& \bibinfo{author}{{G{\"a}nsicke}, B.~T.}
\newblock \bibinfo{title}{{Polarimetric evidence of a white dwarf pulsar in the binary system AR Scorpii}}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{1}}, \bibinfo{pages}{0029} (\bibinfo{year}{2017}).
\bibitem{pelisoli2023}
\bibinfo{author}{Pelisoli, I.} \emph{et~al.}
\newblock \bibinfo{title}{A 5.3-min-period pulsing white dwarf in a binary detected from radio to x-rays}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{7}}, \bibinfo{pages}{931--942} (\bibinfo{year}{2023}).
\bibitem{2022Natur.601..526H}
\bibinfo{author}{{Hurley-Walker}, N.} \emph{et~al.}
\newblock \bibinfo{title}{{A radio transient with unusually slow periodic emission}}.
\newblock \emph{\bibinfo{journal}{Nature}} \textbf{\bibinfo{volume}{601}}, \bibinfo{pages}{526--530} (\bibinfo{year}{2022}).
\bibitem{Hurleywalker2023}
\bibinfo{author}{Hurley-Walker, N.} \emph{et~al.}
\newblock \bibinfo{title}{A long-period radio transient active for three decades}.
\newblock \emph{\bibinfo{journal}{Nature}} \textbf{\bibinfo{volume}{619}}, \bibinfo{pages}{487--490} (\bibinfo{year}{2023}).
\bibitem{2022ApJ...934..184R}
\bibinfo{author}{{Ronchi}, M.}, \bibinfo{author}{{Rea}, N.}, \bibinfo{author}{{Graber}, V.} \& \bibinfo{author}{{Hurley-Walker}, N.}
\newblock \bibinfo{title}{{Long-period Pulsars as Possible Outcomes of Supernova Fallback Accretion}}.
\newblock \emph{\bibinfo{journal}{Astrophys. J.}} \textbf{\bibinfo{volume}{934}}, \bibinfo{pages}{184} (\bibinfo{year}{2022}).
\bibitem{genccali2022evolution}
\bibinfo{author}{Gen{\c{c}}ali, A.~A.}, \bibinfo{author}{Ertan, {\"U}.} \& \bibinfo{author}{Alpar, M.~A.}
\newblock \bibinfo{title}{Evolution of the long-period pulsar gleam-x j162759. 5--523504.3}.
\newblock \emph{\bibinfo{journal}{Monthly Notices of the Royal Astronomical Society: Letters}} \textbf{\bibinfo{volume}{513}}, \bibinfo{pages}{L68--L71} (\bibinfo{year}{2022}).
\bibitem{de2025sporadic}
\bibinfo{author}{de~Ruiter, I.} \emph{et~al.}
\newblock \bibinfo{title}{Sporadic radio pulses from a white dwarf binary at the orbital period}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \bibinfo{pages}{1--13} (\bibinfo{year}{2025}).
\bibitem{hurley20242}
\bibinfo{author}{Hurley-Walker, N.} \emph{et~al.}
\newblock \bibinfo{title}{A 2.9 hr periodic radio transient with an optical counterpart}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal Letters}} \textbf{\bibinfo{volume}{976}}, \bibinfo{pages}{L21} (\bibinfo{year}{2024}).
\bibitem{rodrigues2025}
\bibinfo{author}{{Rodriguez}, A.~C.}
\newblock \bibinfo{title}{{Spectroscopic detection of a 2.9-hour orbit in a long-period radio transient}}.
\newblock \emph{\bibinfo{journal}{\aap}} \textbf{\bibinfo{volume}{695}}, \bibinfo{pages}{L8} (\bibinfo{year}{2025}).
\bibitem{potter_and_buckley_2018}
\bibinfo{author}{Potter, S.~B.} \& \bibinfo{author}{Buckley, D. A.~H.}
\newblock \bibinfo{title}{Time series photopolarimetry and modelling of the white dwarf pulsar in ar scorpii}.
\newblock \emph{\bibinfo{journal}{Monthly Notices of the Royal Astronomical Society}} \textbf{\bibinfo{volume}{481}}, \bibinfo{pages}{2384--2392} (\bibinfo{year}{2018}).
\bibitem{qu2025magnetic}
\bibinfo{author}{Qu, Y.} \& \bibinfo{author}{Zhang, B.}
\newblock \bibinfo{title}{Magnetic interactions in white dwarf binaries as mechanism for long-period radio transients}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal}} \textbf{\bibinfo{volume}{981}}, \bibinfo{pages}{34} (\bibinfo{year}{2025}).
\bibitem{geng2016model}
\bibinfo{author}{Geng, J.-J.}, \bibinfo{author}{Zhang, B.} \& \bibinfo{author}{Huang, Y.-F.}
\newblock \bibinfo{title}{A model of white dwarf pulsar ar scorpii}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal Letters}} \textbf{\bibinfo{volume}{831}}, \bibinfo{pages}{L10} (\bibinfo{year}{2016}).
\bibitem{2021NatAs...5..648S}
\bibinfo{author}{{Schreiber}, M.~R.}, \bibinfo{author}{{Belloni}, D.}, \bibinfo{author}{{G{\"a}nsicke}, B.~T.}, \bibinfo{author}{{Parsons}, S.~G.} \& \bibinfo{author}{{Zorotovic}, M.}
\newblock \bibinfo{title}{{The origin and evolution of magnetic white dwarfs in close binary stars}}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{5}}, \bibinfo{pages}{648--654} (\bibinfo{year}{2021}).
\bibitem{2024MNRAS.527.3826P}
\bibinfo{author}{{Pelisoli}, I.} \emph{et~al.}
\newblock \bibinfo{title}{{Unveiling the white dwarf in J191213.72 - 441045.1 through ultraviolet observations}}.
\newblock \emph{\bibinfo{journal}{\mnras}} \textbf{\bibinfo{volume}{527}}, \bibinfo{pages}{3826--3836} (\bibinfo{year}{2024}).
\bibitem{doi:10.1126/sciadv.adp6351}
\bibinfo{author}{Men, Y.}, \bibinfo{author}{McSweeney, S.}, \bibinfo{author}{Hurley-Walker, N.}, \bibinfo{author}{Barr, E.} \& \bibinfo{author}{Stappers, B.}
\newblock \bibinfo{title}{A highly magnetized long-period radio transient exhibiting unusual emission features}.
\newblock \emph{\bibinfo{journal}{Science Advances}} \textbf{\bibinfo{volume}{11}}, \bibinfo{pages}{eadp6351} (\bibinfo{year}{2025}).
\bibitem{1969ApL.....3..225R}
\bibinfo{author}{{Radhakrishnan}, V.} \& \bibinfo{author}{{Cooke}, D.~J.}
\newblock \bibinfo{title}{{Magnetic Poles and the Polarization Structure of Pulsar Radiation}}.
\newblock \emph{\bibinfo{journal}{Astrophys. J. Lett.}} \textbf{\bibinfo{volume}{3}}, \bibinfo{pages}{225} (\bibinfo{year}{1969}).
\bibitem{1997A&A...327..155G}
\bibinfo{author}{{Gangadhara}, R.~T.}
\newblock \bibinfo{title}{{Orthogonal polarization mode phenomenon in pulsars.}}
\newblock \emph{\bibinfo{journal}{\aap}} \textbf{\bibinfo{volume}{327}}, \bibinfo{pages}{155--166} (\bibinfo{year}{1997}).
\bibitem{2005ApJ...628L.143W}
\bibinfo{author}{{Wood}, B.~E.}, \bibinfo{author}{{M{\"u}ller}, H.~R.}, \bibinfo{author}{{Zank}, G.~P.}, \bibinfo{author}{{Linsky}, J.~L.} \& \bibinfo{author}{{Redfield}, S.}
\newblock \bibinfo{title}{{New Mass-Loss Measurements from Astrospheric Ly{\ensuremath{\alpha}} Absorption}}.
\newblock \emph{\bibinfo{journal}{\apjl}} \textbf{\bibinfo{volume}{628}}, \bibinfo{pages}{L143--L146} (\bibinfo{year}{2005}).
\bibitem{2024arXiv241116606W}
\bibinfo{author}{{Wang}, Z.} \emph{et~al.}
\newblock \bibinfo{title}{{Detection of X-ray Emission from a Bright Long-Period Radio Transient}}.
\newblock \emph{\bibinfo{journal}{Nature}} (\bibinfo{year}{2025}).
\bibitem{Lyutikov2020}
\bibinfo{author}{{Lyutikov}, M.} \emph{et~al.}
\newblock \bibinfo{title}{{Magnetospheric interaction in white dwarf binaries AR Sco and AE Aqr}}.
\newblock \emph{\bibinfo{journal}{arXiv e-prints}} \bibinfo{pages}{arXiv:2004.11474} (\bibinfo{year}{2020}).
\bibitem{Geng2016}
\bibinfo{author}{{Geng}, J.-J.}, \bibinfo{author}{{Zhang}, B.} \& \bibinfo{author}{{Huang}, Y.-F.}
\newblock \bibinfo{title}{{A Model of White Dwarf Pulsar AR Scorpii}}.
\newblock \emph{\bibinfo{journal}{\apjl}} \textbf{\bibinfo{volume}{831}}, \bibinfo{pages}{L10} (\bibinfo{year}{2016}).
\bibitem{Katz2017}
\bibinfo{author}{{Katz}, J.~I.}
\newblock \bibinfo{title}{{AR Sco: A Precessing White Dwarf Synchronar?}}
\newblock \emph{\bibinfo{journal}{\apj}} \textbf{\bibinfo{volume}{835}}, \bibinfo{pages}{150} (\bibinfo{year}{2017}).
\bibitem{takata2017model}
\bibinfo{author}{Takata, J.}, \bibinfo{author}{Yang, H.} \& \bibinfo{author}{Cheng, K.}
\newblock \bibinfo{title}{A model for ar scorpii: emission from relativistic electrons trapped by closed magnetic field lines of magnetic white dwarfs}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal}} \textbf{\bibinfo{volume}{851}}, \bibinfo{pages}{143} (\bibinfo{year}{2017}).
\bibitem{2025ApJ...981...34Q}
\bibinfo{author}{{Qu}, Y.} \& \bibinfo{author}{{Zhang}, B.}
\newblock \bibinfo{title}{{Magnetic Interactions in White Dwarf Binaries as Mechanism for Long-period Radio Transients}}.
\newblock \emph{\bibinfo{journal}{\apj}} \textbf{\bibinfo{volume}{981}}, \bibinfo{pages}{34} (\bibinfo{year}{2025}).
\bibitem{buckley2017polarimetric}
\bibinfo{author}{Buckley, D.}, \bibinfo{author}{Meintjes, P.}, \bibinfo{author}{Potter, S.}, \bibinfo{author}{Marsh, T.} \& \bibinfo{author}{G{\"a}nsicke, B.}
\newblock \bibinfo{title}{Polarimetric evidence of a white dwarf pulsar in the binary system ar scorpii}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{1}}, \bibinfo{pages}{0029} (\bibinfo{year}{2017}).
\bibitem{du2019constraining}
\bibinfo{author}{Du~Plessis, L.}, \bibinfo{author}{Wadiasingh, Z.}, \bibinfo{author}{Venter, C.} \& \bibinfo{author}{Harding, A.~K.}
\newblock \bibinfo{title}{Constraining the emission geometry and mass of the white dwarf pulsar ar sco using the rotating vector model}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal}} \textbf{\bibinfo{volume}{887}}, \bibinfo{pages}{44} (\bibinfo{year}{2019}).
\bibitem{2025NatAs...9..393L}
\bibinfo{author}{{Lee}, Y.~W.~J.} \emph{et~al.}
\newblock \bibinfo{title}{{The emission of interpulses by a 6.45-h-period coherent radio transient}}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{9}}, \bibinfo{pages}{393--405} (\bibinfo{year}{2025}).
\bibitem{sam2025}
\bibinfo{author}{{Samuel J. McSweeney}} \emph{et~al.}
\newblock \bibinfo{title}{{Confirmation of J1755\ensuremath{-}2527 as a long period radio transient}} (\bibinfo{year}{in prep}).
\bibitem{caleb2024emission}
\bibinfo{author}{Caleb, M.} \emph{et~al.}
\newblock \bibinfo{title}{An emission-state-switching radio transient with a 54-minute period}.
\newblock \emph{\bibinfo{journal}{Nature Astronomy}} \textbf{\bibinfo{volume}{8}}, \bibinfo{pages}{1159--1168} (\bibinfo{year}{2024}).
\bibitem{dong2024discoverynearby421stransient}
\bibinfo{author}{Dong, F.~A.} \emph{et~al.}
\newblock \bibinfo{title}{The discovery of a nearby 421~s transient with chime/frb/pulsar} (\bibinfo{year}{2024}).
\bibitem{2021PASA...38....9H}
\bibinfo{author}{{Hotan}, A.~W.} \emph{et~al.}
\newblock \bibinfo{title}{{Australian square kilometre array pathfinder: I. system description}}.
\newblock \emph{\bibinfo{journal}{Publ. Astron. Soc. Aust.}} \textbf{\bibinfo{volume}{38}}, \bibinfo{pages}{e009} (\bibinfo{year}{2021}).
\bibitem{2016mks..confE...1J}
\bibinfo{author}{{Jonas}, J.} \& \bibinfo{author}{{MeerKAT Team}}.
\newblock \bibinfo{title}{{The MeerKAT Radio Telescope}}.
\newblock In \emph{\bibinfo{booktitle}{MeerKAT Science: On the Pathway to the SKA}}, \bibinfo{pages}{1} (\bibinfo{year}{2016}).
\bibitem{2011ApJ...739L...1P}
\bibinfo{author}{{Perley}, R.~A.}, \bibinfo{author}{{Chandler}, C.~J.}, \bibinfo{author}{{Butler}, B.~J.} \& \bibinfo{author}{{Wrobel}, J.~M.}
\newblock \bibinfo{title}{{The Expanded Very Large Array: A New Telescope for New Science}}.
\newblock \emph{\bibinfo{journal}{\apjl}} \textbf{\bibinfo{volume}{739}}, \bibinfo{pages}{L1} (\bibinfo{year}{2011}).
\bibitem{perley2017accurate}
\bibinfo{author}{Perley, R.~A.} \& \bibinfo{author}{Butler, B.~J.}
\newblock \bibinfo{title}{An accurate flux density scale from 50 mhz to 50 ghz}.
\newblock \emph{\bibinfo{journal}{The Astrophysical Journal Supplement Series}} \textbf{\bibinfo{volume}{230}}, \bibinfo{pages}{7} (\bibinfo{year}{2017}).
\bibitem{perley2022evla}
\bibinfo{author}{Perley, R.}, \bibinfo{author}{Greisen, E.} \& \bibinfo{author}{Hugo, B.}
\newblock \bibinfo{title}{{EVLA Memo 219 Enabling MeerKAT Polarimetric Imaging in AIPS}}.
\newblock \bibinfo{type}{Tech. Rep.}, \bibinfo{institution}{National Radio Astronomy Observatory} (\bibinfo{year}{2022}).
\bibitem{2014MNRAS.444..606O}
\bibinfo{author}{{Offringa}, A.~R.} \emph{et~al.}
\newblock \bibinfo{title}{{WSCLEAN: an implementation of a fast, generic wide-field imager for radio astronomy}}.
\newblock \emph{\bibinfo{journal}{Mon. Not. R. Astron. Soc.}} \textbf{\bibinfo{volume}{444}}, \bibinfo{pages}{606--619} (\bibinfo{year}{2014}).
\bibitem{2024zndo..13626183P}
\bibinfo{author}{{Pritchard}, J.}
\newblock \bibinfo{title}{{askap-vast/dstools: v1.0.0}} (\bibinfo{year}{2024}).
\bibitem{2013PASP..125..306F}
\bibinfo{author}{{Foreman-Mackey}, D.}, \bibinfo{author}{{Hogg}, D.~W.}, \bibinfo{author}{{Lang}, D.} \& \bibinfo{author}{{Goodman}, J.}
\newblock \bibinfo{title}{{emcee: The MCMC Hammer}}.
\newblock \emph{\bibinfo{journal}{\pasp}} \textbf{\bibinfo{volume}{125}}, \bibinfo{pages}{306} (\bibinfo{year}{2013}).
\end{thebibliography}