On the irreducibility of slice algebraic sets
Extracted source text · arXiv 2507.21674
\documentclass[12pt]{amsart}
\makeatletter \@addtoreset{equation}{section}
\def\theequation{\thesection.\arabic{equation}}
\makeatother \makeatletter
\newtheorem{theorem}{Theorem}[section]
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{example}[theorem]{Example}
\newtheorem*{theorem*}{Theorem}
\newtheorem*{proposition*}{Proposition}
\def\prodstar{\prod\hskip -2.45ex * \hskip 1.3 ex}
\def\pr1{\prod\hskip -2.07ex * \hskip 0.9 ex}
\newcommand{\dm}{{\em Proof. }}
\newcommand{\aaa}{ \textbf{\em a}}
%\DeclareMathOperator{\aaa}{\textbf{\em a}}
\newcommand{\cvd}{\hfill $\Box $}
\def\Kls{\mbox{{\rm K-}}\!ambdaim_{\hspace{-.4cm}z\rightarrow\sigma}}
\def\Klss{\mbox{{\rm K-}}\!\lim_{\hspace{-.4cm}w\rightarrow\sigma}}
\def\Klsss{\mbox{{\rm K-}}\!\lim_{\hspace{-.4cm}w\rightarrow 0}}
\oddsidemargin = 8pt
\evensidemargin = 8pt
\usepackage{latexsym} \usepackage{amsmath} \usepackage{amssymb}
\usepackage{amsthm}
\usepackage{centernot}
\usepackage{cancel}\usepackage{color}
\pagestyle{plain}
\addtolength{\textwidth}{3cm}
\begin{document}
\noindent
\title{On the irreducibility of slice algebraic sets}
\author{Anna Gori $^1$}
\thanks{$^1$ Dipartimento di Matematica - Universit\`a di Milano,
Via Saldini 50, 20133 Milano, Italy}
\author{Giulia Sarfatti $^2$}
\thanks{ $^2$ DIISM - Universit\`a Politecnica delle Marche,
Via Brecce Bianche 12, 60131, Ancona, Italy}
\author{Fabio Vlacci $^3$}
\thanks{$^3$ MIGe - Universit\`a di Trieste, Piazzale Europa 1, 34100, Trieste, Italy}
\begin{abstract}
{
In the present paper we investigate the {relations} between {irreducible slice algebraic} sets in $\mathbb{H}^n$ and {quasi prime} right ideals of the ring of slice regular polynomials in $n$ quaternionic variables. We
provide algebraic conditions
on right ideals of slice regular polynomials {which guarantee the}
irreducibility of the corresponding {slice} algebraic sets
and show that radical ideals associated with irreducible slice algebraic sets are quasi prime.
Furthermore we establish that this correspondence is an equivalence
in the case of principal right ideals.}
\end{abstract}
\keywords{Nullstellensatz, quaternionic slice regular polynomials, irreducible algebraic sets\\
{\bf MSC:} 30G35, 16S36}
\maketitle
\section*{Acknowledgments}
The authors are partially supported by:
%The first and third authors were partly supported
GNSAGA-INdAM via the project ``Hypercomplex function theory and
applications''; the first author is also partially supported by MUR
project PRIN 2022 ``Real and Complex Manifolds: Geometry and
Holomorphic Dynamics'', the second and the third authors are also
partially supported by MUR projects PRIN 2022 ``Interactions between
Geometric Structures and Function Theories''.
%\title{}
%\author
\section{Introduction}
%\begin{remark}
%{\color{blue} Io scriverei right ideal dappertutto}\\
\noindent In Algebraic Geometry ideals of polynomials and algebraic varieties are deeply related and
this correspondence often refers to the Hilbert Nullstellensatz.
\noindent {In the quaternionic setting, a version of the so called {\em Strong} Hilbert Nullstellensatz was proven, independently, in \cite{israeliani-3} and in \cite{Nul2}. The rings of polynomials taken into account in these two papers are different but isomorphic. Let $\mathbb H$ denote the skew-field of quaternions. In the present paper we will consider, as in \cite{Nul2},
%the so called
%{\em slice regular polynomials:}
polynomial functions with quaternionic coefficients defined on $\mathbb H^n$ as %(here $\mathbb H$ denotes the skew-field of quaternions) of the form
\[P(q_1,\ldots, q_n)=
\sum_{ \substack{\ell_1=0,\ldots, L_1 \\
\cdots\\ \ell_n=0,\ldots, L_n} }{q_1}^{\ell_1}\cdots {q_n}^{\ell_n}a_{\ell_1,\ldots,\ell_n}, \]
% $a_{\ell_1,\ldots,\ell_n}\in\mathbb{H}$,
%where $\deg_{{q_{\ell_j}}}P:=L_j$.\\
which belong to the class of {\em slice regular functions} according to Definition 2.5 in \cite{severalvariables}.
The set $\mathbb{H}[q_1,\ldots,q_n]$ of slice regular polynomials in $n$ quaternionic variables
can be equipped with a suitable
notion of (non-commutative) product, called {\em slice product} and denoted by the symbol $*$, endowing $(\mathbb{H}[q_1,\ldots,q_n], +,*)$ with the structure of non-commutative ring.
%used in the one-variable
%case. %We refer to \cite{severalvariables} for a precise definition,
%We shortly recall here how it works for
%% monomials, and use the fact that we can extend it by linearity to define it for polynomials and
%slice regular polynomials in two quaternionic variables.
%If $P(q_1,q_2)=\sum_{ \substack{n=0,\ldots,N_1 \\ m=0,\ldots,N_2} } q_1^nq_2^m a_{n,m}$ and
% $Q(q)=\sum_{ \substack{n=0,\ldots,L_1 \\ m=0,\ldots,L_2} }q_1^nq_2^m b_{n,m}$ are two
% slice regular polynomials,
% then the {\em $*$-product} of $P$ and $Q$ is the slice regular polynomial
% defined by
%$$
%P*Q (q_1,q_2):=\sum_{ \substack{n=0,\ldots,N_1+L_1 \\ m=0,\ldots,N_2+L_2}}q_1^nq_2^m\sum_{ \substack{r=0,\ldots,n \\ s=0,\ldots,m} }a_{r,s} b_{n-r,m-s}.
%$$
The set of slice regular polynomials vanishing on a given subset $Z$ of $\mathbb{H}^n$ is not in general an ideal of
$\mathbb{H}[q_1,\ldots,q_n]$. Therefore we define $\mathcal{J}(Z)$ to be the right ideal
% associated with a given subset $Z$ of $\mathbb{H}^n$, namely the ideal
generated by
slice regular polynomials in several variables which vanish on the subset $Z$. The choice of working with right ideals is due to the definition of slice regularity we refer to (see \cite{severalvariables}).
On the other hand, starting from a right ideal $I$ of $\mathbb{H}[q_1,\ldots,q_n]$, we
define $\mathcal{V}(I)$ to be the set of common zeros of slice regular polynomials
in $I$ and
%{If $\mathcal{V}(I)$ is the algebraic set in $\mathbb{H}^n$ associated with the ideal $I$
% in $\mathbb{H}[q_1,\ldots,q_n]$},
%we denote
$\mathcal V_c(I)$ to be the subset of $\mathcal V(I)$ whose points have commuting components. If $\mathbb S$ denotes the two-dimensional sphere of imaginary units in $\mathbb H$, namely $\mathbb{S}=\{J\in\mathbb{H}\ J^2=-1\}$, then
$\mathcal{V}_c(I)=\mathcal{V}(I)\cap\bigcup_{K\in \mathbb{S}}\mathbb{C}_K^n$,
where $\mathbb{C}_K:=\mathbb{R}+\mathbb{R} K$ can be
identified with the complex plane $\mathbb{C}$ { for any $K \in \mathbb S$}.
\\
%The subsets of $\mathbb{H}^n$ considered
%In \cite{israeliani-1}, the authors consider the vanishing sets of quaternionic polynomials with central variables; this means that these sets in $\mathbb{H}^n$
%consist only of points
%with commuting components and thus have rather limited geometric properties.
%In our setting,
%% we can actually consider any
%vanishing sets of slice regular quaternionic polynomials can also contain points with non--commutative components.
%, therefore also sets of points
%with not-commuting components.
%A reasonable approach to study zeros of a polynomial is to begin with the description of zeros of its factors.
%However,
%when studying
%the
%vanishing set of a $*$-product of slice regular polynomials,
%the problem that zeros of this type
% it turns out that this set
%is not always immediately deducible from
%the vanishing sets of the polynomial factors. % involved in the $*$-product.
%After some preliminary results, the present paper focuses on two
%classes of vanishing sets %in several quaternionic variables
%%; in particular, different types of vanishing sets are considered, primarily those
%which have specific
%symmetries and are preserved
%under the $*$-product. \\
%\noindent For any $(a_1,\ldots,a_n)\in \mathbb {H}^n$, we define
%\[\mathbb S_{(a_1,\ldots,a_n)}=\{(g^{-1}a_1g, \ldots, g^{-1}a_ng) : g \in \mathbb{H}\setminus\{0\} \}\]
%%by $\mathbb S_{(a_1,\ldots,a_n)}
%to be the set obtained by simultaneously rotating each coordinate of the point $(a_1,\ldots,a_n)$.
%Any set of this form will be called a {\em spherical set} in analogy with the terminology adopted
%in the one quaternionic variable setting.
%In particular, if $(a_1,\ldots,a_n)$ has commuting components, i.e. $a_la_m=a_ma_l$ for any $l,m$, then the spherical set will be said an {\em arranged} spherical set.
%Any set of the form $\mathbb S_{(a_1,\cdots,a_k)}\times \{a_{k+1}\} \times \cdots\times \{ a_{n}\}$, with $\mathbb S_{(a_1,\cdots,a_k)}$ an arranged spherical set and
%$a_la_m=a_ma_l$ for any $l,m\geq k+1$,
%will be called a {\em balloon}.
%
%\noindent {\color{red} In order to prove that balloons and subsets of the form $\mathbb S_{a_1}\times \cdots \times \mathbb S_{a_k}\times \cdots \times \{a_{k+1}\}\times \{a_n\}$ are zero loci of right ideals in $\mathbb H[q_1,\ldots,q_n]$, we provide an explicit description of slice regular polynomials vanishing on these sets.
%}
%
%%associated with $(a_1,\cdots,a_n)$.
%%We say %introduce the following notion
%%that a subset $U$ of $\mathbb H^n$ is \emph{balloon symmetric} if, for any $(a_1,\ldots,a_n)\in U$, %with
%%with $a_\ell a_m\neq a_m a_\ell$ for some $\ell,m\in\{1,\ldots,n\}$, if
%%$t=\min\{ k : \ a_ra_s=a_sa_r \text{ for any } \ r,s\in\{k+1,\ldots,n \}\}$ greater than 1,
%%at least one
%%arranged spherical set
%%$\mathbb S_{(\widetilde{a_1},\cdots, \widetilde{a_t })}$ with $\widetilde{a_l}\in\mathbb{S}_{a_l}, \ l=1,\ldots, t$, the
%%appropriate \emph{balloon} built starting from $(a_1,\cdots,a_n)$ (see Definition \ref{BS} for the precise construction)
%%$\mathbb{S}_{(\widetilde{ a_1},\ldots,\widetilde{ a_t})}\times\{a_{t+1}\}\times\cdots\times \{a_n\}$
%%is contained in $U$.
%%Any set of the form $\mathbb S_{(a_1,\cdots,a_t)}\times \{a_{t+1}\} \times \cdots\times \{ a_{n}\}$,
%%where $\mathbb S_{(a_1,\cdots,a_t)}$ is an arranged spherical set and $a_la_m=a_ma_l$ for any $t+1\leq l,m\leq n$,
%%$ t:=\min\{ k : \ a_ra_s=a_sa_r \text{ for any } \ r,s\in\{k+1,\ldots,n \}\}\geq 2$,
%%will be called a (non--trivial) {\em balloon}.
%%We say %introduce the following notion
%%that a subset $U$ of $\mathbb H^n$ is \emph{balloon symmetric} if for any $(a_1,\ldots,a_n)\in U$
%% with
%%with $a_\ell a_m\neq a_m a_\ell$ for some $\ell,m\in\{1,\ldots,n\}$, if
%%$t=\min\{ k : \ a_ra_s=a_sa_r \text{ for any } \ r,s\in\{k+1,\ldots,n \}\}$ greater than 1,
%%{\color{red} at least one balloon obtained starting from the point $(a_1,\ldots,a_n)$ is contained in $U$ .\\}
%%then at least one % {\color{red} for at least one} arranged spherical set
%%$\mathbb S_{(\widetilde{a_1},\cdots, \widetilde{a_t })}$ with $\widetilde{a_l}\in\mathbb{S}_{a_l}, \ l=1,\ldots, t$, the
%%\emph{balloon} $\mathbb{S}_{(\widetilde{ a_1},\ldots,\widetilde{ a_t})}\times\{a_{t+1}\}\times\cdots\times \{a_n\}$ is contained in $U$.
%\noindent Given a right ideal $I\subseteq \mathbb{H}[q_1,\ldots, q_n]$, we prove that for any $(a_1,\ldots,a_n)\in%either $\mathcal V(I)$ consists only of points with commuting components or
%\mathcal V(I)$ appropriate balloons, built starting from $(a_1,\cdots,a_n),$ are contained in $\mathcal{V}(I)$. %\end{Proposition}
%\noindent This is the key ingredient to show that the right ideal $\mathcal{J}(\mathcal{ V }(I))$ coincides
%with the set of polynomials vanishing on $\mathcal{ V }(I)$.
\noindent Introducing, as in \cite{israeliani-1}, the radical $\sqrt{I}$ of a right ideal $I$ as the intersection
% right ideal $L$ in $\mathbb{H}[q_1,\ldots,q_n]$ is {\em
% completely prime} if for any $P,Q \in
% \mathbb{H}[q_1,\ldots,q_n]$ such that $P*Q \in L$ and
% $P*L\subseteq L$ we have that $P\in L$ or $Q\in L$, we have that the
% {\em right radical} $\sqrt{I}$ of $I$ is the intersection of
of all {\em completely prime} right ideals (see Definition \ref{cprime}) that contain $I$,
the Strong Hilbert Nullstellensatz in $\mathbb{H}^n$ reads as follows
\begin{theorem*}[{\cite[Theorem 4.3]{Nul2}}]
Let $I$ be a right ideal in $\mathbb{H}[q_1,\ldots, q_n]$. Then
\[\mathcal{J}(\mathcal{V}(I))=\sqrt{I}.\]
\end{theorem*}
\noindent Motivated by this connection between ideals of polynomials and subsets of $\mathbb H^n$, in \cite{Nul2} we introduced the notion of {\em slice algebraic
sets} which defines a Zariski-type topology on $\mathbb H^n$ and we proved that every subset of the form
$\mathcal{V}(I)$ or $\mathcal V_c(I)$ is a slice algebraic set.}
The aim of the present paper is to investigate the relations between the geometric notion of {\em irreducibility}, namely the fact that
%{ the impossibility to decompose $\mathcal{V}(I)$ as the union of two different algebraic sets},
{$\mathcal{V}(I)$ (or $\mathcal V_c(I)$) cannot be obtained as the union of two different slice algebraic sets},
with the algebraic properties of the corresponding right ideal $I$ of {slice regular} polynomials.
%{\color{red} TOGLIEREI with respect to the $*$ product as introduced in Section \ref{prel}.}
%All the statement concerning ideals will refer to {\em right} ideals with respect of the $*$ product as introduced in Section \ref{prel}.
In the complex setting, it is a well-known fact that an algebraic variety is irreducible if and only if its associated ideal is {\em prime}. % {\color{blue}$*$}.
In view of the statement of the Strong Nullstellensatz, a first guess would be to replace prime ideals with completely prime ideals to characterize irreducible slice algebraic sets in $\mathbb{H}^n$.
%prime one could think that
%
% { Due to the non-commutativity of the multiplication operation, denoted by the symbol $*$, the notion of {\em completely prime} (right) ideal in $(\mathbb{H}[q_1,\ldots,q_n], +, *)$ is used as a counterpart of prime ideal. In fact
% one can define the radical of an ideal $I$ to be the intersection of all completely prime ideals containing $I$ (see Section \ref{prel}).}
However, this translation does not work:
there are irreducible slice algebraic sets whose corresponding ideals are not completely prime.
As an example consider the set $\mathbb{S}$ which is the zero locus
of slice regular polynomials belonging to the right ideal generated by $q^2+1$
in $\mathbb H[q]$. While $\mathbb{S}$ is irreducible, %that is it is irreducible, %(see Proposition \ref{boh}),
the right ideal of $\mathbb{H}[q]$ generated by $q^2+1=(q+i)*(q-i)=:(q+i)^s$ is not {completely prime}. %{\color{blue} and is {\em reducible} as ideal}
%(see Definiton \ref{cprime}).% {\color{blue} and \ref{redI}}). %\marginpar{se lo teniamo, anticipiamo il concetto {\color{blue}$*$}}
%
%
%
%, some interesting results which show the relation between the geometric notion of irreducibility for an algebraic set in $\mathbb{H}^n$ and the corresponging algebraic characterizion of the associated ideals have been obtained in this paper.
%
%
%
%%If $\mathcal{V}(I)$ is an algebraic set in $\mathbb{H}^n$,
{In order to fill the gap in the correspondence between completely prime ideals and irreducible slice algebraic sets}, we introduce the new and wider class of {\em quasi prime} right ideals (see Definition \ref{qpideal}).
%, containing that of completely prime ideals, related to the geometry of the zero sets of slice regular polynomials.
This notion is linked to the geometry of the zero sets of slice regular polynomials and turns out to be the proper tool to
%The new notion of {\em quasi prime (right) ideals}
% seems to be the proper tool to introduce in order to
characterize the irreducibility of algebraic sets. In fact,
%In Proposition \ref{vc-v}
in one direction we prove
%e show %(see Prop. \ref{cp-qp})
% that any completely prime ideal is quasi prime.
\begin{theorem*} Let $I$ be a radical right ideal of $(\mathbb{H}[q_1,\ldots,q_n], +, *)$ such that $\mathcal{V}_c(I)$ is irreducible, then $I$ is a quasi prime ideal.
\end{theorem*}
%\noindent We show that the class of quasi prime ideals contains the one of completely prime ideals.
% Using this relevant fact we are able to prove the following result that provides
\noindent To provide a converse of the previous result, starting from a right ideal $I$, we introduce
the {\em symmetrized ideal} $\mathcal{S}(I)$ as the right ideal generated by the symmetrizations $P^s$ (see Definition \ref{R-coniugata2}) of slice regular polynomials $P$ belonging to the ideal $I$. \\
Given a set $Z_c\subseteq \mathbb H^n$ of points with commuting components, we define its {\em symmetrization} as $\mathbb{S}_{Z_c}:=\bigcup_{\textbf{{\em a}} \in Z_c}\mathbb S_{\textbf {\em a}}$, where
$\mathbb S_{\textbf {\em a}}=\{ (ga_1g^{-1},\ldots, ga_ng^{-1}) \ : \ g\in\mathbb{H}\setminus\{0\}\}$ if $\textbf {\em a}=(a_1,\ldots, a_n)$.
%
{ Symmetrized ideals and symmetrized subsets are related by the following result.
\begin{theorem*}%\label{simmalg}
Let $I$ be a right ideal of $\mathbb H[q_1,\ldots,q_n]$. Then the symmetrization of $\mathcal V_c(I)$ is a slice algebraic set, more precisely
%\begin{equation*}%\label{inclusion}
$ \mathbb{S}_{\mathcal{V}_c(I)}= \mathcal{V}_c( \mathcal{S}(I)).$
%:=\bigcup_{\mathbf{a} \in \mathcal{V}_c(J)}\mathbb S_{\mathbf a}
%=\bigcup_{\mathbf{a}\in\bigcap\limits_{K\in J}\mathcal{V}_c(\langle K\rangle)} \mathbb S_{\mathbf a}
%\end{equation*}
\end{theorem*}
}% (see Proposition \ref{simmalg})
\noindent Using this relevant fact, we are able to prove the following
\begin{theorem*}%\marginpar{\color{blue} Chiamarlo teorema?}
Let $I$ be a {quasi} prime {and radical} right ideal of $\mathbb H[q_1,\ldots,q_n]$. If the radical of $\mathcal S(I)$ is quasi prime, then $\mathcal V_c(I)$ is irreducible.
\end{theorem*}
\noindent Furthermore, we prove that the irreducibility of $\mathcal{V}_c(I)$ implies the irreducibility of $\mathcal{V}(I)$ (see Proposition \ref{vc-v}) and provide examples that the converse is not true.
\noindent We then consider the case of principal right ideals, namely those generated by a single polynomial. Using factorization properties of the generator, we completely characterize principal ideals whose corresponding algebraic sets are irreducible.
% is given in the following
\begin{theorem*} A principal right ideal $I$ of $\mathbb H[q_1,\ldots,q_n]$ is quasi prime if and only if $I$ is radical
and $\mathcal{V}_c(I)$ is irreducible.
\end{theorem*}
%
%{\color{red} ERASE}
%\noindent Where also a characterization of the generator of principal ideals $I$ with $\mathcal V_c(I)$ irreducible is provided.\\
%
%
%
\noindent The paper is organized as follows: after recalling in Section 2 some background results concerning ideals of
$(\mathbb{H}[q_1,\ldots, q_n], +,*)$,
together with several preliminary results and definitions, in Section 3 we introduce the two notions of symmetrization of an ideal and of a slice algebraic set and we link these two concepts in Theorem \ref{simmalg}. In Section 4 we further investigate the relations between algebraic sets and ideals, and in Section 5
we provide necessary and sufficient conditions for the irreducibility of a slice algebraic set (see
Theorem \ref{VirredIprime}, Theorem \ref{nosfere}, Theorem \ref{cpirred} and Corollary \ref{coro3}).
In Section 6 we focus our attention on principal right ideals giving, in Theorem \ref{principal}, a complete characterization of right ideals $I$ whose corresponding slice algebraic sets $\mathcal V_c(I)$ are irreducible.
%, in terms of { factorization properties} of the generator and of the fact that $I$ is a quasi prime ideal.
%\end{remark}
\section{Preliminaries}\label{prel}
\noindent Let $\mathbb{H}{=\mathbb{R}+i\mathbb{R}+j\mathbb{R}+k\mathbb{R}}$ denote the skew field of
quaternions and let $\mathbb{S}=\{J \in \mathbb{H} \ : \ J^2=-1\}$ be
the two dimensional sphere of quaternionic imaginary units. Then
\[ \mathbb{H}=\bigcup_{J\in \mathbb{S}}(\mathbb{R}+\mathbb{R} J),
%\hskip 1 cm \mathbb{R}=\bigcap_{I\in \mathbb{S}}(\mathbb{R}+\mathbb{R} I),
\]
where, for any $J\in \mathbb{S}$, the ``slice'' $\mathbb{C}_J:=\mathbb{R}+\mathbb{R} J$ can be
identified with the complex plane $\mathbb{C}$. { Hence, any $q\in \mathbb{H}$ can be
written as $q=x+yJ$ with $x,y \in \mathbb{R}$ and $J \in
\mathbb{S}$. The {\it real part }of $q$ is ${\rm Re}(q)=x$ and its
{\it imaginary part} is ${\rm Im}(q)=yJ$; the {\it conjugate}
of $q$ is $\bar q:={\rm Re}(q)-{\rm Im}(q)$. For any non-real
quaternion $a\in \mathbb{H}\setminus \mathbb{R}$ we will
denote by $J_a:=\frac{{\rm Im}(a)}{|{\rm Im}(a)|}\in
\mathbb{S}$ and by $\mathbb{S}_a:=\{{\rm Re}(a)+J |{\rm
Im}(a)| \ : \ J\in \mathbb{S}\}$. If $a\in \mathbb{R}$,
then $J_a$ is any imaginary unit.}
Similarly, if $\textbf{\em a}=(a_1,\ldots,a_n)\in \mathbb H^n$, then the
{\em spherical set} \[\mathbb S_{(a_1,\ldots,a_n)}=\mathbb S_{\textbf{\em a}}
:=\{(g^{-1}a_1g, \ldots, g^{-1}a_ng) : g \in \mathbb{H}\setminus\{0\} \}\]
%by $\mathbb S_{(a_1,\ldots,a_n)}
is the set obtained by simultaneously rotating each coordinate of the point
$\textbf{\em a}=(a_1,\ldots,a_n)$. If there exists $J\in \mathbb S$ such that $\textbf{\em a}=(a_1,\ldots,a_n) \in \mathbb C_J^n$, then $\mathbb S_{\textbf{\em a}}$ is called {\em arranged spherical set}.
Slice regular polynomials in $n$ quaternionic variables are polynomial functions
$P:{\mathbb{H}^n} \to \mathbb{H}$, of the form
\[ (q_1,\ldots, q_n)\mapsto P(q_1,\ldots, q_n)=
\sum_{ \substack{\ell_1=0,\ldots, L_1 \\
\cdots\\ \ell_n=0,\ldots, L_n} }{q_1}^{\ell_1}\cdots {q_n}^{\ell_n}a_{\ell_1,\ldots,\ell_n} \]
with $a_{\ell_1,\ldots,\ell_n}\in\mathbb{H}$,
where $\deg_{{q_{\ell_j}}}P:=L_j$.
Slice regular polynomial functions of several variables can be endowed with an
appropriate notion of product, the so called {\em slice product}, that
will be denoted by the symbol $*$.
%used in the one-variable
%case. %We refer to \cite{severalvariables} for a precise definition,
Let us recall here how it works for
% monomials, and use the fact that we can extend it by linearity to define it for polynomials and
slice regular polynomials in two variables.
\begin{definition}
If $P(q_1,q_2)=\sum_{ \substack{n=0,\ldots,N_1 \\ m=0,\ldots,N_2} } q_1^nq_2^m a_{n,m}$ and
$Q(q)=\sum_{ \substack{n=0,\ldots,L_1 \\ m=0,\ldots,L_2} }q_1^nq_2^m b_{n,m}$ are two
slice regular polynomials,
then the {\em $*$-product} of $P$ and $Q$ is the slice regular polynomial
defined by
$$
P*Q (q_1,q_2):=\sum_{ \substack{n=0,\ldots,N_1+L_1 \\ m=0,\ldots,N_2+L_2}}q_1^nq_2^m\sum_{ \substack{r=0,\ldots,n \\ s=0,\ldots,m} }a_{r,s} b_{n-r,m-s}
$$
\end{definition}
%\begin{itemize}
% \item $q_1*q_2=q_2*q_1=q_1q_2$;
% \item $a*(q_1q_2)=(q_1q_2)*a=q_1q_2a$;
% \item $(q_1^nq_2^ma)*(q_1^rq_2^sb)=q_1^{n+r}q_2^{m+s}ab$.
%\end{itemize}
%\noindent Moreover we point out that, if $P$ or $Q$ have real coefficients, then $P*Q=Q*P$.
%
%\noindent For slice regular polynomials of one variable, the relation of the $*$-product with the usual pointwise product is the following (see \cite[Theorem 3.4]{libroGSS}):
% \begin{equatio\noindent For example, if $a,b \in \mathbb{H}$, thenn}\label{PQ}
% P* Q(q)=\left\{\begin{array}{c cl}
% 0 & &\text{if $P(q)=0 $}\\
% P(q)\cdot Q(P(q)^{-1}\cdot q \cdot P(q)) & &\text{if $P(q)\neq 0 $\,,}
% \end{array}\right.
% \end{equation}
% for any $P,Q$ slice regular polynomials.
% \noindent
% Notice that $P(q)^{-1}\cdot q\cdot P(q)$ belongs to the %same
% sphere $\mathbb{S}_q$. Hence each zero of $P*Q$ in
% $\mathbb{S}_q$ is given either by a zero of $P$ or by a point
% which is a conjugate of a zero of $Q$ in the same sphere.
%
%
%%\begin{example}\label{valutazione}
%\noindent Observe that
%(\ref{PQ}) shows that
%%despite the multiplicative structure given by the $*$ product in $\mathbb{H}[q_1,q_2]$ resembles the usual one,
%the point-wise evaluation of slice regular polynomials %power series
%is not a multiplicative homomorphism.
%Moreover, the vanishing locus of the $*$-product of two slice regular polynomials is not in general the union of the zeros of each of the factors.
%
%
%\noindent For slice regular polynomials in several variables, the situation is even more complicated; in general, the zeros of $P$ are not necessarily zeros of $P*Q$. For instance, while the polynomial $q_1-i$ vanishes on $\{i\}\times \mathbb{H}$,
% the polynomial $q_1q_2-q_1j-q_2i+k$ obtained by $*$--multiplication of $(q_1-i)$ times $(q_2-j)$, vanishes on the pair $(i, q_2)$ if and only if $q_2\in\mathbb{C}_i$.
%
%
%
%
%
%
%The set $\mathbb{H}[q_1,\ldots,q_n]$ of slice regular polynomials in $n$ quaternionic variables
%can be endowed with an
%appropriate notion of (non--commutative) product, the so called {\em slice product}, denoted by the symbol $*$;
\noindent Notice that the $*$ product is not commutative in general.
In this way, $(\mathbb{H}[q_1,\ldots,q_n], +,*)$ has the structure of a non-commutative ring.
\noindent For points with commuting components, the $*$-product {has the following explicit expression.}
%can be expressed in terms of the point-wise product as follows.
\begin{proposition}[{\cite[Proposition 3.1]{Nul2}}]\label{prodstar}
Let ${ \textbf{a}}=(a_1,\ldots,a_n)\in \mathbb H^n$ be
such that
$a_la_m=a_ma_l$ for any $1\leq l,m\leq n$
and let $P,Q \in \mathbb H[q_1,\ldots,q_n]$. Then
\[P*Q(\textbf{a})=\left\{\begin{array}{lr}
0 & \text {if}\ P(\textbf{a})=0\\
P(\textbf{a})\cdot Q(P(\textbf{a})^{-1}a_1P(\textbf{a}), P(\textbf{a})^{-1}a_2P(\textbf{a}),\ldots,P(\textbf{a})^{-1}a_nP(\textbf{a})) & \text{if}\ P(\textbf{a})\neq0
\end{array}\right.\]
\end{proposition}
{
\noindent As in the one-variable case, it is possible to introduce two
operators on slice regular functions. For the sake of
simplicity, we define them only for polynomials of two
quaternionic variables.
% $\mathbb{H}[q_1,q_2]$:
\begin{definition}\label{R-coniugata2}
Let $P(q_1,q_2)=
\sum_{ \substack{n=0,\ldots,N_1 \\ m=0,\ldots,N_2} }
q_1^nq_2^ma_{n,m}$ be
a slice regular polynomial. Then the \emph{regular conjugate} of $P$ is the
slice regular polynomial defined by
\[P^c(q_1,q_2)=\sum_{ \substack{n=0,\ldots,N_1 \\ m=0,\ldots,N_2} }
q_1^nq_2^m\overline{a_{n,m}},\]
and the
\emph{symmetrization} of $P$ is the slice regular polynomial defined by
\[P^s= P*P^c = P^c*P.\]
%Moreover the regular reciprocal $f^{-*}$ of $f$ is defined as $f^{-*}=f^{-s}f^c$ on $\Omega\setminus Z_f$
\end{definition}
}
%
% Let us now recall the
%definition of slice regular functions of a quaternionic variable. In
%view of the setting we will be working on, we will restrict to the
%simplest case possible, that is to slice regular functions defined on
%open balls centered at the origin, $B_R:=\{q \in \mathbb{H} \ :
%\ |q|<R \}$, with radius $R>0$, possibly infinite.
%
%
%%For the sake of simplicity we will restrict our attention to a
%%particular class of domains, where the preliminary results are more
%%easy to state.
%For all details and a more general version of this theory
%%on
%%axially symmetric domains $\Omega$,
%we refer the reader to
%\cite{libroGSS}.
%A domain $\Omega\subseteq \mathbb{H}$ is called {\em
% axially symmetric} if it intersects the real axis and, for any $I\in
%\mathbb S$, $\Omega_I:=\Omega\cap \mathbb C_I$ is a symmetric domain
%with respect to the real axis.
%{\color{red} Spostare nei prerequisiti}
{
%In analogy with the Splitting Lemma for slice regular functions in one quaternionic variable (see \cite[Lemma 1.3]{libroGSS}), for slice regular polynomial functions in several quaternionc variables
\noindent We
recall the following version of Splitting Lemma which is the several variable counterpart of Lemma 1.3 in \cite{libroGSS}.
\begin{lemma}\label{splitting}%\marginpar{ $K\perp L$??}
Let $P$
be a slice regular polynomial in $n$ quaternionic variables. For any $K\in \mathbb S$ and for any $L\in\mathbb S$ orthogonal to $K$ (with respect to the standard scalar product in $\mathbb{R}^3$), there exist two complex polynomials $F,G:\mathbb C_K^{n}\to \mathbb C_K^{n} $ such that for any $(z_1,\ldots,z_n)\in \mathbb C_K^n$
\begin{equation}\label{split}
P(z_1,\ldots,z_n)=F(z_1,\ldots,z_n)+G(z_1,\ldots,z_n)L
\end{equation}
and
\begin{equation}\label{splitconj}
P^c(z_1,\ldots,z_n)=\overline{F(\bar z_1,\ldots,\bar z_n)}-G(z_1,\ldots,z_n)L.
\end{equation}
\end{lemma}
%
% \begin{lemma}\label{splittingconj} %\marginpar{ $K\perp L$??}
% Let $P \in \mathbb H[q_1,\ldots,q_n]$
% be a slice regular polynomial, and let $K,L\in \mathbb S$ be two orthogonal imaginary units. If the splitting of $P$ on $\mathbb C_K^n$ with respect to $L$ is
%$P(z_1,\ldots,z_n)=F(z_1,\ldots,z_n)+G(z_1,\ldots,z_n)L$,
%where $(z_1,\ldots,z_n)\in \mathbb C_K^n$ and $F,G:\mathbb C_K^{n}\to \mathbb C_K^{n} $ are complex polynomials, then the splitting of $P^c$ on $\mathbb C^n_K$ with respect to $L$ is
% \begin{equation}\label{splitconj}
% P^c(z_1,\ldots,z_n)=\overline{F(\bar z_1,\ldots,\bar z_n)}-G(z_1,\ldots,z_n)L.
% \end{equation}
%
%\end{lemma}
\begin{proof}
Equation \eqref{split} is the statement of Lemma 2.1 in \cite{Nul2}.
% [{\cite[Lemma 2.1]{Nul2}}]
\noindent As a consequence, if $P(q_1,\ldots,q_n)=\sum_{\ell_1,\ldots,\ell_n}q_1^{\ell_1}\cdots q_n^{\ell_n}a_{\ell_1,\ldots,\ell_n}$,
since $\{1, K, L, KL\}$ is an orthonormal basis of $\mathbb H$ over $\mathbb R$,
we can write %any coefficient of $P$ as
\[a_{\ell_1,\ldots,\ell_n}=\alpha_{\ell_1,\ldots,\ell_n}+\beta_{\ell_1,\ldots,\ell_n}L\]
with $\alpha_{\ell_1,\ldots,\ell_n},\beta_{\ell_1,\ldots,\ell_n} \in \mathbb C_K$, so that the splitting components of $P$ are
\[F(z_1,\ldots,z_n)=\sum_{\ell_1,\ldots,\ell_n}z_1^{\ell_1}\cdots z_n^{\ell_n}\alpha_{\ell_1,\ldots,\ell_n}, \ \ \text{ and } \ \
G(z_1,\ldots, z_n)=\sum_{\ell_1,\ldots,\ell_n}z_1^{\ell_1}\cdots z_n^{\ell_n}\beta_{\ell_1,\ldots,\ell_n}.\]
Thanks to the fact that for any quaternions $u,v$ we have $\overline{uv}=\bar v \bar u$ and to the fact that orthogonal imaginary units anticommute, we have
\[\overline{a_{\ell_1,\ldots,\ell_n}}=%\overline{\alpha_{\ell_1,\ldots,\ell_n}+\beta_{\ell_1,\ldots,\ell_n}L}=
\overline{\alpha_{\ell_1,\ldots,\ell_n}} + \overline{\beta_{\ell_1,\ldots,\ell_n}L}= \overline{\alpha_{\ell_1,\ldots,\ell_n}} -L\overline{\beta_{\ell_1,\ldots,\ell_n}}= \overline{\alpha_{\ell_1,\ldots,\ell_n}} -\beta_{\ell_1,\ldots,\ell_n} L.\]
Hence
\[P^c(z_1,\ldots,z_n)=
%\sum_{\ell_1,\ldots,\ell_n}z_1^{\ell_1}\cdots z_n^{\ell_n}\bar a_{\ell_1,\ldots,\ell_n}=
\sum_{\ell_1,\ldots,\ell_n}z_1^{\ell_1}\cdots z_n^{\ell_n}(\overline{\alpha_{\ell_1,\ldots,\ell_n}} -\beta_{\ell_1,\ldots,\ell_n} L)=\overline{F(\bar z_1,\ldots,\bar z_n)}-G(z_1,\ldots,z_n)L.\]
\end{proof}
}
\noindent {In what follows, we will mainly consider {\em right ideals} of $\mathbb{H}[q_1,\ldots,q_n]$. In particular, we will use the notation
$\langle P : \ P \in \mathcal{E}\rangle $
for the {\em right} ideal generated in $(\mathbb H[q_1,\ldots, q_n], +, *)$ by the slice regular polynomials belonging to the subset $ \mathcal E \subset \mathbb H[q_1,\ldots, q_n]$.}
%{\color{blue} and, for the sake of brevity, we will often refer to them simply as {\em ideals}, unless explicitly stated. }
\begin{definition}
\noindent Given a right ideal $I$ in $\mathbb{H}[q_1,\ldots,q_n]$,
we define $\mathcal{V}(I)$ to be the set of common zeros of $P\in I$, i.e.,
if $Z_P\subset \mathbb{H}^n$ denotes the zero set of a slice regular polynomial $P\in I$, then
\[ \mathcal{V}(I):=\bigcap_{P\in I} Z_P.\]
\noindent Furthermore, we set
\[ \mathcal{V}_c(I):=\mathcal{V}(I)\cap\bigcup_{K\in \mathbb{S}}\mathbb{C}_K^n.\]
\end{definition}
%% {\color{blue} Cancellerei questa frase: \noindent The notation is in accordance with the one in \cite{israeliani-1}, where $\mathbb{H}^n_c$ denotes the set of quaternionic $n$-tuples with commuting components, so that
%% $\mathcal{V}_c(I)=\mathcal{V}(I)\cap \mathbb{H}^n_c$.}
%%
\noindent Notice that for any right ideal $I$ the set $\mathcal V_c(I)$ is contained in $\mathcal{V}(I)$.
\begin{definition}\label{idealinuovi}
Let $Z$ be a non-empty subset of $\mathbb{H}^n$.
\noindent We denote by
$\mathcal{J}(Z)$ the right ideal generated in $\mathbb{H}[q_1,\ldots,q_n]$
by slice regular polynomials which vanish on $Z$,
\[\mathcal{J}(Z):=\left\{\sum_{k=1}^N P_k*Q_k\ :\ P_k,Q_k\in\mathbb{H}[q_1,\ldots,q_n]\ {\rm with}\ {P_k}_{|_Z}\equiv 0\right\}.\]
\end{definition}
\noindent In general
$\mathcal{J}(Z)$ does not coincide with the set of polynomials vanishing on $Z$, but if $Z=\mathcal{ V }_c(I)$ where $I$ is a right ideal of $\mathbb{H}[q_1,\ldots,q_n]$, then it can be proved directly, using Proposition \ref{prodstar}, that
\begin{equation}\label{jvci}
\mathcal J(\mathcal V_c(I))=\{P\in \mathbb H[q_1,\ldots,q_n] : \ P_{|_{\mathcal V_c(I)}} = 0 \}.
\end{equation}
\noindent We can say also that, if $Z_c$ is any subset of $\mathbb H^n$ of points with commuting components $$
\mathcal J(Z_c)=\{P\in \mathbb H[q_1,\ldots,q_n] : \ P_{|_{Z_c}} = 0 \}.$$
\noindent Furthermore, notice that if $Z'\subseteq Z$ then $\mathcal J(Z')\supseteq \mathcal J (Z)$.
\noindent Combining Theorem 4.1 and Corollary 4.2 in \cite{Nul2} we have
\begin{proposition} \label{vanishing}
Let $I$ be a right ideal of $\mathbb H [q_1,\ldots,q_n]$. Then
\begin{enumerate}
\item $\mathcal{J}(\mathcal{V}_c(I))=\mathcal{J}(\mathcal{V}(I))$;
\item $\mathcal{J}(\mathcal{V}(I))$ coincides with the set of polynomials vanishing on $\mathcal V(I)$.
\end{enumerate}
\end{proposition}
\noindent Therefore it is not difficult to show that, if $I_1$ and $I_2$ are two ideals in $\mathbb H[q_1,\ldots,q_n]$,
\begin{equation}\label{Jcap1}
\mathcal{J}(\mathcal{V}(I_1)\cup \mathcal{V}(I_2))=\mathcal{J}(\mathcal{V}(I_1))\cap \mathcal{J}(\mathcal{V}(I_2)),
\end{equation}
and
\begin{equation}\label{Jcap}
\mathcal{J}(\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2))=\mathcal{J}(\mathcal{V}_c(I_1))\cap \mathcal{J}(\mathcal{V}_c(I_2)).
\end{equation}
\noindent To state the Strong Nullstellensatz for slice regular polynomials we need the following generalized notion of {\em prime} ideal in the non-commutative setting (see \cite{reyes}).
\begin{definition}\label{cprime} A right ideal $I$ in $\mathbb{H}[q_1,\ldots,q_n]$ is said to be {\em completely prime} if and only if given two polynomials $P,Q$ such that
$P*Q\in I$
and $P*I\subseteq I$, %(forse non serve questa richiesta)
then either $P\in I$ or $Q\in I$.
\end{definition}
\begin{theorem}[{Strong Nullstellensatz}]\label{snull}
Let $I$ be a right ideal of $\mathbb{H}[q_1,\ldots,q_n]$. Then
\[\mathcal{J}(\mathcal{V}(I))=\sqrt I:= \bigcap_{ I \subset J\\ \; {J\mathrm{ completely\ prime}}}J\]
\end{theorem}
%\noindent As a consequence of the Strong Nullstellensatz, we have that equality \eqref{jvci} holds also for points with non-commuting components.
%\begin{corollary}\label{vanishing}
% Let $I$ be a right ideal in $\mathbb H [q_1,\ldots,q_n]$. Then
%$\mathcal{J}(\mathcal{V}(I))$ coincides with the set of polynomials vanishing on $\mathcal V(I)$.
%\end{corollary}
%more in general,
% \ \text{and} \ \mathcal{J}(\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2))=\mathcal{J}(\mathcal{V}_c(I_1))\cap \mathcal{J}(\mathcal{V}_c(I_2))
%\noindent Notice that if $\mathcal{V}(I)\subseteq \mathcal{V}(I')$
%\[\boxed{\mathcal{J}(\mathcal{V}(I'))\subseteq \mathcal{J}(\mathcal{V}(I))}\]
%\end{iteze}
%if $I\subseteq\mathbb{H}[q_1,\ldots,q_n]$ is an ideal,
%\[\boxed{\mathcal{V}_c(I):=\mathcal{V}(I)\bigcap_{J\in\mathbb{S}}( \mathbb{C}_J)^n}\]
\begin{definition} An ideal $I$ in $\mathbb{H}[q_1,\ldots,q_n]$ is said to be {\em radical} if and only if $\sqrt I=I$.
\end{definition}
\noindent Notice that $ I$ is contained in $\sqrt{I}$.
\begin{remark} By definition, any completely prime ideal $I$ in $\mathbb{H}[q_1,\ldots,q_n]$ is a radical ideal. The converse is not true: the ideal $I$ in $\mathbb{H}[q]$ generated by the polynomial $q^2+1$ is radical. Indeed $$\sqrt{I}=\mathcal{J}(\mathcal{V}(I))=\mathcal{J}(\mathbb{S}_i)=I,$$
but it is not completely prime, thanks to \cite[Lemma 4.7]{israeliani-1}.
\end{remark}
% \\
% \fbox{$I$ radical $\centernot\implies$ $I$ completely prime (see Ex. \ref{radnoprime})}
%
%\end{center}
\begin{proposition}\label{vrad} We have
\[{\mathcal{V}_c(I)=\mathcal{V}_c(\sqrt{I}) }\] and
\[{\mathcal{V}(I)=\mathcal{V}(\sqrt{I}) }\]
\end{proposition}
\begin{proof}
Since $\sqrt{I}\supseteq I$, then $\mathcal V_c(\sqrt{I})\subseteq \mathcal V_c({I})$ and $\mathcal V(\sqrt{I})\subseteq \mathcal V({I})$.
The other inclusion follows from the fact that $\mathcal J(\mathcal V(I))$ coincides with the set of polynomials vanishing on $\mathcal V(I)$, and the same holds for $\mathcal V_c(I)$ (see Proposition \ref{vanishing}).
\end{proof}
\noindent Besides its algebraic meaning, the Strong Nullstellesantz is relevant also from a geometric viewpoint. In fact it allows to introduce a notion of algebraic set in this quaternionic setting which gives rise to a Zariski type topology, see \cite{Nul2}.
\begin{definition} \label{slicealg}
A subset $V\subseteq \mathbb H^n$ is called {\em slice algebraic} if for any $K\in \mathbb S$, $V\cap \mathbb C_K^n$ is a complex algebraic subset of $\mathbb C_K^n$.
\end{definition}
\noindent In particular, one can prove (see \cite[Theorem 4.6]{Nul2}) that if $I$ is a right ideal of $\mathbb H [q_1,\ldots,q_n]$, then
$\mathcal V(I)$ is a slice algebraic set in $ \mathbb H^n$.
Moreover, since for any $K\in \mathbb S$
\[\mathcal V_c(I) \cap \mathbb C_K^n=\mathcal V(I) \cap \mathbb C_K^n,\]
Definition \ref{slicealg} implies that {any set of the form} $\mathcal V_c(I)$ is slice algebraic as well.
\section{Symmetrization of algebraic sets and of ideals}
\noindent To begin with, we give the following
%\begin{definition} %(DA DEFINIRE SIMMETRIZZATO ANCHE NON DI INSIEME CON PUNTI CENTRALI?)S
% Starting from a set $\mathcal{V}_c(I)$, we denote its {\em symmetrization} by $$\mathbb{S}_{\mathcal{V}_c(I)}:=\bigcup_{\textbf{a} \in \mathcal{V}_c(I)}\mathbb S_{\textbf a}.$$\end{definition}
{ \begin{definition} %(DA DEFINIRE SIMMETRIZZATO ANCHE NON DI INSIEME CON PUNTI CENTRALI?)S
Starting from a set $Z_c\subseteq \mathbb H^n$ containing only points with commuting components, we denote its {\em symmetrization} by $$\mathbb{S}_{Z_c}:=\bigcup_{\textbf{a} \in Z_c}\mathbb S_{\textbf a}.$$\end{definition}
}
\noindent {Note that, if all the points of $Z_c$ have commuting components, then the same holds true for all the points of $\mathbb{S}_{Z_c}$.}
\noindent Let us recall the following propositions which enlighten the analogies between spheres in $\mathbb H$ and spherical sets in $\mathbb H^n$ as vanishing sets for slice regular polynomials in one and several quaternionic variables.
\begin{proposition}[{{\cite[Proposition 3.12]{Nul2}}}] \label{zerisferici}
If a slice regular polynomial with real coefficients $R\in
\mathbb R[q_1,\ldots,q_n]$ vanishes at a point
$\textbf{a}\in \mathbb H^n$, then it vanishes on the
entire spherical set $\mathbb S_{\textbf{a}}$.
\end{proposition}
\begin{proposition}[{\cite[Proposition 3.18]{Nul2}}]\label{zericinesi}
Let $P\in \mathbb{H}[q_1,\ldots,q_n]$ and let ${\textbf{a}}\in \mathbb{C}_J^n$ for some $J\in \mathbb{S}$.
If $P$ vanishes on two different points of the arranged spherical set $\mathbb S_{\textbf{a}}$, then $P$ vanishes on the entire $\mathbb S_{\textbf{a}}$.
\end{proposition}
\noindent Thanks to the previous propositions, we can prove the following result, which generalizes to the several variable case Proposition 3.10 in \cite{libroGSS}.
\begin{proposition}\label{corrsimm}
Let $\textbf{a}\in \mathbb C^n_J$. Then the zeros of $P$ on $\mathbb{S}_{\textbf{a}}$ are in one-to-one correspondence with those of $P^c$. \\
Moreover $P$ has a zero on $\mathbb{S}_{\textbf{a}}$ if and only if $P^s$ has a zero on $\mathbb{S}_{\textbf{a}}$ (and hence vanishes identically on it).
\end{proposition}
\begin{proof}
Let $\textbf{\em b}\in \mathbb S_{\textbf{\em a}}$ be such that $P(\textbf{\em b})=0$. Then $P^s=P*P^c$ vanishes on $\mathbb S_{\textbf{\em a}}$, and in particular $P^s(\overline{ \textbf{\em b}})=0$.
Suppose first that $P(\overline{\textbf{\em b}})\neq 0$, { so that the only zero of $P$ on $\mathbb S_{\textbf{\em a}}$ is $\textbf{\em b}$.} Then
\[0=P^s(\overline{ \textbf{\em b}})=P(\overline{ \textbf{\em b}})\cdot P^c(P(\overline{ \textbf{\em b}})^{-1}\overline{\textbf{\em b}}P(\overline{ \textbf{\em b}})),\]
and thus $P^c$ vanishes on $P(\overline{ \textbf{\em b}})^{-1}\overline{\textbf{\em b}}P(\overline{ \textbf{\em b}}) \in \mathbb S_{\textbf{\em a}}$.
If instead $P(\overline{\textbf{\em b}})= 0$, then, $P$ vanishes on the entire $\mathbb S_{\textbf{\em a}}$.
Let $K,L$ be two orthogonal imaginary units and consider the splitting $P=F+GL$. Since $P$ vanishes on $\mathbb S_{\textbf{\em a}}\cap \mathbb C_{K}^n=\{\textbf{\em c},\overline{\textbf {\em c}}\}$, we get that $F(\textbf{\em c})=F(\overline{\textbf{\em c}})=0$ and $G(\textbf{\em c})=G(\overline{\textbf{\em c}})=0$.
Using the splitting formula for $P^c$, we get that $P^c$ vanishes both at ${\textbf{\em c}}$ and at $\overline{\textbf{\em c}}$, and hence on the entire $\mathbb S_{\textbf{\em a}}$. { Since the regular conjugation is an involution in $\mathbb H[q_1,\ldots,q_n]$ (see Definition \ref{R-coniugata2}), we get the claim. }
\end{proof}
\noindent Let us now investigate geometric and algebraic properties of the symmetrization of a slice algebraic set of the form $\mathcal V_c(I)$. \\
\noindent We need the following definition.
\begin{definition}
Given a right ideal $I\subseteq \mathbb H[q_1,\ldots,q_n]$, we can define the {\em symmetrized ideal} $\mathcal{S}(I)$ as the ideal generated by $P^s$ with $ P\in I$, i.e.
$\mathcal{S}(I)=\langle P^s\;|\;P\in I\rangle$.
\end{definition}
\noindent Notice that $ \mathcal{S}(I)$ is contained in $I$. Moreover, since $\mathbb H[q_1,\ldots,q_n]$ is a Noetherian ring (see, e.g., \cite{reyes}), the symmetrized ideal $\mathcal S(I)$ can be generated by a finite number of symmetrized polynomials.
\begin{theorem}\label{simmalg}
Let $I$ be a right ideal of $\mathbb H[q_1,\ldots,q_n]$. Then the symmetrization of $\mathcal V_c(I)$ is a slice algebraic set, more precisely
\begin{equation}\label{inclusion}
\mathbb{S}_{\mathcal{V}_c(I)}= \mathcal{V}_c( \mathcal{S}(I)).
%:=\bigcup_{\mathbf{a} \in \mathcal{V}_c(J)}\mathbb S_{\mathbf a}
%=\bigcup_{\mathbf{a}\in\bigcap\limits_{K\in J}\mathcal{V}_c(\langle K\rangle)} \mathbb S_{\mathbf a}
\end{equation}
\end{theorem}
\begin{proof} The inclusion $\mathbb{S}_{\mathcal{V}_c(I)}\subseteq\mathcal{V}_c( \mathcal{S}(I))$ can be proved easily. Indeed
every polynomial $P$ in $I$ vanishes on $\mathcal{V}_c(I)$, and thus $P^s$ vanishes on $\mathbb{S}_{\mathcal{V}_c(I)}$. \\
For the other inclusion, assume by contradiction that there exists $\textbf{\em a}\in \mathcal{V}_c(\mathcal{S}(I))\setminus \mathbb{S}_{\mathcal{V}_c(I)}$.
%For all $P\in I$ its symmetrisation $P^s$ vanishes on $\mathbb{S}_{\textbf{\em a}}$.
%Moreover for all $g\neq 0$ there exists a polynomial $Q_g\in I$ such that $Q_g(g^{-1}ag)\neq 0$.
Since $\textbf{\em a}\notin \mathbb{S}_{\mathcal{V}_c(I)}$, we get that $\textbf{\em a}\notin \mathbb{R}^n$ and $\mathbb S_{\textbf{\em a}} \cap \mathcal{V}_c(I)=\varnothing$. Hence there exist at least two polynomials $P_1$ and $P_2$ in $I$ without a common zero on $\mathbb{S}_{\textbf{\em a}}$. Call $\textbf{\em a}_1$ and $\textbf{\em a}_2$
the two points on $\mathbb{S}_{\textbf{\em a}}$ on which $P_1$ and $P_2$ vanish respectively. Note that $\textbf{\em a}_1\neq \textbf{\em a}_2.$ Observe that for all $\alpha,\beta\in \mathbb{H},$ $P_1*\alpha+P_2*\beta\in I$. Thus $F_{\alpha,\beta}=(P_1*\alpha+P_2*\beta)^s\in S(I)$ vanishes on
$\mathbb{S}_{\textbf{\em a}}$. We evaluate $F_{\alpha,\beta}=P_1^s|\alpha|^2+P_2^s|\beta|^2+P_1*\alpha*\overline{\beta}*P_2^c+P_2*\beta*\overline{\alpha}*P_1^c$ on $\textbf{\em a}_1$.
{If $\alpha,\beta\in \mathbb{H}\setminus\{0\}$}, using the notation of Proposition \ref{prodstar},
we get $$0=F_{\alpha,\beta}(\textbf{\em a}_1)=[P_2{ { *}}\beta*\overline{\alpha}*P_1^c] (\textbf{\em a}_1)=P_2(\textbf{\em a}_1)\beta\overline{\alpha}\cdot P_1^c({(P_2(\textbf{\em a}_1)\beta\overline{\alpha})}^{-1}\textbf{\em a}_1P_2(\textbf{\em a}_1)\beta\overline{\alpha}),$$
{since $P_2(\textbf{\em a}_1)\beta\overline{\alpha}\neq 0$.}
Denote by $\textbf{\em b}_1$
the unique zero on $\mathbb{S}_{\textbf{\em a}}$ of $ P_1^c$. Thus we have {$\forall \alpha,\beta\in \mathbb{H}\setminus\{0\}$}
$$\textbf{\em b}_1={(P_2(\textbf{\em a}_1)\beta\overline{\alpha})}^{-1}\textbf{\em a}_1P_2(\textbf{\em a}_1)\beta\overline{\alpha}=\overline{\alpha}^{-1}\beta^{-1}P_2(\textbf{\em a}_1)^{-1}\textbf{\em a}_1P_2(\textbf{\em a}_1)\beta\overline{\alpha}.$$
For $\alpha=\beta=1$ we get
$$\textbf{\em b}_1={(P_2(\textbf{\em a}_1))}^{-1}\textbf{\em a}_1P_2(\textbf{\em a}_1),$$
and hence for $\alpha=1, \beta\notin \mathbb{R}$
$${\beta}^{-1} \textbf{\em b}_1 \beta=\textbf{\em b}_1.$$ Thus $\textbf{\em b}_1 \beta= \beta\textbf{\em b}_1$ but this can happen only if $\textbf{\em b}_1\in \mathbb{R}^n$. A contradiction since $\textbf{\em a}\notin\mathbb{R}^n$.
% $$\mathbb{S}_{\mathcal{V}_c(J)} \subseteq \mathcal{V}_c( \mathcal{S}(J)).$$
\end{proof}
%
%{\noindent In the particular case of a principal right ideal, i.e. a right ideal generated by a unique slice regular polynomial, we have the following expression.% the symmetrization of the slice algebraic set associated with it has the following expression.
%
%\begin{corollary}\label{zeriPs} Let $P\in \mathbb{H}[q_1,\ldots,q_n]$. Then $\mathbb{S}_{\mathcal{V}_c(\langle P\rangle )}=\mathcal{V}_c(\langle P^s\rangle).$
%\end{corollary}
%}%\noindent As a consequence, we have that $\mathbb{S}_{\mathcal{V}_c(\langle P\rangle )}$ is a slice algebraic set.
%
\noindent From the algebraic viewpoint we have the following description.
\begin{corollary}\label{essealg}
Let $I$ be an ideal in $\mathbb H[q_1,\ldots,q_n]$. Then
\[\mathcal J (\mathbb S_{\mathcal V_c(I)})=\sqrt{\mathcal S(I)}.\]
Furthermore $\mathcal J (\mathbb S_{\mathcal V_c(I)})$ is a two-sided ideal.
% the symmetrization of $\mathcal V_c(I)$ is a slice algebraic set, more precisely
% \[\mathbb S_{\mathcal V_c(I)}= \mathcal V(\mathcal J (\mathbb S_{\mathcal V_c(I)}))=\mathcal V_c(\mathcal J (\mathbb S_{\mathcal V_c(I)})).\]
% The symmetrisazion $$ of a slice algebraic set of the form $$, for any ideal
\end{corollary}
\begin{proof}
The equality $\mathcal J (\mathbb S_{\mathcal V_c(I)})=\sqrt{\mathcal S(I)}$ follows directly from Theorem \ref{simmalg} and from the Strong Nullstellesatz \ref{snull}.\\
For the second part, let $Q$ be any polynomial in $\mathbb{H}[q_1,\ldots,q_n]$; we have to show that $Q * \mathcal J (\mathbb S_{\mathcal V_c(I)}) \subseteq \mathcal J (\mathbb S_{\mathcal V_c(I)})$. Let $P \in \mathcal J (\mathbb S_{\mathcal V_c(I)})$, and let $\textbf{\em a} \in \mathbb S_{\mathcal V_c(I)}$. Then either $Q(\textbf{\em a})=0$, and hence $Q*P(\textbf{\em a})=0$; or
\[Q*P(\textbf{\em a})=Q(\textbf{\em a})P(Q(\textbf{\em a})^{-1}\textbf{\em a}Q(\textbf{\em a}))=0\]
since $P$ vanishes on $\mathbb S_{\mathcal V_c(I)}$.
Therefore $Q*P \in \mathcal J (\mathbb S_{\mathcal V_c(I)})$.
\end{proof}
\noindent The second part of the statement of the previous corollary can be easily generalized as follows.
\begin{proposition}\label{Zc}
Let $Z_c$ be a subset of $\mathbb H^n$ of points with commuting components. Then $\mathcal J(\mathbb S_{Z_c})$ is a two-sided ideal.
% of $\mathbb H_c^n$
\end{proposition}
%The same does not hold for $\mathbb S_{Z_c}$ where $Z_c$ is any subset of $\mathbb H^n$ of points with commuting components.\\
%Consider for example $Z_c=\cup_{n \in \mathbb{N}} \{i/n\} \cup \{0\}$. Then $\mathcal J (\mathbb S_{Z_c})=\{0_{\mathbb H[q]}\}$. Hence $\mathcal V_c (\mathcal J (\mathbb S_{Z_c}))=\mathbb H \supsetneq \mathbb S_{Z_c}=\cup_{n \in \mathbb{N}} \mathbb S_{i/n} \cup \{0\}$. But note that $Z_c$ is not an algebraic set.
%More in general, we also have that any subset of $\mathbb H^n$ of points with commuting components turns out to be slice algebraic of the form $Z_c=\mathcal V_c(\mathcal {J}(Z_c)).$} NON è vero!!!
%\begin{center} HO COMMENTATO LA PARTE SULLA SIMMETRIZZAZIONE\end{center}
%Given a slice algebraic set $V$ we can define its {\em symmetrisation} generalising the following procedure that applies on a "balloon with a tail".
%If one considers the baloon symmetric set, balloon with a tail,
%\[ \mathbb{S}_{(a_1,\ldots, a_t)}\times\{a_{t+1}\}\times \ldots \times \{a_n\}\]
%with $a_la_m=a_ma_l$ for $l,m>t$, there exists $a\in\mathbb{H}\setminus\{0\}$
%sucht that each $\widetilde{a_j}=a^{-1} a_j a$ for $j=1,\ldots, t$ commutes with any $a_l$ for $l>t$.
%The symmetrization of
%$[\mathbb{S}_{(a_1,\ldots, a_t)}\times\{a_{t+1}\}\times \ldots \times \{a_n\}]=
%\mathbb{S}_{(\widetilde{a_1},\ldots, \widetilde{a_t}, a_{t+1}\ldots a_{n})}.$
%%With this definition we have that $[V]^s=[V_c]^s.$\\
%\begin{remark}Note that $V$ can be NOT CONTAINED in $[V]^s$. Take for example $V={(j,i,i)},$ its symmetrisation can be done taking the shadow of ${(j,i,i)}$ in $V_c$ and then sqymmetrizing, we obtain the spherical set $\mathbb{S}_{(i,i,i)}= [V_c]^s= [V]^s$ which does not contain $V$ {\color{red} In realtà per costruzione ci starebbe anche $\mathbb{S}_{(-i,i,i)}$ }. Note that this set is not of the form $\mathcal{V}(I),$ but also if $V=\mathcal{V}(I)$, we can have $V$ not contained on $[V]^s$. Take $\mathbb{S}_i\times i\times i=V=\mathcal{V}(\langle q_1^2+1,q_2-i,q_3-i\rangle)$ is not contained in $\mathbb{S}_{(i,i,i)}=V^s$.
%
%\end{remark}
%\noindent {\color{blue} Starting from an ideal $I$ we denote with $I^s$ the ideal whose generators are the symmetrisation of the generators of $I$.}
%\begin{lemma} Let $I$ be a right ideal in $\mathbb H[q_1,\ldots,q_n]$. Then
% \begin{equation}\label{inclusion}
% \mathbb{S}_{\mathcal{V}_c(I)}\subseteq \mathcal{V}_c( \mathcal{S}(I))
% %:=\bigcup_{\mathbf{a} \in \mathcal{V}_c(J)}\mathbb S_{\mathbf a}
% %=\bigcup_{\mathbf{a}\in\bigcap\limits_{K\in J}\mathcal{V}_c(\langle K\rangle)} \mathbb S_{\mathbf a}
% \end{equation}
%
% and \begin{equation}\label{unisfere} \mathcal{V}_c( \mathcal{S}(I))=\bigcap_{K\in I}\mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}.
% %{=\bigcap_{K\in J}{\mathcal{V}_c(\langle K^s\rangle)=\bigcap_{K\in J}\bigcup_{\mathbf {a}\in\mathcal{V}_c(\langle K\rangle)} \mathbb{S}_{\mathbf a}
% %=\bigcup_{\mathbf{b}\in\bigcap\limits_{K\in J}\mathcal{V}_c(\langle K^s\rangle)} \mathbb{S}_{\mathbf b}}}
%\end{equation}\end{lemma}
%
%\begin{proof}
% We start from a polynomial $K$ in $I$, $K$ vanishes on $\mathcal{V}_c(I)$, and thus $K^s$ vanishes on $\mathbb{S}_{\mathcal{V}_c(I)}$. Thus the inclusion \eqref{inclusion} holds true.
% % $$\mathbb{S}_{\mathcal{V}_c(J)} \subseteq \mathcal{V}_c( \mathcal{S}(J)).$$
%
% For the second equality, $\textbf{\em a}\in \mathcal{V}_c( \mathcal{S}(I))$ if and only if $K^s(\textbf{\em a})=0$ for any $K\in I$.\\
% Recalling Proposition \ref{corrsimm}, if $\textbf{\em a}\in \mathcal{V}_c( \mathcal{S}(I))$ for any $K\in I$ there exists $\tilde{ \textbf{\em a}} \in \mathbb S_{\textbf{\em a}}$ such that $K(\tilde {\textbf{\em a}})=0$. That is $\mathbb S_{\textbf{\em a}} \subseteq \mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}$, hence $$\mathcal{V}_c( \mathcal{S}(I))\subseteq \bigcap_{K\in I}\mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}.$$
% Conversely if $a\in \bigcap_{K\in I}\mathbb{S}_{\mathcal{V}_c(K)}$ then $\mathbb{S}_{\textbf{\em a}} \subseteq \bigcap_{K\in I}\mathbb{S}_{\mathcal{V}_c(K)}$, and hence $K^s(\mathbb S_{\textbf{\em a}})=0$ for any $K\in I$. Since ${\textbf{\em a}}$ has commuting components, we conclude that
% $\mathbb S_{\textbf{\em a}}\subseteq \mathcal{V}_c( \mathcal{S}(I)).$
%\end{proof}
%
%{\color{red} Lo teniamo?
% \noindent Notice that
%
% \begin{equation}
% \mathbb{S}_{\mathcal{V}_c(I)}
% % :=\bigcup_{\textbf{\em a} \in \mathcal{V}_c(J)}\mathbb S_{\textbf \em a}
% =\bigcup_{\textbf{\em a}\in\bigcap\limits_{K\in I}\mathcal{V}_c(\langle K\rangle)} \mathbb S_{\textbf{\em a}}
% \end{equation}
%
% and
% \begin{equation} \mathcal{V}_c( \mathcal{S}(I))=\bigcap_{K\in I}\mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}
% {=\bigcap_{K\in I}{\mathcal{V}_c(\langle K^s\rangle)=\bigcap_{K\in I}\bigcup_{\textbf {\em a}\in\mathcal{V}_c(\langle K\rangle)}
% \mathbb{S}_{\textbf {\em a}}
% =\bigcup_{\textbf{\em b}\in\bigcap\limits_{K\in I}\mathcal{V}_c(\langle K^s\rangle)} \mathbb{S}_{\textbf {\em b}}}}
% \end{equation}
%}
%\noindent Notice that the inclusion in Formula \eqref{inclusion} is strict in general.
%\begin{example}
% Consider for instance $I=\langle q-i,q-j \rangle$. Then $\mathcal V_c(I)=\varnothing$, $\mathbb S_{\mathcal V_c(I)}=\varnothing$, while $\mathcal V_c(\mathcal S(I))=\mathbb S_i$.
% %{\color{blue}(In this case $\mathbb S_{\mathcal V_c(I)}$ is however slice algebraic.)}
%
%\end{example}
%{\color{blue} NOOO! Se $1\in I$ allora $1 \in \mathcal S(I)$, quindi se $\mathcal V_c(I)=\varnothing$ anche $\mathcal V_c(\mathcal S(I))=\varnothing$ }
%\noindent One can construct also examples in which $\mathcal V_c(I)\neq\varnothing$.
%{\color{blue} Nemmeno quello sotto è un controesempio! Se due polinomi non hanno zeri comuni, si può scrivere 1 come loro combinazione (almeno in una variabile) e quindi $I=<q-2i>$ e $\mathcal V_c(\mathcal S(I))=\mathbb S_{\mathcal V_c(I)}$.
%
%Sembra proprio che $\mathcal V_c(\mathcal S(I))=\mathbb S_{\mathcal V_c(I)}$ sempre... }
%\begin{example}
%
% \noindent Consider the ideal $I=\langle K_1, K_2\rangle$ where
% \[K_1(q)=(q-2i)*(q-k)*(q-j)\quad \text{and} \quad K_2(q)=(q-2i)*(q-i)*(q-j).\]
% \noindent If $q\neq 2i$, setting
% \[P_1(q)=(q-k)*(q-j), \ P_2=(q-i)*(q-j)\ \text{ and } \ T(q)=(q-2i)^{-1}q(q-2i),\] we can write
% \[K_1(q)=(q-2i)P_1(T(q)) \quad \text{and} \quad K_2(q)=(q-2i)P_2(T(q)).\]
% Thanks to the one variable theory (see \cite[Proposition 3.21]{libroGSS}), we have that $P_1$ and $P_2$ do not have common zeros,
% therefore $\mathcal V(I)=\mathcal{V}_c(I)=\{ 2i\}$ and
% \[\varnothing\neq \mathbb{S}_{\mathcal{V}_c((I)}=\mathbb{S}_{2i}\subsetneq \mathcal{V}_c(( \mathcal{S}(I))=\mathbb{S}_{2i}\cup \mathbb{S}_{j}{\color{red}=\mathbb{S}_{\mathcal{V}_c({\langle K_1, K_2\rangle})}\cup \mathbb{S}_{\mathcal{V}_c({\langle K_1^c, K_2^c\rangle})}.} \]
%
%
% {\color{red}
%
%
% Finally, the equality
%
% \[\mathcal{V}_c( \mathcal{S}(I))=\mathbb{S}_{\mathcal{V}_c({\langle K_1, K_2\rangle})}\cup \mathbb{S}_{\mathcal{V}_c({\langle K_1^c, K_2^c\rangle})}; \]
% always holds for ideals with two generators (E' vero??).
%
%
% \noindent In particular, since $\mathbb{S}_{\mathcal{V}_c(I)}=\bigcup_{\textbf{a }\in \mathcal{V}_c(I)} \mathbb{S}_{\textbf a}$, it follows that
% when
% $\bigcap_{\ell=1,2} \mathcal{V}_c(\langle K_\ell\rangle)$
% and $\bigcap_{\ell=1,2} \mathcal{V}_c(\langle K_\ell^c\rangle)$
% are finite sets, then $\mathcal{V}_c( \mathcal{S}(\langle K_1, K_2\rangle))$ is reducible.}
%
%
%\end{example}
\begin{definition}
We say that a slice algebraic set $\mathcal V_c(I)$ has {\em spherical symmetry} if $\mathcal V_c(I)= { \mathcal V_c(\mathcal S(I))=}\mathbb S_{\mathcal V_c(I)}$.
\end{definition}
%\begin{remark}\label{sfere}
% If $\mathcal V_c(I)$ has {\em spherical symmetry} then $\mathcal V_c(\mathcal S (I))=\mathbb S_{\mathcal V_c(I)}$.
% In fact,
% \[\mathcal V_c(\mathcal S (I))=\mathcal V_c(I)\subseteq \mathbb S_{\mathcal V_c(I)} \subseteq \mathcal V_c(\mathcal S(I)) \] where the last inclusion is guaranteed by \eqref{inclusion}.
%
%% \noindent In this case the symmetrization of $\mathcal V_c(I)$ is a slice algebraic set itself.
%\end{remark}
%\noindent The notion of spherical symmetry has a geometric interpretation.
%{\color{red}\begin{proposition}\label{spherical}
%A slice algebraic set $\mathcal V_c(I)$ has spherical symmetry if and only if $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$.% Then $\mathcal V_c(I)$ has spherical symmetry.
%\end{proposition}
%\begin{proof}
%Suppose first that $\mathcal V_c(I)$ has spherical symmetry. Then
%\[\mathcal V_c(\mathcal S (I))=\mathcal V_c(I)\subseteq \mathbb S_{\mathcal V_c(I)} \subseteq \mathcal V_c(\mathcal S(I)) \] where the last inclusion is guaranteed by \eqref{inclusion}.
%
%In the other direction, if $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$, Proposition \ref{zeriPs} implies that the zero set of $K^s$ coincides with the zero set of $K$ for any $K$ in $I$.
%\end{proof}
%\noindent Notice that whenever $\mathcal V_c(I)$ has spherical symmetry, $\mathbb S_{\mathcal V_c(I)}$ is a slice algebraic set.
\noindent As a particular case, we have the following result.
\begin{proposition}\label{twosided}
Let $I$ be a two-sided ideal of $\mathbb H[q_1,\ldots, q_n]$. Then $\mathcal V_c(I)$ has spherical symmetry.
\end{proposition}
\begin{proof}
Let $\textbf{\em a} \in \mathcal V_c(I)$. We want to show that $\mathbb S_{\aaa}\subset \mathcal V_c(I)$. Let $P\in I$, for any $b \in \mathbb H\setminus \{0\}$ we have that $b*P \in I$, hence, using the notation of Proposition \ref{prodstar},
\[bP(b^{-1}{\aaa}b)=0,\]
which implies that $P$ vanishes on the entire sphere $\mathbb S_\aaa$.
\end{proof}
\noindent As a consequence we have
\begin{corollary} If an ideal $I\subseteq \mathbb H[q_1,\ldots,q_n]$ is generated by polynomials with real coefficients, then $\mathcal V_c(I)$ has spherical symmetry. %$\mathcal{V}_c(I)=\mathcal{V}_c( \mathcal{S}(I))$.
\end{corollary}
\noindent Furthermore we have the following equivalence.
\begin{proposition}\label{equivalence}
{The slice algebraic set} $\mathcal V_c(I)$ has spherical symmetry if and only if $\sqrt{I}$ is a two-sided ideal.
\end{proposition}
\begin{proof}
Indeed $\sqrt{I}=\mathcal J(\mathcal V_c(I))=\mathcal J(\mathbb S_{\mathcal V_c(I)})$ which is two-sided, thanks to Proposition \ref{essealg}.
The other direction follows directly from Proposition \ref{twosided}, recalling that $\mathcal V_c(I)=\mathcal V_c(\sqrt{I})$.
\end{proof}
%\begin{proof}
% Proposition \ref{zerisferici} implies that $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$.
%%
%% $\mathcal V_c(\langle K\rangle)=\mathbb S_{\mathcal V_c(\langle K\rangle)}$ whenever $K$ has real coefficients. Corollary \ref{zeriPs} allows us to conclude that $\mathcal V_c(I)=\mathcal V_c(\mathcal S(I))$.
%\end{proof}
%Does the converse hold?
%}
\section{On the correspondence between algebraic sets and ideals}
\noindent Given two { right} ideals $I_1,I_2$ of $\mathbb{H}[q_1,\ldots,q_n]$, we define the $*$-product ideal $I_1*I_2$ to be the { right} ideal generated by the $*$-product of polynomials of
$I_1$ and $I_2$, namely $I_1*I_2:=\langle P*Q,\ P\in I_1, \ Q\in I_2\rangle$.
\noindent We have the following relations concerning algebraic sets associated with $*$-product, intersection and sum of two ideals.
\begin{proposition}\label{corr}
Let $I_1,I_2$ be { right} ideals of $\mathbb H[q_1,\ldots,q_n]$. Then
\begin{enumerate}
\item $\mathcal{V}_c(I_1*I_2)\subseteq \left(\mathcal{V}_c(I_1)\cup\mathcal{V}_c( \mathcal{S}(I_2))\right)$;
\item $\mathcal V_c(I_1 \cap I_2)\supseteq \left(\mathcal{V}_c(I_1)\cup\mathcal{V}_c(I_2) \right)$;
\item $\mathcal{V}_c(I_1+I_2)=\left(\mathcal{V}_c(I_1)\cap\mathcal{V}_c(I_2)\right)$.
\end{enumerate}
\end{proposition}
\begin{proof}
The proof follows taking into account the expression of the $*$-product of polynomials evaluated on points with commuting components given in Proposition \ref{prodstar} and the fact that the operator $\mathcal V$ reverses the inclusions.
\end{proof}
\noindent In the particular case in which one of the two ideals is generated by polynomials with real coefficients we have the following result.
\begin{proposition}
Let $I_1,I_2$ be { right} ideals of $\mathbb H[q_1,\ldots,q_n]$ such that $I_2=\langle G_1,\ldots,G_n\rangle$ with $G_\ell \in \mathbb R[q_1,\ldots,q_n]$ for any $\ell=1,\ldots,n$. Then
\[\mathcal V_c(I_1 \cap I_2)=\mathcal V_c(I_1 * I_2)=\mathcal V_c(I_1)\cup \mathcal V_c(I_2).\]
\end{proposition}
\begin{proof}
The fact that $I_2$ is generated by polynomials with real coefficients, guarantees that $I_1*I_2\subseteq I_1\cap I_2 $, and hence
\[\mathcal V_c(I_1\cap I_2)\subseteq \mathcal V_c(I_1*I_2). \]
Consider now $\textbf{\em a}\in \mathcal V_c(I_1 * I_2)$. Then, for any $F\in I_1, G \in I_2$, $F*G(\textbf{\em a})=0$. If $\textbf{\em a}\not \in \mathcal V_c(I_1)$, then there exists $F_0\in I_1$ such that $F_0(\textbf{\em a})\neq 0$. Then, recalling Proposition \ref{prodstar}, for any $\ell=1,\ldots,n$, we have \[0=F_0(\textbf{\em a})G_\ell(F_0(\textbf{\em a})^{-1}\textbf{\em a}F_0(\textbf{\em a}))=G_\ell(\textbf{\em a})F_0(\textbf{\em a})\]
where the last equality follows from the fact that $G_\ell$ has real coefficients. Hence $G_\ell(\textbf{\em a})=0$ for any $\ell$, that is $\textbf{\em a}\in \mathcal V_c(I_2)$.
Thus \[\mathcal V_c(I_1 * I_2)\subseteq \mathcal V_c(I_1)\cup \mathcal V_c(I_2).\]
To complete the proof, notice that, since $I_1 \cap I_2$ is contained in $I_1$ and in $I_2$, we have
\[\mathcal V_c(I_1)\cup \mathcal V_c( I_2)\subseteq \mathcal V_c(I_1\cap I_2).\]
\end{proof}
\noindent As a particular case, we have
\begin{corollary}\label{essei}
Let $I_1,I_2$ be {right} ideals of $\mathbb H[q_1,\ldots,q_n]$. Then
\[\mathcal{V}_c(I_1* \mathcal{S}(I_2))=\mathcal{V}_c(I_1)\cup\mathcal{V}_c( \mathcal{S}(I_2))=\mathcal{V}_c(I_1\cap \mathcal{S}(I_2)).\]
\end{corollary}
%\begin{proof}
% The proof follows immediatlySince $\mathcal S(I_2)$ is generated by polynomials with real coefficients, we have that $I_1*\mathcal S(I_2) \subseteq I_1 \cap \mathcal S(I_2)$. Combining point (1) and point (2) of Proposition \ref{corr}, we have
% \[\mathcal{V}_c(I_1)\cup\mathcal{V}_c( \mathcal{S}(I_2))\subseteq \mathcal{V}_c(I_1\cap \mathcal{S}(I_2)) \subseteq \mathcal{V}_c(I_1* \mathcal{S}(I_2))\subseteq \mathcal{V}_c(I_1)\cup\mathcal{V}_c( \mathcal{S}(I_2)).\]
%
%\end{proof}
%Therefore it is easy to show that
%\begin{equation}\label{Jcap}
% \mathcal{J}(\mathcal{V}(I_1)\cup \mathcal{V}(I_2))=\mathcal{J}(\mathcal{V}(I_1))\cap \mathcal{J}(\mathcal{V}(I_2)) \ \text{and} \ \mathcal{J}(\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2))=\mathcal{J}(\mathcal{V}_c(I_1))\cap \mathcal{J}(\mathcal{V}_c(I_2))
%\end{equation}
%%more in general,
%
%\noindent Notice that if $\mathcal{V}(I)\subseteq \mathcal{V}(I')$
%\[\boxed{\mathcal{J}(\mathcal{V}(I'))\subseteq \mathcal{J}(\mathcal{V}(I))}\]
%%\end{iteze}
\section{Irreducible slice algebraic sets}\label{S1}
%\begin{itemize}
%\item
%\item
\noindent {Let us begin by giving the definition of reducibility of slice algebraic sets.}
% We begin with this fundamendal
\begin{definition} \label{red}
Given a right ideal $I\subseteq \mathbb{H}[q_1,\ldots,q_n]$, the slice algebraic set $\mathcal{V}(I)$,
is said to be {\em reducible} whenever there exist two proper, non-trivial, right ideals $I_1$ and $I_2$ in $\mathbb{H}[q_1,\ldots,q_n]$, not contained one into another, such that
$$\mathcal{V}(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2).$$
%with $\mathcal{V}(I_1)\not\subset \mathcal{V}(I_2)\neq \mathcal{V}(I)$ and
% $\varnothing\neq\mathcal{V}(I_2)\not\subset \mathcal{V}(I_1)\neq \mathcal{V}(I)$.
Analogously, $\mathcal{V}_c(I)$ is said to be {\em reducible} if there exist two proper, non-trivial, right ideals $I_1$ and $I_2$ not contained one into another, such that
$$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2).$$
\noindent If $\mathcal{V}(I)$ (or $\mathcal{V}_c(I)$) is not reducible, it is called {\em irreducible}.
\end{definition} %
%\noindent Recall that for a balloon symmetric set $V$, the set $\mathcal{J}(V)$ coincides with the set of polynomials vanishing on $V$.
\noindent Thanks to Proposition 3.14 in \cite{Nul2}, we have the following result.
{\begin{proposition}\label{GeneratoriSfere}
Let $\mathbb{S}_{(a_1,\cdots,a_n)}\subset\mathbb{H}^n$ be an arranged spherical set and suppose $a_\ell \not \in \mathbb{R}$ for any $\ell$. Then $\mathbb S_{(a_1,\ldots, a_n)}=\mathcal{V}(I)$ where $I$ is a two-sided radical ideal generated by polynomials with real coefficients.
\end{proposition}
}
%{\color{blue}\begin{proposition}\label{GeneratoriSfere1}
% Let $\mathbb{S}_{(a_1,\cdots,a_n)}\subset\mathbb{H}^n$ be an arranged spherical set and suppose $a_\ell \not \in \mathbb{R}$ for any $\ell$. Then $\mathbb S_{(a_1,\ldots, a_n)}=\mathcal{V}(I)$ where
% \[I=\langle P_1,\ldots, P_{n-1}, (q_n-a_n)^s \rangle \] is the radical ideal generated by the (irreducible) polynomials $P_\ell$
% \[P_{\ell}(q_{\ell},q_{\ell+1})=q_{\ell}+q_{\ell +1} \gamma_{\ell} +\delta_{\ell}\]
% with real coefficients
%% \[\gamma_{\ell}=\pm\dfrac{|{\rm Im} (a_{\ell})|}{|{\rm Im} ( a_{\ell +1})|}, \quad \text{and} \quad \delta_{\ell}=-{\rm Re} (a_{\ell})\mp\dfrac{{\rm Re} (a_{\ell+1})|{\rm Im} (a_{\ell})|}{|{\rm Im} (a_{\ell +1})|}, \]
%% for any $\ell=1,\ldots,n-1$, where the signs are determined by the sign of the product ${\rm Im}(a_{\ell}){\rm Im }(a_{\ell+1})$.
%%
%% -------------------
%%
%% Let $\mathbb{S}_{(a_1,\cdots,a_n)}\subset\mathbb{H}^n$ be an arranged spherical set and suppose $a_\ell \not \in \mathbb{R}$ for any $\ell$. A slice regular polynomial $P\in\mathbb{H}[q_1,\ldots,q_n]$ vanishes on $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}$ if and only if $P$ is of the form
%% \[Q_1*P_1(q_1,\ldots,q_n)+Q_2*P_2(q_2,\ldots,q_n)+\cdots+Q_{n-1}*P_{n-1}(q_{n-1},q_n)+S_{a_n}*P_n(q_n)\]
%% where
%% $Q_\ell$ are the polynomials
%% \begin{equation}\label{Ql}
%% Q_{\ell}(q_{\ell},q_{\ell+1})=q_{\ell}+q_{\ell +1} \gamma_{\ell} +\delta_{\ell}
%% \end{equation}
%% with real coefficients
% \[\gamma_{\ell}= -\dfrac{|{\rm Im} (a_{\ell})|}{|{\rm Im} ( a_{\ell +1})|}, \quad \text{and} \quad \delta_{\ell}=-{\rm Re} (a_{\ell})+\dfrac{{\rm Re} (a_{\ell+1})|{\rm Im} (a_{\ell})|}{|{\rm Im} (a_{\ell +1})|}, \]
% for any $\ell=1,\ldots,n-1$, if ${\rm Im}(a_{\ell}){\rm Im }(a_{\ell+1})>0$;
% while
% \[\gamma_{\ell}=\dfrac{|{\rm Im} (a_{\ell})|}{|{\rm Im} ( a_{\ell +1})|}, \quad \text{and} \quad \delta_{\ell}=-{\rm Re} (a_{\ell})-\dfrac{{\rm Re} (a_{\ell+1})|{\rm Im} (a_{\ell})|}{|{\rm Im} (a_{\ell +1})|}, \]
% for any $\ell=1,\ldots,n-1$, if ${\rm Im}(a_{\ell}){\rm Im }(a_{\ell+1})<0$.
%% where we take the $+$ in $\gamma_\ell$ and the $-$ in $\delta_\ell$ if ${\rm Im}(a_{\ell}){\rm Im }(a_{\ell+1})>0$, and the opposite otherwise.% he signs are determined by the sign of the product .
%
%
% \end{proposition}
%\begin{proof}
% By direct computation, each $P_\ell$ vanishes on $(a_\ell,a_{\ell+1})$ and hence, considered as a polynomial in $\mathbb H[q_1,\ldots,q_n]$, also on $(a_1,\ldots,a_n)$. Since they have real coefficients, they all vanish on $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}$ (see lemma \ref{zerisferici}).
% Consider $Q$ vanishing on ${\mathbb S_{(a_1,a_2,a_3,\ldots, a_n)}}$. Performing subsequent divisions by the monic polynomials $P_\ell$, we can write
% \[Q(q_1,\ldots,q_n)=P_1*Q_1(q_1,\ldots,q_n)+P_2*Q_2(q_2,\ldots,q_n)+\cdots+P_{n-1}*Q_{n-1}(q_{n-1},q_n)+R(q_n).\]
% Since $Q,P_1,\ldots,P_{n-1}$ vanish on $\mathbb S_{(a_1,a_2,a_3,\ldots, a_n)}$ and $R$ depends only on the variable $q_n$, recalling the one-variable theory (see Proposition 3.18 in \cite{libroGSS}), we get that $R(q_n)$ vanishes on $\mathbb S_{a_n}$, and thus is of the form $R(q_n)=(q_n-a_n)^s*R_1(q_n)$.
%\end{proof}\noindent
%Note that the polynomials $P_{\ell}$ are irreducible and the ideal $I$ is radical.
%and $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}$ is irreducible. We consider the example in dimension $2$, the same argument holds in any dimension for arranged spheres.
%Let $V(I)=\mathbb{S}_{(a,b)}$ where $(a,b)$ are such that $ab=ba$, where $$I=\langle P(q_1,q_2),(q_2-b)^{s}\rangle$$ with $P(q_1,q_2)$
%a polynomial of the form $P(q_1,q_2)=q_1+q_2\alpha+\beta$ and $\alpha,\beta\in \mathbb{R}.$
%Note that a similar polynomial vanishes on $\mathbb{S}_{(a,b)}$. (see Proposition \ref{GeneratoriSfere}) Starting from a generic polynomial that vanishes on $\mathbb{S}_{(a,b)}$ and then dividing it by $P$, the rest $R$
%has degree less than $1$ in $q_1,$ that is it depends only on $q_2$, i.e. $R=R(q_2)$. This vanishes on the sphere of $b$ i.e. it is a multiple of $(q_2-b)^s.$\\
%Note that $P(q_1,q_2)$ is {\bf irreducible}.
% We can assume that $\alpha=1$. Suppose that there exist $c,d\in \mathbb{H}$ such that $q_1q_2+q_1\beta+q_2\gamma+\delta=(q_1-c)*(q_2-d)$ we get that $d=-\beta$ and $c=-\gamma$ and thus $c$ and $d$ are both real, this implies that the *-product is pointwise. But $(q_1-c)\cdot(q_2-d)$ cannot vanish on the spherical set, a contradiction. \\
%
\noindent Moreover
%it is impossible to find two ideals $I_1$ and $I_2$ such that $\mathcal{V}(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2)$, that is $V$ is {\em irreducible}.
\begin{proposition}\label{S_a} Any arranged spherical set $\mathbb{S}_{(a_1,\cdots,a_n)}\in\mathbb{H}^n$ { with $a_\ell \not \in \mathbb{R}$ for any $\ell$} is an irreducible slice algebraic set.
\end{proposition}
\begin{proof}
%Let us prove the statement in two variables.
{ Proposition \ref{GeneratoriSfere} yields that arranged spherical sets are always slice algebraic sets of the form $\mathcal V(I)$.} Suppose now, by contradiction, that $\mathbb{S}_{(a_1,\cdots,a_n)}=\mathcal{V}(I_1)\cup\mathcal{V}(I_2)$ is reducible. Then, without loss of generality, there exist at least two points $(a'_1,\ldots, a_n')$ and $(a''_1,\ldots, a_n'')\in \mathcal{V}(I_1)\subset \mathcal{V}(I)$. Since they belong to the same spherical set, the entire sphere through $(a'_1,\ldots, a_n')$ belongs to $\mathcal{V}(I_1)$, that is $\mathcal{V}(I)=\mathbb{S}_{(a'_1,\ldots, a_n')}=\mathbb{S}_{(a''_1,\ldots, a_n'')}\subset \mathcal{V}(I_1)$.
%This procedure can be applied to a spherical set $S_{(a_1,a_2,\cdots,a_n)}\in\mathbb{H}^n$
% taking as generators of $I$ the polynomials with real coefficients introduced in Proposition \ref{GeneratoriSfere} $P_{\ell}(q_{\ell},q_{{\ell}+1})$ (with $\ell:1\cdots ,n-1)$ and $(q_n-a_n)^s$.\\
\end{proof}
%{\color{red} forse si scrive solo per due sfere}
\noindent {Another family of irredudible slice algebraic set is given by the following proposition.}
\begin{proposition}\label{SxS} Any set of the form $\mathbb{S}_{a}\times\mathbb{S}_{b}\subset \mathbb H^2$ with $a,b\notin \mathbb R$
is an irreducible algebraic set.
\end{proposition}
\begin{proof}
Combining Propositions 3.4 and 3.7 in \cite{Nul2}, we have that $\mathbb{S}_{a}\times\mathbb{S}_{b}=\mathcal V(I)$ where $I$ is the (radical) ideal generated by $(q_1-a)^s,(q_2-b)^s$.
Suppose {by contradiction} that $\mathcal V(I)=\mathcal V(I_1)\cup \mathcal V(I_2)$ is reducible. Thanks to Proposition \ref{vrad} we can assume that $I_1$ and $I_2$ are radical ideals. For the points with commuting components we {then} have
\[\mathcal V_c(I)=\mathcal V_c(I_1)\cup \mathcal V_c(I_2)=\mathbb S_{(a,\tilde{b})}\cup \mathbb S_{(\bar{a},\tilde{b})}\]
where $a\tilde{b}=\tilde{b}a$.
Since $\mathbb S_{(a,\tilde{b})}$ and $\mathbb S_{(\bar{a},\tilde{b})}$ are irreducible, up to swapping the subsets, we get that
$\mathcal V_c(I_1)=\mathbb S_{(a,\tilde{b})}$ and $\mathcal V_c(I_2)=\mathbb S_{(\bar{a},\tilde{b})}$. Thus, recalling Proposition \ref{GeneratoriSfere}, $I_1$ and $I_2$ are such that $\mathcal V(I_1)=\mathcal V_c(I_1)$ and $\mathcal V(I_2)=\mathcal V_c(I_2)$.
Therefore% $\mathcal V(I_1)\cup \mathcal V(I_2)=\mathcal V_c(I_1)\cup \mathcal V_c(I_2)$. Hence
\[\mathbb S_a \times \mathbb S_b=\mathcal V(I_1)\cup \mathcal V(I_2)=\mathcal V_c(I_1)\cup \mathcal V_c(I_2)=\mathbb S_{({a},\tilde{b})}\cup \mathbb S_{(\bar{a},\tilde{b})},\]
which is a contradiction.
\end{proof}
\noindent The previous proposition easily generalizes to any subset of the form $\mathbb{S}_{(a_1, \ldots,a_k)}\times \mathbb S_{(b_{k+1},\ldots,b_n)}$ in $\mathbb H^n$.\\
\noindent Let us now give a couple of examples of reducible slice algebraic sets.
\begin{example}\label{red1}
The subset of $\mathbb{H}^2$ given by $V=\{(i,i)\}\cup \{(2j,2j)\}$ is a reducible slice algebraic set.
\noindent
Indeed $V=\mathcal{V}(I)$ where $I=\langle q_1-q_2, (q_1-i)*(q_2-2\tilde{J})\rangle$.
\noindent This can be seen by considering a polynomial $P$ that vanishes on $V$; if we divide $P$ by $-q_2+q_1$ (as a monic polynomial in the variable $q_2$), we obtain a rest $R(q_1)$ which vanishes again on $V$, that is for $q_1=i$ and $q_1=2j$.
Recalling Proposition 3.21 in \cite{libroGSS},
$$R(q_1)=(q_1-i)*(q_1-2\tilde{J})*R_1(q_1)$$ where $\tilde{J}=(2j-i)^{-1}j(2j-i)=\frac{-4i+3j}{5}$.
Thus $P$ can be written as a combination of $q_1-q_2$ and $(q_1-i)*(q_2-2\tilde{
J})$. \\
Therefore we have that $V=\mathcal{V}(I)=\mathcal{V}(I_1) \ \cup \ \mathcal{V}(I_2)$ is reducible, with $I_1=\langle (q_1-q_2), (q_1-i)\rangle$ and $I_2=\langle (q_1-q_2), (q_2-2j)\rangle$.
\end{example}
\begin{example} \label{red2}
Let $I=\langle q_1^2+1,(q_1-q_2)*(q_2-i)\rangle$. Then
$$\mathcal{V}(I)=(\mathbb{S}\times\mathbb{H})\cap (\{q_1=q_2\}\cup (\mathbb{H}\times \{i\}))$$
is reducible since $\mathcal{V}(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2)$ where $I_1=\langle q_1^2+1,q_1-q_2\rangle$ and $I_2=\langle q_1^2+1,q_2-i\rangle $.
Indeed
$\mathcal{V}(I_1) =\mathbb{S}_{(i.i)}=(\mathbb{S}\times\mathbb{H})\cap \{q_1=q_2\}$ and $\mathcal{V}(I_2)=\mathbb{S}_i\times \{i\}=(\mathbb{S}\times\mathbb{H})\cap (\mathbb{H}\times \{i\})$.
%Note that it is different from $\mathbb{S}_i\times \mathbb{S}_i$ (indeed, for example, it does not contain $(j,k)$).
\end{example}
\noindent {Let us now describe the relation between the reducibility of a slice algebraic set and the reducibility of its points with commuting components. %$\mathcal V(I)$ and that of $\mathcal V_c(I)$.
}\begin{proposition}\label{vc-v}
{Let $I$ be a right ideal of $\mathbb H[q_1,\ldots,q_n]$.}
If $\mathcal{V}_c(I)$ is irreducible, then $\mathcal{V}(I)$ is irreducible.
\end{proposition}
\begin{proof}
Suppose by contradiction that $\mathcal V(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2)$ is reducible. Then
$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$, where $\mathcal V_c(I_1)$ and $\mathcal V_c(I_2)$ are not empty thanks to the Weak Nullstellensatz (see \cite[Theorem 1.1]{israeliani-1}). {Since $\mathcal{V}_c(I)$ is irreducible, we can suppose without loss of generality that $\mathcal{V}_c(I_1)\subseteq \mathcal{V}_c(I_2)$.} Recalling the Strong Nullstellensatz \ref{snull}, we then have
\[\sqrt{I_1}=\mathcal J(\mathcal V(I_1))=\mathcal J(\mathcal V_c(I_1)) \supseteq \mathcal J(\mathcal V_c(I_2))=\mathcal J(\mathcal V(I_2))=\sqrt{I_2},\]
which implies that
\[\mathcal V(I_1)=\mathcal V(\sqrt{I_1})\subseteq \mathcal V(\sqrt{I_2})=\mathcal V({I_2}),\]
in contradiction with the reducibility of $\mathcal V(I)$.
\end{proof}
{
\begin{remark}
The converse of the previous proposition is not true in general. Consider for instance $\mathbb{S}_i\times \mathbb{S}_i$, which is an irreducible slice algebraic set of the form $\mathcal{V}(I)$ with $I=\langle(q_1-i)^s,(q_2-i)^s\rangle$, thanks to Proposition \ref{SxS}. In this case %, the points of $\mathcal V(I)$ with commuting components are
$\mathcal{V}_c(I)=\mathbb S_{(i,i)}\cup \mathbb S_{(-i,i)}$, which is a reducible slice algebraic set, thanks to Proposition \ref{S_a}.
\end{remark}}
%\noindent
%
%\begin{corollary}
%Let $J$ be an ideal whose radical $\sqrt {J}$ is irreducible, then $\mathcal V(J)$ is irreducible.
%
%
%\end{corollary}
%\subsection{Main results}
\noindent Irreducible complex algebraic sets are characterized by the fact that their associated ideals are {\em prime}. From the quaternionic Strong Nullstellesatz \ref{snull}, one might think that {\em completely prime} ideals could play the analogue role in our setting.
This is not the case, as shown by the ideal $I$ generated by $q^2+1$ in $\mathbb H[q]$: {as already recalled in the Introduction, $I$ is not completely prime, while its associated slice algebraic set is $\mathbb S_i$ which is irreducible.}
\noindent {Motivated by this gap in the correspondence}, we introduce a new class of ideals, related to the geometry of the zero sets of slice regular polynomials.
\begin{definition}\label{qpideal} A right ideal $I\subseteq \mathbb H[q_1,\ldots,q_n]$ is said to be {\em quasi prime} if and only if given two polynomials $P,Q$ such that
$P*Q\in I$,
%\item $P*I\subseteq I$ (forse non serve questa richiesta)
then either $P\in I$ or $Q^s\in I$.
\end{definition}
\noindent The class of quasi prime ideals contains the one of completely prime ideals.
\begin{proposition} \label{cp-qp}
If $I\subseteq \mathbb H[q_1,\ldots,q_n]$ is a completely prime right ideal, then $I$ is a quasi prime ideal.
\end{proposition}
\begin{proof} Let $P,Q$ be {slice regular polynomials }such that $P*Q\in I$. Then $P*Q*Q^c=P*Q^s=Q^s*P\in I$, moreover $Q^s*I=I*Q^s\subseteq I$. The fact that $I$ is completely prime implies that either $P\in I$ or $Q^s\in I$.
\end{proof}
%{\color{red}\noindent{\bf DOMANDA}dall'ultima formula applicando la distributiva tra intersezione e unione segue che sopravvivono solo un certo numero delle sfere arrangiate e che quindi $\mathcal{V}_c( \mathcal{S}(J))$ risulta essere unione di sfere arrangiate?}\\
%\begin{proof}
%(We have not a counterexample that the inclusion in equation \ref{unisfere} is strict) We start from a polynomial $K$ in $J$, $K$ vanishes on $\mathcal{V}_c(J)$, and thus $K^s$ vanishes on $\mathbb{S}_{\mathcal{V}_c(J)}$. Thus
% $$\mathcal{V}_c( \mathcal{S}(J))\supseteq \mathbb{S}_{\mathcal{V}_c(J)}.$$
%Moreover, $a\in \mathcal{V}_c( \mathcal{S}(J))$, if and only if $K^s(a)=0$ for any $K\in J$.\\
% Recalling Proposition \ref{corrsimm}, if $a\in \mathcal{V}_c( \mathcal{S}(J))$ for any $K\in J$ there exists $\tilde a \in \mathbb S_a$ such that $K(\tilde a)=0$. That is $\mathbb S_a \subseteq \mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}$, hence $$\mathcal{V}_c( \mathcal{S}(J))\subseteq \bigcap_{K\in J}\mathbb{S}_{\mathcal{V}_c(\langle K\rangle)}.$$
% Conversely if $a\in \bigcap_{K\in J}\mathbb{S}_{\mathcal{V}_c(K)}$ then $\mathbb{S}_a \subseteq \bigcap_{K\in J}\mathbb{S}_{\mathcal{V}_c(K)}$, and hence $K^s(\mathbb S_a)=0$ for any $K\in J$. Since $a$ has commuting components, we conclude that
%$\mathbb S_a\subseteq \mathcal{V}_c( \mathcal{S}(J)).$
% \end{proof}
%\noindent Note that $ \mathcal{S}(J)$ is an ideal, by definition. Moreover $ \mathcal{S}(J)\subseteq J$.\\
%Note that $\mathcal{V}_c(I)=\mathcal{V}_c( \mathcal{S}(I))$ whenever $I$ is generated by polynomials with real coefficients as stated in the following
%\
%\begin{proof} Suppose $I=\langle P_1,\cdots,P_l\rangle$ which $P_i\in \mathbb{R}{[q_1,q_2,\cdots,q_n]}$ then $ \mathcal{S}(I)$ is generated by $\langle P^s_1,\cdots,P^s_l, P_i*P_j\rangle$, with $i<j$. Moreover $\mathcal{V}_c(I)=\mathcal{V}_c( \mathcal{S}(I))$ because we evaluate polynomials on points with commuting components.
%\end{proof}
%\noindent This happens when $I=\mathcal{J}(V)$ and $V$ is an arranged spherical set by Proposition \ref{GeneratoriSfere}.
% \noindent We can now state the following questions,
% \begin{enumerate}
% \item In case of reducibility. Which is the relation between $I$, $I_1$ and $I_2$? Maybe the generators of $I$ can be obtained as $*$-product of the generators of $I_1$ and $I_2$, (see Examples \ref{red1}, \ref{red2})?
% \item The irreducibility of $V=\mathcal{V}(I)$ has some relations with the irreducibility of generators of $I$?
% \item $V_c(I)$ irreducible in our sense if and only if $\mathbb{S}_{\mathcal{V}_c{(I)}}$ is irreducible in our sense?
% \item $\mathbb{S}_{\mathcal{V}_c(J)}=\mathcal{V}_{\color{red}c}( \mathcal{S}(J))$?
% %\item $V^s$ irreducible in our sense if and only if it is irreducible in the sense of Fabio?
% %\item the symmetrisation of an algebraic set $\mathcal{V}(I)$ is still of the form $\mathcal{V}(\tilde{I})$, that is $[\mathcal{V}(I)]^s=\mathcal{V}(\tilde{I})$?
% %\item which is the relation between $\tilde I$ and $I^s$?
% \end{enumerate}
%
% For $(4)$ we have proved an inclusion.\\
% \begin{proposition} The symmetrization of an algebraic set $V(I)$, $[V(I)]^s,$ is contained in $V(I^s)$
% \end{proposition}
%\begin{proof} This easily follows, taking into account that $\mathcal{V}_c(I)\subseteq \mathcal{V}_c(I^s)$ and that, by construction, $[V]^s=[V_c]^s$ (application of Proposition 0.20).
%\end{proof}
%\noindent The converse is not true in general. For example consider $I=\langle q_1-j,q_2-j\rangle$ and $I^s=\langle q_1^2+1,q_2^2+1\rangle,$ we get $$[\mathcal{V}(I)]^s=\mathbb{S}_{(i,i)}\subseteq \mathbb{S}\times \mathbb{S}=\mathcal{V}(I^s).$$
%Note that, in this example, we have $[\mathcal{V}(I)]^s=\mathcal{V}(\tilde{I})\subset \mathcal{V}(I^s)$ for $\tilde{I}=\langle q_1-q_2,q_1^2+1\rangle$.\\\\
%\noindent These first examples lead us to formulate the following\\\\
% {\bf{Conjecture 1}} An algebraic set $\mathcal{V}(I)$ is irreducible if and only if $I$ admits as generators irreducible polynomials or the symmetrisation of linear polynomials of the form $(q_i-c)$ (or of an irreducible polynomial?).\\
%
%\noindent Note that it is easy to prove that if $\mathcal{V}_c(I)$ is
%irreducible then $\mathcal{V}(I)$ is irreducible.
% We can prove a particular case of the conjecture
% \begin{proposition} \label{unique}Let $I$ be a radical ideal generated by a {unique} polynomial $H$
% %{\color{red} It suffices to assume that $\mathcal{J}(\mathcal{V}(I))=\sqrt{I}$ is generated by a unque polynomial $H$, $\langle H\rangle=\sqrt{I}$}
%such that $\mathcal{V}_c(I)$ is irreducible, then $H$ is irreducible or the symmetrisation of an irreducible polynomial.
% \end{proposition}
% \begin{proof} Take $V=\mathcal{V}(I)$ irreducible. Take $H$ the generator of $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$.
% Suppose $H$ is reducible that is $H=P*Q$, we can assume that $Q$ is a non constant irreducible factor. We have that $P*Q$ vanishes on $\mathcal{V}(I)$, and thus it vanishes also on $\mathcal{V}_c(I).$ On this set $P*Q$ has a manageable expression. Indeed for all $\textbf{\em a}=(a_1,a_2\cdots,a_n)\in \mathcal{V}_c(I)$ with $a_l a_m=a_m a_l$ for all $l,m=1,\cdots,n$ $$(P*Q)(\mathbf{ a})=P(\textbf{\em a})\cdot (Q({P}^{-1}(\textbf{\em a})\textbf {\em a}P(\textbf{\em a}))$$
% thus either $P(\textbf {\em a})=0$ or $(Q({P}^{-1}(\textbf{\em a})\textbf{\em a}P(\textbf{\em
% a}))=0$. Take now the two subsets of $\mathcal{V}_c(I),$ given
% by $$V_1=\mathcal{V}(I)\cap \mathcal{V}_c(\langle P\rangle)=\mathcal{V}_c(\langle P\rangle)$$
% and
%
% $$ V_2=\mathcal{V}(I)\cap \mathbb{S}_{\mathcal{V}_c(\langle
% Q\rangle)}=
% \mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
%{Q^s}\rangle)=
% \mathcal{V}_c(\langle {Q^s},H\rangle)$$
% Now either ${\bf a}$ belongs to $V_1$ or one of
% its conjugates belongs to $V_c(\langle Q\rangle)$, thus ${\bf a}$
% belongs to $V_2$; therefore $$\mathcal{V}_c(I)= V_1\cup {V_2}$$
%
% In general
% there are two possibilities:
% \begin{enumerate}
% \item $\mathcal{V}_c(I)$ irreducible (and $\mathcal{V}(I)$ irreducible);
% \item $\mathcal{V}(I)$ irreducible and $\mathcal{V}_c(I)$ reducible;
% \end{enumerate}
% Under our assumptions we are in case $(1)$: Since $\mathcal{V}_c(I)$ is irreducible, then
% $$\mathcal{V}_c(I)=V_1=\mathcal{V}_c(\langle P\rangle)$$
% and thus $P$ vanishes on $\mathcal{V}_c(I)$ and hence $P\in \mathcal{I}(\mathcal{V}_c(I))=\sqrt{I}=\langle H\rangle$ (here we use the fact that $I$ is radical). Therefore we can take $P=H*P_1=P*Q*P_1$ that is $Q*P_1=1$ and thus $Q$ is constant and $H$ is therefore irreducible.
%\noindent The other possibility is that $\mathcal{V}_c(\langle
%H\rangle)=V_2$. From the Remark after Proposition \ref{corrsimm} we have
%that $$V_2=\mathcal{V}{(I)}\cap\mathcal{V}_c(\langle {Q^s}\rangle).$$
%Note that $$\mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
%{Q^s}\rangle)=\mathcal{V}_c(\langle {Q^s},H\rangle)$$ we thus
%have $$\mathcal{V}_c(\langle H\rangle)=\mathcal{V}_c(\langle
%{Q^s},H\rangle)$$ that is $Q^s\in \langle H\rangle$. Thus $Q^s$ is a
%polynomial that can be written as a right multiple of $H$.
%Take $H^s=P^s*Q^s=P^s*H*Q_1=H*Q_1*P^s$, that is $H^c=Q_1*P^s=Q^c*P^c$ which implies that $Q_1*P=Q^c$, but recalling that $Q^c$ is irreducible, we get that $Q_1$ is a constant and thus $H$ is a multiple of $Q^s$.
%%Thanks
%%to the irreducibility of $Q$ this can only occur if $P$ is a multiple
%%of $Q^c$, that is $H$ is a multiple of $Q^s$,
% i.e. $H$ is the
%symmetrization of an irreducible factor.
% \end{proof}
% \noindent
\noindent
From the irreducibility of a slice algebraic set we can deduce some algebraic properties of the corresponding ideal.
\begin{theorem}\label{VirredIprime} Let $I$ be a radical ideal of $\mathbb{H}[q_1,\ldots,q_n]$ such that $\mathcal{V}_c(I)$ is irreducible. Then $I$ is a quasi prime ideal.
\end{theorem}
% {\color{red} Maybe we can say more: i.e. If $I$ is such that $\mathcal{V}_c(I)$ is irreducible, then $\sqrt{I}$ is a quasi prime ideal. The proof is the same taking into account that $\mathcal{V}_c(\sqrt{I})=\mathcal{V}_c(I).$ }
%
\begin{proof}
Suppose $P*Q\in I$, with $Q$ a non-constant polynomial. We have that $P*Q$ vanishes on $\mathcal{V}(I)$, and thus it vanishes also on $\mathcal{V}_c(I).$ On this set $P*Q$ has a manageable expression. Indeed for all $\textbf{\em a}=(a_1,a_2\cdots,a_n)\in \mathcal{V}_c(I)$, either $P(\textbf {\em a})=0$, or
% with $a_l a_m=a_m a_l$ for all $l,m=1,\cdots,n$
$$(P*Q)(\textbf{\em a})=P(\textbf{\em a})\cdot Q({P}^{-1}(\textbf{\em a})\textbf {\em a}P(\textbf{\em a})),$$
that is $Q({P}^{-1}(\textbf{\em a})\textbf{\em a}P(\textbf{\em
a}))=0$.
Take now the two subsets of $\mathcal{V}_c(I),$ given
by $$V_1=\mathcal{V}(I)\cap \mathcal{V}_c(\langle P\rangle)=\mathcal{V}_c(I+\langle P\rangle)\subseteq \mathcal{V}_c(\langle P\rangle)$$
and
$$ V_2=\mathcal{V}(I)\cap \mathbb{S}_{\mathcal{V}_c(\langle
Q\rangle)}=
\mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
{Q^s}\rangle)=\mathcal{V}_c(I+\langle Q^s\rangle)\subseteq
\mathcal{V}_c(\langle {Q^s}\rangle).$$
Now either ${\textbf{\em a}}$ belongs to $V_1$ or one of
its conjugates belongs to $V_c(\langle Q\rangle)$, thus ${\textbf{\em a}}$
belongs to $V_2$; therefore $$\mathcal{V}_c(I)= V_1\cup {V_2}.$$
Since $\mathcal{V}_c(I)$ is irreducible,{ the first possibility is that
$$\mathcal{V}_c(I)=V_1\subseteq \mathcal{V}_c(\langle P\rangle).$$
In this case $P$ vanishes on $\mathcal{V}_c(I),$ and hence $P\in \mathcal{J}(\mathcal{V}_c(I))=I$ since $I$ is radical; the second possibility is that $\mathcal{V}_c(I)=V_2\subseteq \mathcal{V}_c(\langle {Q^s}\rangle)$. In this case $Q^s\in \mathcal{J}(\mathcal{V}_c(I))=I$, since $I$ is radical}.
\end{proof}
\noindent {The first result in the other direction is the following.}
\begin{theorem} \label{nosfere}Let $I$ be a quasi prime and radical ideal of $\mathbb{H}[q_1,\ldots,q_n]$.
% Then $\mathcal{V}_c(I)$ cannot split as the union of two sets $\mathcal{V}_c(I_1)\cup\mathcal{V}_c(I_2)$ in which at least one is an algebraic set of the form $\mathcal{V}_c(J)=\mathcal{V}_c( \mathcal{S}(J)),$ for a certain ideal $J$. {\color{red} We have proven more: indeed, whenever $I$ is a quasi prime radical ideal
Then $\mathcal{V}_c(I)$ cannot be written as a non-trivial union of the form $\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$ where either $\mathcal{V}_c(I_1)$ or $\mathcal{V}_c(I_2)$ has spherical symmetry.
%$\mathcal{V}_c(I_1)$ and $\mathcal{V}_c(I_2)$
% where $\mathcal{V}_c(I_2)$ has spherical symmetry,
% % i.e. $\mathcal{V}_c( \mathcal{S}(I_2))=\mathcal{V}_c(I_2)$,
% then
% $\mathcal{V}_c(I)$
% %=\mathcal{V}_c( \mathcal{S}(I))$
%%$\mathcal{V}_c(I)$
%has spherical symmetry and is irreducible.
%
\end{theorem}
\begin{proof}
%Take $I$ quasi prime and radical (we still not know if quasi prime implies radical). We want to prove that $\mathcal{V}_c(I)$ is irreducible.
Suppose, by contradiction, that $\mathcal{V}_c(I)$ can be written as $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$$ where $\mathcal{V}_c(I_2)$ has spherical symmetry. {Suppose that there exists} $H\in \mathcal{J}(\mathcal{V}_c(I_1))\setminus{I}$ and take any $K\in \mathcal{J}(\mathcal{V}_c(I_2))$.
Consider the polynomial $H*K$, and take a point $\textbf{\em a}\in \mathcal{V}_c(I)$. {If $\textbf{\em a}\in \mathcal{V}_c(I_1)$, then $H(\textbf{\em a})=0$ and hence $H*K(\textbf{\em a})=0$ as well; %if otherwise $\textbf{\em a}\in \mathcal{V}_c(I_2)$. In the first case
%H(\textbf{\em a})K(H^{-1}(\textbf{\em a})\textbf{\em a}H(\textbf{\em a}))=0.$$
if, otherwise, $\textbf{\em a}\in \mathcal{V}_c(I_2)$, then $\mathbb{S}_{\textbf{\em a}}\subseteq \mathcal{V}_c(I_2)$ since $\mathcal{V}_c(I_2)=\mathcal{V}_c( \mathcal{S}(I_2))$ and hence
%Therefore,%for any $\textbf{\em a} \in \mathcal{V}_c(I)$,
$$H*K(\textbf{\em a})=H(\textbf{\em a})K(H^{-1}(\textbf{\em a})\textbf{\em a}H(\textbf{\em a}))=0.$$
Thus $H*K\in \mathcal{J}(\mathcal{V}_c(I))=\sqrt{I}=I$. Recalling that $I$ is quasi prime and that $H\notin I$ we necessarily have that $K^s\in I$.
Since $K$ was chosen arbitrarily in $\mathcal{J}(\mathcal{V}_c(I_2))$, which contains $I_2$, we get that
%$I_2\subseteq \mathcal{J}(\mathcal{V}_c(I_2))$ we have $K^s\in I$, and thus
$$\mathcal{V}_c(I_2)\subseteq \mathcal{V}_c(I) \subseteq \bigcap_{K\in I_2}{\mathcal{V}_c(\langle K^s\rangle)}= \mathcal{V}_c( \mathcal{S}(I_2))$$ %={\color{blue}\bigcap_{K\in I_2}\mathbb{S}_{{\mathcal{V}}_c(\langle K\rangle)}}=\supseteq \mathcal{V}_c(I)\supseteq .$$ \marginpar{togliamo il blu?}
\noindent Since by hypothesis $\mathcal{V}_c(I_2)=\mathcal{V}_c( \mathcal{S}(I_2)),$ we have that $\mathcal{V}_c(I)=\mathcal{V}_c(I_2)$ and hence we conclude that $\mathcal{V}_c(I)$ is irreducible. }
\end{proof}
{\noindent As a consequence of the previous theorem we have the following.
{\begin{corollary}\label{coro}
Let $I$ be a quasi prime and radical right ideal of $\mathbb H[q_1,\ldots,q_n]$ such that $\mathcal{V}_c(I)$ has spherical symmetry. Then $\mathcal V_c(I)$ is irreducible.
\end{corollary}}
\begin{proof}
If $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$ and $\mathcal V_c(I)=\mathcal V_c(I_1)\cup \mathcal V_c(I_2)$ is reducible, then $\mathbb S_{\mathcal V_c(I)}=\mathbb S_{\mathcal V_c(I_1)}\cup \mathbb S_{\mathcal V_c(I_2)}=\mathcal V_c(I_1)\cup \mathcal V_c(I_2)$. Therefore $\mathcal V_c(I_1)$ and $\mathcal V_c(I_2)$ consist of the union of arranged spheres. Then $\mathbb S_{\mathcal V_c(I_1)}=\mathcal V_c(I_1)$ and $\mathbb S_{\mathcal V_c(I_2)}=\mathcal V_c(I_2)$, in contradiction with Theorem \ref{nosfere}.
%The proof follows from and Proposition \ref{spherical}.
\end{proof}
}
%\begin{remark}
\noindent In view of Proposition \ref{equivalence}, Corollary \ref{coro} can be restated as follows
\begin{corollary}\label{coro2}
If $I$ is a quasi prime and radical two-sided ideal of $\mathbb H[q_1,\ldots,q_n]$, then $\mathcal V_c(I)$ is irreducible.
\end{corollary}
\noindent Notice that if $\mathcal{V}_c(I)$ is irreducible and has spherical symmetry, then, either $\mathcal{V}_c(I)$ consists of a unique arranged sphere or it is the union of an infinite number of arranged spheres. {We will give an explicit example of this phenomenon at the end of Section \ref{Sprinc}.}
% \end{remark}
\noindent Theorem \ref{nosfere} can be used as a tool to establish if an ideal is quasi prime, as shown by the following examples.
\begin{example}
The ideal $I=\langle q_1^2+1,q_2^2+1\rangle$ is not quasi prime. Indeed $\mathcal{V}(I)=\mathbb{S}\times \mathbb{S}$ is irreducible while $\mathcal{V}_c(I)=\mathbb{S}_{(i,i)}\cup\mathbb{S}_{(-i,i)}$ is reducible. Thanks to the Strong Nullstellensatz \ref{snull} we have that
\[\sqrt{I}=\mathcal J(\mathbb{S}\times \mathbb{S})=\langle q_1^2+1,q_2^2+1\rangle=I,\]
that is $I$ is radical. Theorem \ref{nosfere} implies then that $I$ is not quasi prime.
% Actually we have a direct proof that $I$ is not quasi-prime: the polynomial $q_1^2-q_2^2=(q_1+q_2)*(q_1-q_2) \in I$, but neither $q_1\pm q_2 \in I $ nor $(q_1\pm q_2)^s=q_1^2+q_2^2\pm 2q_1q_2 \in I$. In fact they do not vanish on the entire $\mathbb S \times \mathbb S$.
\end{example}
\begin{example}
The ideal $I=\langle (q_1^2+1)*(q_2^2+1)\rangle$ is not quasi prime. In fact, as above, the Strong Nullstellensatz yields that $I$ is radical but $\mathcal{V}(I)=(\mathbb{S}\times\mathbb{H})\cup (\mathbb{H}\times\mathbb{S})$ and hence
\[\mathcal{V}_c(I)=\bigcup_{\substack{x,y \in \mathbb R \\ I\in \mathbb S}}\mathbb{S}_{(I,x+yI)} \cup \bigcup_{\substack{x,y \in \mathbb R \\ I\in \mathbb S}}\mathbb{S}_{(x+yI,I)}.\]
Thanks to Theorem \ref{nosfere}, the ideal $I$ is not quasi prime.
\end{example}
%\begin{theorem}\label{cpirred}
% Let $I$ be a completely prime right ideal such that $\mathbb S_{\mathcal V_c(I)}$ is irreducible. Then $\mathcal V_c(I)$ is irreducible.
%\end{theorem}
%\begin{proof}
%Suppose by contradiction that $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$$ is reducible. Then
%\[\mathbb S_{\mathcal{V}_c(I)}=\mathbb S_{\mathcal{V}_c(I_1)}\cup \mathbb S_{\mathcal{V}_c(I_2)}.\]
%Thus we can assume, without loss of generality, that $\mathbb S_{\mathcal{V}_c(I_1)}\subseteq\mathbb S_{\mathcal{V}_c(I_2)}$.
%%or $\mathbb S_{\mathcal{V}_c(I_1)}\supseteq\mathbb S_{\mathcal{V}_c(I_2)}.$
%This implies that for any ${\aaa} \in \mathcal{V}_c(I_1) \setminus \mathcal{V}_c(I_2)$ there exists ${\bf \tilde \aaa} \in \mathcal{V}_c(I_2) \cap \mathbb S_{\aaa}$, and hence the entire $\mathbb{S}_{\aaa}\subset \mathcal V_c(I)$. Since $\textbf{\em a} \notin \mathcal V_c(I_2)$, then necessarily $\mathbb S_{\aaa}$ is contained in $\mathcal{V}_c(I_1)$.
%Denote by $V_1=\mathbb S_{\mathcal V_c(I_1)\setminus \mathcal V_c(I_2)}$. Notice that $V_1 \subseteq \mathcal V_c(I_1)$ and \[\mathcal V_c(I)=V_1\cup \mathcal V_c(I_2).\]
%Consider now a polynomial $P \in \mathcal J(\mathcal V_c(I_2))\setminus I$, and let $K$ be any polynomial in $\mathcal J(V_1)$. We want to prove that $P*K \in I$. For any $\aaa \in \mathcal V_c(I)$, if $\aaa \in \mathcal V_c(I_2)$, $P(\aaa)=0$ and hence $P*K(\aaa)=0$. Let $\aaa \in V_1\setminus \mathcal V_c(I_2)$. If $P(\aaa)=0$, as before we conclude that $P*K(\aaa)=0$. Otherwise, since $\mathbb S_\aaa \subset V_1$, using Proposition \ref{prodstar}, we have
%\[P*K(\aaa)=P(\aaa)K(P(\aaa)^{-1}\aaa P(\aaa))=0.\]
%Thus $P*K \in \mathcal J(\mathcal V_c(I))=I$ since $I$ is radical. \\
%Notice that $\mathcal J(V_1)\supseteq I$, since $P*K \in I$ for any $K\in \mathcal J(V_1)$, we obtain that $P*I \subseteq I$.\\
%Since $I$ is completely prime and $P$ does not belong to $I$, we conclude that for any $K \in \mathcal J(V_1)$, $K$ belongs to $I$, that is $\mathcal J(V_1)\subseteq I$ and then $\mathcal J(V_1)= I$. Using Proposition \ref{Zc} we get that $I$ is a two-sided ideal and then, thanks to Proposition \ref{twosided}, $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$ which is irreducible.
%\end{proof}
\noindent Another result in the direction of characterizing ideals whose corresponding algebraic set are irreducible is the following.
{\begin{theorem}\label{cpirred}
Let $I$ be a quasi prime and radical right ideal of $\mathbb H[q_1,\ldots,q_n]$ such that $\mathbb S_{\mathcal V_c(I)}$ is irreducible. Then $\mathcal V_c(I)$ is irreducible.
\end{theorem}
\begin{proof}
Suppose by contradiction that $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$$ is reducible. Then
\[\mathbb S_{\mathcal{V}_c(I)}=\mathbb S_{\mathcal{V}_c(I_1)}\cup \mathbb S_{\mathcal{V}_c(I_2)}.\]
Thus we can assume, without loss of generality, that $\mathbb S_{\mathcal{V}_c(I_1)}\subseteq\mathbb S_{\mathcal{V}_c(I_2)}$.
%or $\mathbb S_{\mathcal{V}_c(I_1)}\supseteq\mathbb S_{\mathcal{V}_c(I_2)}.$
This implies that for any ${\aaa} \in \mathcal{V}_c(I_1) \setminus \mathcal{V}_c(I_2)$ there exists ${\bf \tilde \aaa} \in \mathcal{V}_c(I_2) \cap \mathbb S_{\aaa}$, and hence the entire $\mathbb{S}_{\aaa}\subset \mathcal V_c(I)$. Since $\textbf{\em a} \notin \mathcal V_c(I_2)$, then necessarily $\mathbb S_{\aaa}$ is contained in $\mathcal{V}_c(I_1)$.
Denote by $V_1=\mathbb S_{\mathcal V_c(I_1)\setminus \mathcal V_c(I_2)}$. Notice that $V_1 \subseteq \mathcal V_c(I_1)$ and \[\mathcal V_c(I)=V_1\cup \mathcal V_c(I_2).\]
Consider now a polynomial $P \in \mathcal J(\mathcal V_c(I_2))\setminus I$, and let $K$ be any polynomial in $\mathcal J(V_1)$. We want to prove that $P*K \in I$. For any $\aaa \in \mathcal V_c(I)$, if $\aaa \in \mathcal V_c(I_2)$, $P(\aaa)=0$ and hence $P*K(\aaa)=0$. Let $\aaa \in V_1\setminus \mathcal V_c(I_2)$. If $P(\aaa)=0$, as before we conclude that $P*K(\aaa)=0$. Otherwise, since $\mathbb S_\aaa \subset V_1$, using Proposition \ref{prodstar}, we have
\[P*K(\aaa)=P(\aaa)K(P(\aaa)^{-1}\aaa P(\aaa))=0.\]
Thus $P*K \in \mathcal J(\mathcal V_c(I))=I$ since $I$ is radical. \\
Since $I$ is quasi prime and $P$ does not belong to $I$, for any $K \in \mathcal J(V_1)$, $K^s$ belongs to $I$, that is $\mathcal S(\mathcal J(V_1))\subseteq I$. Since $I$ is radical we have that
$\sqrt{\mathcal S(\mathcal J(V_1))}\subseteq I$. Moreover, recalling Corollary \ref{essealg},
\[ \sqrt{\mathcal S(\mathcal J(V_1))}= \mathcal J (\mathbb S_{\mathcal V_c (\mathcal J(V_1))})= \mathcal J ( \mathcal V_c (\mathcal J(V_1)))=\mathcal J(V_1),\]
where we used the fact that $\mathcal V_c(\mathcal J(V_1))$ has spherical symmetry and $\mathcal J(V_1)$ is radical.
Therefore, we deduce that $\mathcal J(V_1)\subseteq I$. The other inclusion holds since $V_1 \subseteq \mathcal V_c(I)$, and thus $I=\mathcal J(V_1)$.
Using Proposition \ref{Zc} we get that $I$ is a two-sided ideal and then, thanks to Proposition \ref{twosided}, $\mathcal V_c(I)=\mathbb S_{\mathcal V_c(I)}$ which is irreducible.
\end{proof}
}
%{\color{blue} Non ho capito: dimostriamo anche che $\mathcal V_c(I)$ ha simmetria sferica??}
\noindent Consequently, we also have
\begin{corollary}\label{coro3}
If $I$ is a {quasi} prime {and radical} ideal of $\mathbb H[q_1,\ldots,q_n]$ such that $\sqrt{\mathcal S(I)}$ is quasi prime, then $\mathcal V_c(I)$ is irreducible.
\end{corollary}
\begin{proof}
{The ideal $\sqrt{\mathcal S(I)}$ is by assumption quasi prime and radical. Moreover, recalling that $
\mathcal V_c(\sqrt{\mathcal S(I)})=\mathcal V_c({\mathcal S(I)})$, we have that
% =\mathbb S_{\mathcal V_c(I)},\]
% and
\[\mathbb S_{\mathcal V_c(\sqrt{\mathcal S(I)}) }=\mathbb S_{\mathcal V_c({\mathcal S(I)}) }=\mathcal V_c({\mathcal S(I)})\]
since $\mathcal V_c ({\mathcal S(I)})$ has always spherical symmetry.
Thus $\mathcal V_c(\sqrt{\mathcal S(I)})$ has spherical symmetry as well.
% \mathcal V_c({\mathcal S(I)})=\mathcal V_c(\sqrt{\mathcal S(I)})=\mathcal V_c (\mathcal J(\mathbb S_{\mathcal V_c(I)}))=\mathbb S_{\mathcal V_c(I)}\] has spherical symmetry,
Corollary \ref{coro} then yields that $\mathcal V_c (\sqrt{\mathcal S(I)})$ is irreducible. Since $\mathcal V_c (\sqrt{\mathcal S(I)})=\mathcal V_c({\mathcal S(I)})=\mathbb S_{\mathcal V_c(I)}$,
%$\mathbb S_{\mathcal V_c(I)}$ is irreducible.
Theorem \ref{cpirred} allows us to conclude.
}
\end{proof}
\subsection{Irreducibility of ideals}
% {{\color{blue} Diamo le definizioni nell'anello dei polinomi? con right ideals?}
% }
\noindent In Algebra there exists a classical notion of irreducibility for ideals, {which in our framework reads as follows}.
\begin{definition}\label{redI}
A right ideal $I$ {of $\mathbb H[q_1,\ldots,q_n]$} is said to be {\em irreducible} if, whenever $I=I_1\cap I_2$ {for two right ideals $I_1,I_2$}, it follows that either $I=I_1$ or $I=I_2$.
If not, then
$I$ is said to be {\em reducible}.
\end{definition}
%\marginpar{\color{red} {Pensare a una definizione di $s$--irreducibility e legami con ideali quasi prime}}
\noindent In view of the geometric properties of the zeros of quaternionic slice regular polynomials, it is useful to introduce also the following definition. \begin{definition}
A right ideal $I$ {of $\mathbb H[q_1,\ldots,q_n]$} is said to be {\em s-irreducible} if, whenever $I=I_1\cap \mathcal S(I_2)$ for some right ideals $I_1,I_2$, it follows that either $I=I_1$ or $I=\mathcal S(I_2)$.
If not, then
$I$ is said to be {\em s-reducible}.
\end{definition}
% {\color{red} Forse togliamo solo questa
% \begin{definition}
% An ideal $I$ is said to be {\em r-irreducible} if, whenever $I=I_1\cap I_2$, with $I_2$ generated by polynomials with real coefficients, it follows that either $I=I_1$ or $I=I_2$.
% If not, then
% $I$ is said to be {\em r-reducible}.
% \end{definition}}
% }
\noindent The slice algebraic set associated with an irreducible radical ideal is irreducible.
\begin{proposition}\label{idirred}
Let $I$ be a right ideal {of $\mathbb{H}[q_1,\ldots,q_n]$} whose radical $\sqrt {I}$ is an irreducible right ideal. Then $\mathcal V_c(I)$ is an irreducible {slice} algebraic set.
\end{proposition}
\begin{proof}
Suppose by contradiction that
\[\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)\]
is reducible. Hence, thanks to the Strong Nullstellensatz \ref{snull} and to Proposition \ref{vanishing},
\[\sqrt{I}=\mathcal{J}(\mathcal{V}_c(I))=\mathcal{J}(\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2))= \mathcal{J}(\mathcal{V}_c(I_1))\cap \mathcal{J}(\mathcal{V}_c(I_2))=\sqrt{I_1}\cap \sqrt{I_2},\]
that is $\sqrt{I}$ is reducible.
\end{proof}
\noindent The converse is not true in general.
\begin{example}\label{qp_not_irred}%varietà irriducibile con ideale riducibile.
Let $I=\langle q^2+1\rangle$. Then, $\mathcal V_c(I)=\mathbb S_i$ is irreducible, but the ideal $I$ is radical and reducible, as $I=\langle q-j \rangle \cap \langle q-k \rangle$.
\end{example}
\noindent Thanks to Theorem \ref{VirredIprime} and Proposition \ref{idirred}, we have
\begin{corollary}
If $I$ is a radical and irreducible right ideal {of $\mathbb H[q_1,\ldots,q_n]$}, then $I$ is a quasi prime ideal.
\end{corollary}
\noindent {In the other direction, we can prove the following}
{\begin{corollary}
If $I$ is a quasi prime and radical right ideal of $\mathbb H[q_1,\ldots,q_n]$, then $I$ is s-irreducible.
%{\color{red}(and r-irreducible)}.
\end{corollary}
\begin{proof}
Suppose by contradiction that $I=I_1\cap \mathcal S(I_2)$. Then, thanks to Corollary \ref{essei}, $\mathcal V_c(I)=\mathcal V_c(I_1)\cup \mathcal V_c(\mathcal S(I_2))$. Which is in contradiction with Theorem \ref{nosfere}.
\end{proof}
}
%Domanda: $J$ quasi primo implica $\sqrt{J}$ quasi primo?
%\begin{corollary} Let $J$ be a radical and quasi prime ideal generated by polynomials with real coefficients then $\mathcal{V}_c(J)$ is irreducible. \end{corollary}
%
%\begin{corollary} Let $J=\langle L\rangle$ where $L$ is an irreducible polynomial with real coefficients, then $\mathcal{V}_c(I)$ is irreducible.\end{corollary}
%
%{\color{green} Quest'ultimo corollario coincide con l'ultimo della prossima sezione}
\section{Principal ideals}\label{Sprinc}
\noindent If $I$ is a right ideal of $\mathbb{H}[q_1,\ldots,q_n]$ generated by {the polynomial $H$, then
$\mathcal{V}_c(I)$ coincides with the set $Z_c(H)$ of zeros of $H$.}
\begin{proposition} \label{irrcp} Let $H\in \mathbb H[q_1,\ldots,q_n]$. The ideal $I=\langle H\rangle$ is completely prime if and only if the polynomial $H$ is irreducible.\end{proposition}
\begin{proof} Suppose{ that $I$ is completely prime and, by contradiction, that $H=P*Q$, with $Q$ a non-constant irreducible factor, and $P$ non-constant.
Note that $(P*Q)*Q^c=P*Q^s\subseteq I$ and $Q^s*I=I*Q^s\subseteq
I$}. Then either $P$ or $Q^s$ belong to $I$. If $P\in I$, then
$\langle P \rangle \subset I=\langle P*Q \rangle \subset \langle P
\rangle$, hence $I=\langle P\rangle$ and $Q$ is a constant polynomial, a contradiction. % we can take it as a generator of $I$ with lower degree
Otherwise, if $Q^s\in
I$, we
have $$Q^s=P*Q*R$$ where $R$ is a polynomial in $\mathbb H[q_1,\ldots, q_n]$. Then, $P=Q^s*R^{-*}*Q^{-*}=R^{-*}*Q^c$, and
thus $Q^c=R*P$ and $Q=P^c*R^c$. Therefore, taking into account that $Q$ is irreducible, $R^c$ is constant that is $\langle H\rangle=\langle Q^s\rangle$. Now note that a completely prime ideal cannot be generated by the symmetrisation of an irreducible factor. Indeed, if $I=\langle Q^s\rangle$ we have $Q^s=Q*Q^c\in I$ and $Q*I\subseteq I$, thus either $Q\in I$ or $Q^c\in I$, but none of them can be written as a multiple of $Q^s$, {leading to a contradiction.}
Conversely, {suppose that $H$ is irreducible and
take $P$ and $Q$ in $\mathbb{H}[q_1,\ldots, q_n]$ such that $P*Q\in I$ and $P*I\subseteq I$. Therefore there exist $R,S$ polynomials in $\mathbb{H}[q_1,\ldots, q_n]$ such that $$P*Q=H*R$$ $$P*H=H*S.$$
Thus
$P=H*S*H^{-*}$ and $R=S*H^{-*}*Q$. Since $H$ is irreducible, we conclude that either $S=S_1*H$ and then $P=H*S_1 \in I$, or $Q=H*Q_1 \in I$. }%Note that If $H=H_1*H_2$ the relation $H*S*H^{-*}*Q=H*R$, would be satisfied also if $S=S_1*H_2$ and $Q=H_1*Q_1$.
%Thus $Q\notin I$, unless $Q_1=H_2*Q_2$ and $P=H*S_1*H_1^{-*}$ belongs to $I$ iff $S_1=S_2*H_1$ (bisogna escludere che $Q_1=H_2*Q_2$ e the $S_1=S_2*H_1$ )
\end{proof}
%
% \begin{proof} Suppose, by contradiction, that $H=P*Q$
%, we can assume that $Q$ is a non constant irreducible factor, and $P$ non constant.
% Note that $(P*Q)*Q^c=P*Q^s\subseteq I$ and $Q^s*I=I*Q^s\subseteq
% I$. Then either $P$ or $Q^s$ belong to $I$. If $P\in I$, then
% $\langle P \rangle \subset I=\langle P*Q \rangle \subset \langle P
% \rangle$, hence $I=\langle P\rangle$ and $Q$ is a constant, a contradiction. % we can take it as a generator of $I$ with lower degree
% Otherwise, if $Q^s\in
% I$, taking into account that $Q$ is irreducible, we
% have $$Q^s=P*Q*R$$ where $R$ is a polynomial and $P=R^{-*}*Q^c$, and
% thus $Q^c=R*P$ and $Q=P^c*R^c$, thus $R^c$ is constant that is $\langle H\rangle=\langle Q^s\rangle$. Now note that a completely prime ideal cannot be generated by the symmetrisation of an irreducible factor. Indeed, if $I=\langle Q^s\rangle$ we have $Q^s=Q*Q^c\in I$ and $Q*I\subseteq I$, thus either $Q\in I$ or $Q^c\in I$, but none of them can be written as a multiple of $Q^s$, a contradiction.\\
% Conversely,
%Take $P$ and $Q$ such that $P*Q\in I$ and $P*I\subseteq I$, that is $$P*Q=H*R$$ $$P*H=H*S$$ for $R,S$ polynomials in $\mathbb{H}[q_1,\ldots, q_n]$. Note that the fact that $H$ is irreducible implies that
%$P=H*S*H^{-*}$ and $R=S*H^{-*}*Q$ that is either $S=S_1*H$ (and then $P=H*S_1$) or $Q=H*T_1$. Note that If $H=H_1*H_2$ the relation $H*S*H^{-*}*Q=H*R$, would be satisfied also if $S=S_1*H_2$ and $Q=H_1*Q_1$.
%%Thus $Q\notin I$, unless $Q_1=H_2*Q_2$ and $P=H*S_1*H_1^{-*}$ belongs to $I$ iff $S_1=S_2*H_1$ (bisogna escludere che $Q_1=H_2*Q_2$ e the $S_1=S_2*H_1$ )
% \end{proof}
% \begin{center} FORSE PARTE COLORATA NON SERVE
% \end{center}
% {\color{blue}\begin{definition} An ideal $I$ is said to be {\em symmetrically prime} if and only if, whenever $P^s*Q^s\in I$ then either $P^s$ or $Q^s\in I$.
% \end{definition}
%
% \begin{proposition} If $H$ is an irreducible polynomial, then $\langle H^s \rangle$ is a symmetrically prime ideal.
% \end{proposition}
%
% \begin{proof} Let $I=\langle H^s \rangle$. Take $P^s$ and $Q^s$ such that their $*$-product belongs to $I$ i.e. $$P^s*Q^s=H^s*R$$ for some polynomial $R$. Hence $R=P^s*({H^s})^{{-*}}*Q^s$
% that is either $P^s$ contains an irreducible factor
% $\widetilde{H}$ such that $\widetilde{H}^s=H^s,$ and thus
% necessarily also its conjugate $\widetilde{H}^c$, (that is it
% contains $H^s$) or the same occurs for $Q^s$.
% \end{proof}
%
% Combining Propositions \ref{unique} and \ref{irrcp} we have proven that
\noindent {Specializing Theorem \ref{VirredIprime} to the case of a principal ideal, we get the following stronger result.}
\begin{theorem} \label{symprime} Let $I=\langle H \rangle$ be a radical right ideal {of $\mathbb{H}[q_1,\ldots, q_n]$} such that $\mathcal{V}_c(I)$ is irreducible.
Then %$I$ is completely prime or symmetrically prime. More precisely,
either $H$ is irreducible
or $H$ is the symmetrization of an irreducible polynomial.
% and hence $I$ is a {\em symmetrically prime} ideal.
% .
\end{theorem}
\begin{proof} %Re $V=\mathcal{V}(I)$ irreducible. Take $H$ the generator of $\mathcal{I}(\mathcal{V}(I))=\sqrt{I}$.
Suppose $H$ is reducible that is $H=P*Q$, {where }we can assume that $Q$ is a non-constant irreducible factor. We have that $P*Q$ vanishes on $\mathcal{V}(I)$, and thus it vanishes also on $\mathcal{V}_c(I).$ On this set $P*Q$ has an {explicit} expression (see Proposition \ref{prodstar}). Indeed for all $\textbf{\em a}\in \mathcal{V}_c(I)$, either $P(\textbf {\em a})=0$, or $$(P*Q)(\textbf{\em a})=P(\textbf{\em a})\cdot Q({P}^{-1}(\textbf{\em a})\textbf {\em a}P(\textbf{\em a})$$
and $Q({P}^{-1}(\textbf{\em a})\textbf{\em a}P(\textbf{\em
a})=0$.
Take now the two subsets of $\mathcal{V}_c(I),$ given
by $$V_1=\mathcal{V}(I)\cap \mathcal{V}_c(\langle P\rangle)=\mathcal{V}_c(\langle P\rangle)$$
and
$$ V_2=\mathcal{V}(I)\cap \mathbb{S}_{\mathcal{V}_c(\langle
Q\rangle)}=
\mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
{Q^s}\rangle)=
\mathcal{V}_c(\langle {Q^s},H\rangle)$$
Now either ${ \textbf{\em a}}$ belongs to $V_1$ or one of
its conjugates belongs to $V_c(\langle Q\rangle)$, thus ${\textbf{\em a}}$
belongs to $V_2$; therefore $$\mathcal{V}_c(I)= V_1\cup {V_2}$$
% In general
% there are two possibilities:
% \begin{enumerate}
% \item $\mathcal{V}_c(I)$ irreducible (and $\mathcal{V}(I)$ irreducible);
% \item $\mathcal{V}(I)$ irreducible and $\mathcal{V}_c(I)$ reducible;
% \end{enumerate}
%Under our assumptions we are in case $(1)$:
\noindent Since $\mathcal{V}_c(I)$ is irreducible, then either
$$\mathcal{V}_c(I)=V_1=\mathcal{V}_c(\langle P\rangle)$$
and thus $P$ vanishes on $\mathcal{V}_c(I)$ that is $P\in \mathcal{J}(\mathcal{V}_c(I))=\sqrt{I}=\langle H\rangle$. {Therefore there exists $P_1 \in \mathbb{H}[q_1,\ldots, q_n]$ such that $P=H*P_1=P*Q*P_1$} that is $Q*P_1=1$ in contradiction with the hypothesis on $Q$.
% and thus $Q$ is constant and $H$ is therefore irreducible.
\noindent The other possibility is that $\mathcal{V}_c(\langle
H\rangle)=V_2$,
% From the Remark after Proposition \ref{corrsimm} we have
% that $$V_2=\mathcal{V}{(I)}\cap\mathcal{V}_c(\langle {Q^s}\rangle).$$
% Note that $$\mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
% {Q^s}\rangle)=\mathcal{V}_c(\langle {Q^s},H\rangle)$$ we thus
% have $$\mathcal{V}_c(\langle H\rangle)=\mathcal{V}_c(\langle
% {Q^s},H\rangle)$$
and hence $Q^s\in \langle H\rangle$.
%Thus $Q^s$ is a
% polynomial that can be written as a right multiple of $H$.
\noindent therefore there exists a polynomial $Q_1$ such that
\[H^s=P^s*Q^s=P^s*H*Q_1=H*Q_1*P^s,\]
that is \[H^c=Q^c*P^c=Q_1*P^s\]
which implies that $Q_1*P=Q^c$. Recalling that $Q^c$ is irreducible (since $Q$ is irreducible), we get that $Q_1$ is a constant and thus $H$ is a multiple of $Q^s$.
%Thanks
%to the irreducibility of $Q$ this can only occur if $P$ is a multiple
%of $Q^c$, that is $H$ is a multiple of $Q^s$,
i.e. $H$ is the
symmetrization of an irreducible factor.
\end{proof}
% We give the following
% \begin{definition} An ideal $I$ is said to be {\em quasi-completely prime} if and only if given two polynomials $P,Q$ such that
% \begin{itemize}
% \item $P*Q\in I$
% \item $P*I\subseteq I$ (forse non serve questa richiesta)
% \end{itemize}
% then or $P$ or $Q^s\in I$.
% \end{definition}
%
%\begin{proof} From the fact that $P*Q\in I$ we have that $P*Q*Q^c=P*Q^s=Q^s*P\in I$, moreover $Q^s*I=I*Q^s\subseteq I$, since $I$ is completely prime we get that either $P\in I$ or $Q^s\in I$.
%\end{proof}
%\noindent Moreover
\begin{proposition}\label{Hsquasiprime} If $H$ is an irreducible polynomial such that $H\neq H^c$, then $\langle H^s \rangle$ is a quasi prime and radical ideal.
\end{proposition}
\begin{proof}
Let $P,Q$ be such that $P*Q=H^s*R$. If $P \notin \langle H^s \rangle$, then:
either $Q \in \langle H^s \rangle$ and hence $Q^s \in \langle H^s \rangle$; or there exists $\hat H$ such that
\[P=P_1*\hat H \quad \text{and} \quad Q=\hat H^c * Q_1, \quad \text{with $\hat H^s=H^s$.}\]
Thus $Q^s \in \langle H^s \rangle$ {and $\langle H^s \rangle$ is a quasi prime ideal.}
{\noindent Consider now} $P\in \sqrt{\langle H^s \rangle}=\mathcal J(\mathcal V_c(\langle H^s \rangle))$. Then $P$ vanishes on $\mathbb S_{\mathcal V_c(H)}$ and thus on $\mathcal V_c(H) \cup \mathcal V_c(H^c)$. Since $H$ is irreducible, we have that both $\langle H\rangle $ and $\langle H^c \rangle$ are radical and hence $P \in \langle H\rangle \cap \langle H^c \rangle$. Therefore there exist $P_1,P_2$ such that
\[P=H*P_1=H^c*P_2.\]
Since $H\neq H^c$ {and are both irreducible}, then necessarily there exists $P_3$ such that $P_2=H*P_3$. Thus $P \in \langle H^s \rangle$ {and $\langle H^s \rangle$ is a radical ideal.}
\end{proof}
\noindent Observe that if $H=H^c$, then $\langle H^s \rangle=\langle H^{*2} \rangle$ which is not a radical ideal.
%{\color{red} Togliere?
%
% Combining Theorem \ref{symprime} and Proposition \ref{Hsquasiprime}, we have
%\begin{corollary}
%Let $I=\langle H \rangle$ be a radical ideal such that $\mathcal{V}_c(I)$ is irreducible.
%Then $I$ is quasi prime.
%\end{corollary}
%}
% \noindent With this notation, following the same steps of the proof of Proposition \ref{unique}, but removing the assumption that $I$ has only one generator, we can prove
%
% \begin{proof}
% Suppose $P*Q\in I$, we can assume that $Q$ is a non constant factor. We have that $P*Q$ vanishes on $\mathcal{V}(I)$, and thus it vanishes also on $\mathcal{V}_c(I).$ On this set $P*Q$ has a manageable expression. Indeed for all $\textbf{\em a}=(a_1,a_2\cdots,a_n)\in \mathcal{V}_c(I)$ with $a_l a_m=a_m a_l$ for all $l,m=1,\cdots,n$ $$(P*Q)(\mathbf{ a})=P(\textbf{\em a})\cdot (Q({P}^{-1}(\textbf{\em a})\textbf {\em a}P(\textbf{\em a}))$$
% thus either $P(\textbf {\em a})=0$ or $(Q({P}^{-1}(\textbf{\em a})\textbf{\em a}P(\textbf{\em
% a}))=0$. Take now the two subsets of $\mathcal{V}_c(I),$ given
% by $$V_1=\mathcal{V}(I)\cap \mathcal{V}_c(\langle P\rangle)=\mathcal{V}_c(I+\langle P\rangle)\subseteq \mathcal{V}_c(\langle P\rangle)$$
% and
% $$ V_2=\mathcal{V}(I)\cap \mathbb{S}_{\mathcal{V}_c(\langle
% Q\rangle)}=
% \mathcal{V}_c{(I)}\cap\mathcal{V}_c(\langle
%{Q^s}\rangle)=\mathcal{V}_c(I+\langle Q^s\rangle)\subseteq
% \mathcal{V}_c(\langle {Q^s}\rangle)$$.
% Now either ${\bf a}$ belongs to $V_1$ or one of
% its conjugates belongs to $V_c(\langle Q\rangle)$, thus ${\bf a}$
% belongs to $V_2$; therefore $$\mathcal{V}_c(I)= V_1\cup {V_2}$$
% Since $\mathcal{V}_c(I)$ is irreducible either
% $$\mathcal{V}_c(I)=V_1\subseteq \mathcal{V}_c(\langle P\rangle)$$
% and thus $P$ vanishes on $\mathcal{V}_c(I)$ and hence $P\in \mathcal{J}(\mathcal{V}_c(I))=I$ (here we use the fact that $I$ is radical); or $\mathcal{V}_c(I)=V_2\subseteq \mathcal{V}_c(\langle {Q^s}\rangle)$ and hence $Q^s\in \mathcal{J}(\mathcal{V}_c(I))=I$.
% \end{proof}
%%\noindent As a corollary we have a refinement of Theorem \ref{symprime}
%% \begin{corollary} Let $I=\langle H \rangle$ be an ideal generated by an irreducible polynomial $H$ such that $\mathcal{V}_c(I)$ is irreducible. Then $I$ is quasi prime.
%%\end{corollary}
%%
%\noindent
% In the other direction we conjecture that\\\\
% { \bf{Conjecture 2}} \label{cpirr}Let $I$ be a quasi prime radical ideal such that, {\color{green} for every polynomial $P$, if $P^s\in I$ then $P^s\in \mathcal{A(I)}$}.
% Then $\mathcal{V}_c(I)$ (or simply $\mathcal{V}(I)$ ) is irreducible (and thus also $\mathcal{V}(I)$).\\\\
%\noindent Note that the assumption $P^s\in I$ then $P^s\in \mathcal{S}(I)$ is
% trivially satisfied for $I$ with a unique generator. \\
% Take $I$ quasi prime and radical (we still not know if quasi prime implies radical). We want to prove that $\mathcal{V}_c(I)$ is irreducible. Suppose, by contradiction, that $\mathcal{V}_c(I)$ is reducible that
% is can be written of the two non empty sets as follows $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2).$$ Take $H\in \mathcal{J}(\mathcal{V}_c(I_1))\setminus{I}$ and $K\in \mathcal{J}(\mathcal{V}_c(I_2))$.
% Consider $H*K$, and take a point $\textbf{\em a}\in \mathcal{V}_c(I)$. There are two possibilities: $\textbf{\em a}\in \mathcal{V}_c(I_1)$ or $\textbf{\em a}\in \mathcal{V}_c(I_2)$. In the first case $$H*K(\textbf{\em a})=H(\textbf{\em a})K(H^{-1}(\textbf{\em a})\textbf{\em a}H(\textbf{\em a}))=0.$$
% Otherwise $\textbf{\em a}\in \mathcal{V}_c(I_2)$.\\
% If $\mathcal{V}_c(I_2)=\mathcal{V}_c( \mathcal{S}(I_2))$ then $\mathbb{S}_{\textbf{\em a}}\subseteq \mathcal{V}_c(I_2)$, again $$H*K(\textbf{\em a})=H(\textbf{\em a})K(H^{-1}(\textbf{\em a})\textbf{\em a}H(\textbf{\em a}))=0.$$
% We have that $H*K\in \mathcal{J}(\mathcal{V}_c(I))=\sqrt{I}=I$. We then use the fact that $I$ is quasi prime, and hence $H\in I$ (a contradiction) or $K^s\in I$.
% That is $\forall K\in I_2\subseteq \mathcal{J}(\mathcal{V}_c(I_2))$ we have $K^s\in I$, and thus $$\mathcal{V}_c( \mathcal{S}(I_2)) =\bigcap_{K\in I_2}\mathbb{S}_{{\mathcal{V}}_c(K)}=\bigcap_{K\in I_2}{\mathcal{V}_c(K^s)}\supseteq \mathcal{V}_c(I)\supseteq \mathcal{V}_c(I_2).$$
% Hence if $\mathcal{V}_c(I_2)=\mathcal{V}_c( \mathcal{S}(I_2)),$ we get that $\mathcal{V}_c(I)=\mathcal{V}_c(I_2)$.
% As a first result we thus have that
% \begin{theorem} \label{nosfere}Let $I$ be a quasi prime radical ideal
% then $\mathcal{V}_c(I)$ cannot split as the union of two sets $\mathcal{V}_c(I_1)\cup\mathcal{V}_c(I_2)$ in which at least one is an algebraic set of the form $\mathcal{V}_c(J)=\mathcal{V}_c( \mathcal{S}(J)),$ for a certain ideal $J$. {\color{red} We have proven more, indeed whenever $I$ is a quasi prime radical ideal such that $\mathcal{V}_c(I)$ has a summand with spherical simmetry we get that $\mathcal{V}_c(I)=\mathcal{V}_c(I_2)$ has spherical simmetry i.e. $\mathcal{V}_c( \mathcal{S}(I_2))=\mathcal{V}_c(I)$}
% \end{theorem}
% \noindent{\color{red} In particular, under the hypotheses above, if $I$ is generated by a single polynomial $L$, from $\mathcal{V}_c(I)=\mathcal{V}_c(I_2)$ we get that $I_2=I$, in fact $I=\mathcal{J}(\mathcal{V}_c(I))=\mathcal{J}(\mathcal{V}_c(I_2))=\sqrt{I_2}\supseteq I_2$. Since $I$ is principal we get also that $I=I_2$, since we can take $L$ has the generator of $I_2$, so $I_2$ is radical and quasi prime and after repeating the same argument we conclude that $\mathcal{V}_c(I_2)=\mathcal{V}_c(I)$ is irreducible.\\ \\If moreover the generator $L$ has real coefficients, we get that $\mathcal{V}_c(I)=\mathcal{V}_c( \mathcal{S}(I))$. If it is true that $\mathcal{V}_c( \mathcal{S}(I))$ is the union of arranged spheres (maybe it follows from equation \eqref{unisfere}) we can argue as above and conclude that $\mathcal{V}_c(I)$ is irreducible. We have thus proved the following
% \begin{proposition} Let $I=\langle L\rangle$ be a quasi prime radical ideal generated by a polynomial with real coefficients then $\mathcal{V}_c(I)$ is irreducible.\end{proposition} CONTROLLARE SE NON risulta in contrasto con quanto detto in irred}\\
% \noindent {\bf{DOMANDA}}Se $L$ ha coefficienti reali l'ideale generato da lui risulta quasi primo? Risulta radicale?\\\\
% \noindent As a consequence of Theorem \ref{nosfere} we also have a way to establish that, for example, the ideal $I=\langle q_1^2+1,q_2^2+1\rangle$ is not quasi prime. Indeed $\mathcal{V}(I)=\mathbb{S}\times \mathbb{S}$ is irreducible while $\mathcal{V}_c(I)=\mathbb{S}_{(i,i)}\cup\mathbb{S}_{(-i,i)}$ is reducible. Thanks to Nullstellensatz we have that $I$ is radical hence, by the previous theorem $I$ is not quasi prime. \begin{remark} The ideal $I$ generated by the polynomial $q_1q_2+1$ is completely prime (because the generator is irreducible) and thus radical and quasi prime.
% This is not a counterexample of the previous theorem, indeed $\mathcal{V}_c(I)=\cup_{a\in \mathbb{H}}\mathbb{S}_{(a,-a^{-1})}$ cannot be written as the union of two sets of the form $\mathcal{V}_c(I_1)\cup\mathcal{V}_c(I_2)$ in which one of this is of the form $\mathcal{V}_c(J)=\mathcal{V}_c( \mathcal{S}(J)),$ for a certain ideal $J$.
%\end{remark}
%The assumption that $\mathcal{V}_c(I_2)=\mathcal{V}_c( \mathcal{S}(I_2)),$ was sufficient to get a contradiction, we now have to consider also the case in which DA FINIRE! COME ESCLUDERE I PUNTI PIATTI? Abbiamo esempi con punti piatti in cui $\mathcal{V}(I)=\mathbb{S}\times {i}$ irriducibile, $\mathcal{V}_c(I)$ riducibile nei due pezzi piatti $(i,i)$ e $(-i,i)$ in questo caso $I=\langle q_1^2+1,q_2-i\rangle$. Nota che $ \mathcal{S}(I)=\langle (q_1^2+1)^2,(q_2-i)^s, (q_1^2+1)*(q_2-i)\rangle$ so $\mathcal{V}( \mathcal{S}(I))=\mathbb{S}\times\mathbb{S}$ and $\mathcal{V}_c( \mathcal{S}(I))= \mathbb{S}_{(i,i)}\cup\mathbb{S}_{(-i,i)}$. Non sappiamo ancora se risulta essere quasi primo (per il Nullstellensatz risulta radicale), non risulta completamente primo.\\\\
%
%
%\end{corollary}\begin{corollary} Let $I=\langle H \rangle$ be an ideal generated by an irreducible polynomial $H$ such that $\mathcal{V}_c(I)$ is irreducible. Then $I$ is a quasi prime ideal.
%\begin{corollary}[of Corollary \ref{coro}] {Let $I=\langle H \rangle$ be a quasi prime, radical ideal with spherical symmetry. Then
%$\mathcal{V}_c(I)$ is irreducible.}
%\end{corollary}
%{\color{red}\noindent A consequence of Corollary \ref{coro} is the following.
%\begin{corollary}{
%Let $I=\langle H \rangle$ be an ideal generated by an irreducible polynomial $H\in\mathbb{R}[q_1,\ldots,q_n]$. Then
%$\mathcal{V}_c(I)$ is irreducible.}
%\end{corollary}
%}
%\noindent Combining Corollary \ref{coro} and Proposition \ref{Hsquasiprime} we get
%\begin{corollary}\label{Hsirred}
%Let $I$ be a principal right ideal generated by an irreducible polynomial $H\in\mathbb{H}[q_1,\ldots,q_n]$. Then
%$\mathcal{V}_c(\langle H^s \rangle)$ is irreducible.
%\end{corollary}
%%Resta ancora da dimostrare che se $H$ è irriduile, allora $\mathcal{V}_c(\langle H\rangle)$ è irriducibile.
%
%As a consequence, we can prove
% \begin{theorem}\label{Hirr}
% If $I$ is a principal right ideal generated by an irreducible polynomial $H$, then $\mathcal{V}_c(I)$ is irreducible.
%\end{theorem}
%
%\begin{proof}
%Thanks to Proposition \ref{irrcp} and Corollary \ref{Hsirred} we can apply Theorem \ref{cpirred} and conclude.
%\end{proof}
%{\color{blue}
%\begin{proposition} Let $I=\langle H\rangle$ be a quasi prime ideal. Then the generator $H$ is irreducible or the symmetrization of an irreducible polynomial (and hence $I$ is radical). \end{proposition}
%\begin{proof} Suppose that $H=P*Q$, with $Q$ a non constant irreducible factor, and $P$ a non constant polynomial.
% Since $I$ is quasi prime, either $P$ or $Q^s$ belong to $I$. If $P\in I$, then
% $\langle P \rangle \subset I=\langle P*Q \rangle \subset \langle P
% \rangle$, hence $I=\langle P\rangle$ and $Q$ is a constant, a contradiction. % we can take it as a generator of $I$ with lower degree
% Otherwise, if $Q^s\in
% I$, we
% have $$Q^s=P*Q*R$$ with $R \in \mathbb H[q_1,\ldots,q_n]$. Then, proceeding as in the proof of Proposition \ref{irrcp}, taking into account the irreducibility of $Q$, we get that $P$ is a multiple of $Q^c$ and thus
%$H=Q^s$ up to a multiplicative constant.
%\end{proof}}
{
\noindent For principal ideals, the following theorem provides equivalent conditions for the irreducibility of the corresponding {slice} algebraic set.
\begin{theorem}\label{principal}
Consider $I=\langle H \rangle$ with
$H\in \mathbb H[q_1,\ldots q_n]$. The following are equivalent:
\begin{enumerate}
\item the ideal $I$ is quasi prime;
\item $H$ is an irreducible polynomial or $H=Q^s$ with $Q$
%the symmetrization of
an irreducible polynomial $Q\neq Q^c$;
\item $I$ is a radical ideal and $\mathcal V_c(I)$ is an irreducible slice algebraic set.
\end{enumerate}
\end{theorem}
\begin{proof}
\noindent{ \bf $(1) \Rightarrow (2)$} Suppose that $H=P*Q$, with $Q$ a non-constant irreducible factor, and $P$ a non-constant polynomial.
Since $I$ is quasi prime, either $P$ or $Q^s$ belong to $I$. If $P\in I$, then
$\langle P \rangle \subset I=\langle P*Q \rangle \subset \langle P
\rangle$, hence $I=\langle P\rangle$ and $Q$ is a constant, a contradiction. % we can take it as a generator of $I$ with lower degree
Otherwise, if $Q^s\in
I$, we
have $$Q^s=P*Q*R$$ with $R \in \mathbb H[q_1,\ldots,q_n]$. Then, proceeding as in the proof of Proposition \ref{irrcp}, taking into account the irreducibility of $Q$, we get that $P$ is a multiple of $Q^c$ and thus
$H=Q^s$ up to a multiplicative constant.\\
\noindent{ \bf $(2) \Rightarrow (3)$}
Assume that
$I$ is a principal ideal generated by an irreducible polynomial $H\in\mathbb{H}[q_1,\ldots,q_n]$. Then, from Proposition \ref{irrcp}, it follows that $I$ is completely prime and hence {also} radical.
Combining Corollary \ref{coro} and Proposition \ref{Hsquasiprime} we get
$\mathcal{V}_c(\langle H^s \rangle)$ is irreducible.
%Follows from Corollary \ref{Hsirred} and Theorem \ref{Hirr}.
We now apply Theorem \ref{cpirred} and conclude that $\mathcal{V}_c(I)$ is irreducible.
If $I$ is a principal ideal generated by $H=Q^s$, with
$Q$ an irreducible polynomial such that $Q\neq Q^c$, then $I$ is quasi prime and radical, thanks to
Proposition \ref{Hsquasiprime}. Since $\mathcal{V}_c(\langle Q^s \rangle)$ has spherical symmetry, from Corollary \ref{coro}, we get that $\mathcal{V}_c(I)$ is irreducible.
\\
\noindent{ \bf $(3) \Rightarrow (1)$} This follows directly from Theorem \ref{VirredIprime}.
\end{proof}
\noindent {{Thanks to the previous result we can exhibit an explicit example of an irreducible slice algebraic set which is the union of an infinite number of spherical sets. %containing Let us conclude with an example.
\begin{example} Consider the right ideal $I$ generated by the polynomial $P(q_1,q_2)=q_1q_2+1$. Since $P$ is irreducible, we have that %$\mathcal V_c(I)$ is irreducible and
% $\mathcal{V}_c(I)$ has spherical symmetry. Corollary \ref{coro} implies then that
$\mathcal{V}_c(I)=\bigcup_{a\in \mathbb{H}}\mathbb{S}_{(a,-a^{-1})}$ is irreducible.
\end{example}}
}
}
%{\color{blue}
% \noindent As a consequence we have
%\begin{corollary} Le $I$ be a quasi prime and radical principal ideal then $\mathcal{V}_c(I)$ is irreducible.
%\end{corollary}
%\begin{proof} From the previous proposition, we get that the generator $H$ of $I$ is irreducible or the symmetrisation of an irreducible polynomial.
%In the first case the proof follows from Theorem \ref{Hirr}, in the second case from Corollary \ref{Hsirred}.
%\end{proof}
%\noindent In the other direction we can apply Theorem \ref{VirredIprime} and therefore we %have the following characterisation of quasi prime principal radical ideals
%\begin{theorem} A principal right ideal $I$ is quasi prime if and only if $I$ is radical and $%\mathcal{V}_c(I)$ is irreducible.
%\end{theorem}
%{\color{red}
%\begin{proof}
%Suppose by contradiction that $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2)$$ is reducible. Then
%\[\mathcal{V}_c(\langle H^s \rangle)=\mathbb S_{\mathcal{V}_c(I)}=\mathbb S_{\mathcal{V}_c(I_1)}\cup \mathbb S_{\mathcal{V}_c(I_2)}.\]
%Corollary \ref{Hsirred} then implies that, either $\mathbb S_{\mathcal{V}_c(I_1)}\subseteq\mathbb S_{\mathcal{V}_c(I_2)}$ or $\mathbb S_{\mathcal{V}_c(I_1)}\supseteq\mathbb S_{\mathcal{V}_c(I_2)}.$ Therefore,
%for any ${\textbf a} \in \mathcal{V}_c(I_1) \setminus \mathcal{V}_c(I_2)$ there exists ${\bf \tilde a} \in \mathcal{V}_c(I_2)\cap \mathbb S_{\textbf a}$. Therefore $H$ vanishes on the entire spherical set $\mathbb{S}_{\bf a}$ and hence $\mathbb S_{\bf a}$ is contained in $\mathcal{V}_c(I_1)$.
%The same holds for any point in $\mathcal V_c(I_2)\setminus \mathcal V_c(I_1)$. Thus
%\[\mathcal{V}_c(I)=\big((\mathcal{V}_c(I_1)\setminus \mathcal{V}_c(I_2))\cup( \mathcal{V}_c(I_2)\setminus \mathcal{V}_c(I_1))\big) \cup \big( \mathcal{V}_c(I_1)\cap \mathcal{V}_c(I_2)\big)=\\
%\big(\mathbb{S}_{(\mathcal{V}_c(I_1)\setminus \mathcal{V}_c(I_2))\cup( \mathcal{V}_c(I_2)\setminus \mathcal{V}_c(I_1))}\big) \cup \big( \mathcal{V}_c(I_1)\cap \mathcal{V}_c(I_2)\big)\]
%Recalling Proposition \ref{corr},
%\[\mathcal{V}_c(I_1)\cap \mathcal{V}_c(I_2)=\mathcal{V}_c(I_1+I_2).\]
%Moreover
%$Z_c:=(\mathcal{V}_c(I_1)\setminus \mathcal{V}_c(I_2))\cup( \mathcal{V}_c(I_2)\setminus \mathcal{V}_c(I_1))$ is a subset containing only points with commuting coordinates, which coincides with its symmetrization $\mathbb S_{Z_c}$ which is an algebraic set with spherical symmetry thanks to Propositions \ref{essealg} and \ref{spherical}. This is in contradiction with Theorem \ref{nosfere}.
%\end{proof}
%
%Maybe the proof can be generalized for any $I$ quasi prime and radical ideal such that $\mathbb S_{ V_c(I)}$ is irreducible. \\We would like to prove that $\mathbb S_{ V_c(I)}$ is always irreducible when $I$ is quasi prime and radical. This would follow directly, applying Theorem \ref{nosfere}, if we were able to show that $J=\mathcal{J}(\mathbb{S}_{\mathcal{V}_c(I)})$ is quasi prime (indeed it is always radical).
%}
%
%
%\begin{proposition}
% If $I=\langle H\rangle$ is an ideal generated by a single irreducible monic polynomial $H$, then $\mathcal{V}_c(I)$ is irreducible.
%\end{proposition}
%\begin{proof}
%Suppose, by contradiction, that $\mathcal{V}_c(I)$ is reducible that
% is it can be written as the union of two non empty sets as follows $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2).$$ If this is the case, thanks to Proposition \ref{idirred}, $I=\sqrt{I_1}\cap\sqrt{I_2}$ is reducible.
% Then,
% for all $K\in \mathcal{J}{(V_c(I_1))}\setminus{I}$ and $L\in \mathcal{J}{(V_c(I_2))}$ we get $K*L^s\in \sqrt{I_1}* \mathcal{S}(\sqrt{I_2})\subseteq \sqrt{I_1}\cap\sqrt{I_2}=I$
%%(thanks to \eqref {xcap})
%and, since $I$ is completely prime and $K\notin I,$ we get $L^s\in I=\sqrt{I_1}\cap\sqrt{I_2}.$ That is
% $$\text{for any } \ L\in\sqrt{I_2}, \ \text{ its symmetrization } \ L^s\in \sqrt{I_1}.$$
% \noindent In the same way
% $$\text{for any } \ K\in\sqrt{I_1}, \, \text{ its symmetrization } \ K^s\in \sqrt{I_2}.$$
% In other terms we get that $ \mathcal{S}(\sqrt{I_\ell})\subseteq I$ for $\ell=1,2$.
% {\color{red} fino qui vale con sola ipotesi completamente primo}
%\\
% We can prove more, i.e that, for $\ell=1,2,$ $ \mathcal{S}(\sqrt{I_\ell})\subseteq \langle H^s\rangle.$ In fact, for any $K\in \sqrt{I_1},$ there exists a polynomial $Q_1\in\mathbb{H}[q_1,\ldots,q_n]$ such that $K^s=H*Q_1$. Thanks to the irreducibility of $H$ we have that $Q_1=H^c*Q_2$, that is $K^s$ is a multiple of $H^s$; the same holds starting from $L$ in $\sqrt{I_2},$ hence we have $$ \mathcal{S}(\sqrt{I_\ell})\subseteq \langle H^s\rangle\;\;\;\rm{for\;\;\ell=1,2}$$
% On the other side $H\in \sqrt{I_\ell}$, for $\ell=1,2.$ Thus $$\langle H^s\rangle \subseteq \mathcal{S}(\sqrt{I_\ell}) \;\;\;\rm{for\;\;\ell=1,2}$$ and hence
% {$$ {\langle H^s\rangle= \mathcal{S}(\sqrt{I_1})= \mathcal{S}(\sqrt{I_2})}.$$}
% {\noindent Now take $K\in \sqrt{I_1}\setminus I.$ Since $H$ is monic we can divide $K$ by $H$. $$K=H*K_1+R$$ with $\deg R<\deg H.$ Recall that $H\in \sqrt{I_1}$ hence $K-H*K_1\in \sqrt{I_1}.$ Thus $$R^s={(K-H*K_1)}^s\in \mathcal{S}(\sqrt{I_1})=\langle H^s\rangle$$ but this is a contradiction since $\deg R^s$ is strictly lower than $2\deg H.$\\}
% Take $K\in \sqrt{I_1}\setminus I$. Consider $$(H+K)^s=(H+K)*(H+K)^c=H^s+K^s+H*K^c+K*H^c\in \mathcal{S}(\sqrt{I_1})=\langle H^s\rangle\subseteq I$$ since $H,K\in \sqrt{I_1}$. Thus, since $K^s$ and $H*K^c\in I,$ we get that $K*H^c\in I$ as well. Now observe that{\color{blue} SBAGLIATA applicazione completely prime $H^c*I\subseteq I$ and $K\notin I$,} recalling that $I$ is completely prime we have that $H^c\in I$ which implies that $H$ has real coefficients. That is $\mathcal{V}_c(I)$ has spherical symmetry and is therefore irreducible.
% \end{proof}
%\noindent{\color{blue} As a consequence of the representation formula, with the same proof of the one variable setting, we can prove that if $P$ is a polynomial with coefficients in $\mathbb{C}_J$ then $\mathcal{V}_c(\langle P\rangle)$ is contained in the product of $(\mathbb{C}_J)^n$ or, if it contains a point not in $(\mathbb{C}_J)^n$, it contains an entire sphere. The second possibility does hold, for example consider $P(q-1,q_2,q_3)=q_1q_2q_3+q_12i+2i$ that vanishes also in $(j,j,0)$ not in $\mathbb{C}_i$.
%%f we prove that we deduce that we can argue for $\mathcal{V}_c(\langle P\rangle)$ as in the complex setting.}
%We can prove that
%\begin{proposition} If $L$ is a reducible polynomial, $L=L_1*L_2\in\mathbb{H}[q_1,\ldots,q_n]$, and
%$Z_c(P)$ represents the set of zeroes of $P$ with commuting components, then
% $$Z_c(L_1)\cup Z_c(L_2^s)\supseteq Z_c(L)\supseteq Z_c(L_1)\cup Z_c(L_2);$$
%\end{proposition}
%\noindent As a corollary we get
%\begin{corollary} If $L$ is a reducible polynomial and $L=L_1*L_2=L_2*L_1$ then $$(Z_c(L_1)\cup Z_c(L_2^s))\cap (Z_c(L_2)\cup Z_c(L_1^s))\supseteq Z_c(L)\supseteq Z_c(L_1)\cup Z_c(L_2);$$
%\end{corollary}
%\noindent and in particular
%\begin{corollary} If one of the factors of $L$, say $L_2$, has real coefficients then $Z_c(L_2)=Z_c(L_2^s)$ and thus $$\mathcal{V}_c(I)=Z_c(L)= Z_c(L_1)\cup Z_c(L_2)=\mathcal{V}_c(\langle L_1\rangle )\cup \mathcal{V}_c(\langle L_2\rangle );$$ i.e. $\mathcal{V}_c(I)$ is reducible, whenever $\mathcal{V}_c(\langle L_1\rangle )\nsubseteq \mathcal{V}_c(\langle L_2\rangle )$ and $\mathcal{V}_c(\langle L_1\rangle )\nsupseteq \mathcal{V}_c(\langle L_2\rangle ).$
%\end{corollary}
%
%}
%\section{Examples}
%
% \begin{remark} With the following example we want to clarify why ``slice reducibility" has not a geometricvery .
% The fact that this first example is reducible seems reasonable, but it is not if considering its intersection with ``leaves", i.e. with $\mathbb{C}_I^+=\{x+Iy\; , y>0\}$
%or $\mathbb{C}_I^-=\{x+Iy\;, y<0\}$ where $I\in\mathbb{S}$. %% (definizione di Fabio).
%\end{remark}
%\begin{example} $V=\mathbb{S}_a\times \mathbb{S}_b$
%is irreducible and $V=\mathcal{V}(I)$ where $I=\langle(q_1-a)^s,(q_2-b)^s\rangle$.
%\end{example}
%\begin{proposition}\label{GeneratoriSfere}
%Let $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}\in\mathbb{H}^n$ be an arranged spherical set, and suppose $a_\ell \not \in \mathbb{R}$ for any $\ell$. Then $\mathbb S_{(a_1,a_2,a_3,\ldots, a_n)}=\mathcal{V}(I)$ where
%\[I=\langle P_1,\ldots, P_{n-1}, (q_n-a_n)^s \rangle \] where $P_\ell$ are the polynomials with real coefficients
%\[P_{\ell}(q_{\ell},q_{\ell+1})=q_{\ell}+q_{\ell +1} \gamma_{\ell} +\delta_{\ell}\]
%with
%\[\gamma_{\ell}=\pm\dfrac{|{\rm Im} (a_{\ell})|}{|{\rm Im} ( a_{\ell +1})|}, \quad \text{and} \quad \delta_{\ell}=-{\rm Re} (a_{\ell})\mp\dfrac{{\rm Re} (a_{\ell+1})|{\rm Im} (a_{\ell})|}{|{\rm Im} (a_{\ell +1})|}, \]
%for any $\ell=1,\ldots,n-1$, where the signs are determined by the sign of the product ${\rm Im}(a_{\ell}){\rm Im }(a_{\ell+1})$.
%\end{proposition}
%\begin{proof}
%By direct computation, each $P_\ell$ vanishes on $(a_\ell,a_{\ell+1})$ and hence, considered as a polynomial in $\mathbb H[q_1,\ldots,q_n]$, also on $(a_1,\ldots,a_n)$. Since they have real coefficients, they all vanish on $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}$ (see lemma \ref{zerisferici}).
%Consider $Q$ vanishing on ${\mathbb S_{(a_1,a_2,a_3,\ldots, a_n)}}$. Performing subsequent divisions by the monic polynomials $P_\ell$, we can write
%\[Q(q_1,\ldots,q_n)=P_1*Q_1(q_1,\ldots,q_n)+P_2*Q_2(q_2,\ldots,q_n)+\cdots+P_{n-1}*Q_{n-1}(q_{n-1},q_n)+R(q_n).\]
%Since $Q,P_1,\ldots,P_{n-1}$ vanish on $\mathbb S_{(a_1,a_2,a_3,\ldots, a_n)}$ and $R$ depends only on the variable $q_n$, recalling the one-variable theory (see Proposition 3.18 in \cite{libroGSS}), we get that $R(q_n)$ vanishes on $\mathbb S_{a_n}$, and thus is of the form $R(q_n)=(q_n-a_n)^s*R_1(q_n)$.
%\end{proof}\noindent We can prove that the polynomials $P_{\ell}$ are irreducible and $\mathbb{S}_{(a_1,a_2,\cdots,a_n)}$ is irreducible. We consider the example in dimension $2$, the same argument holds in any dimension for arranged spheres.
% Let $V(I)=\mathbb{S}_{(a,b)}$ where $(a,b)$ are such that $ab=ba$, where $$I=\langle P(q_1,q_2),(q_2-b)^{s}\rangle$$ with $P(q_1,q_2)$
%a polynomial of the form $P(q_1,q_2)=q_1+q_2\alpha+\beta$ and $\alpha,\beta\in \mathbb{R}.$
%%Note that a similar polynomial vanishes on $\mathbb{S}_{(a,b)}$. (see Proposition \ref{GeneratoriSfere}) Starting from a generic polynomial that vanishes on $\mathbb{S}_{(a,b)}$ and then dividing it by $P$, the rest $R$
%%has degree less than $1$ in $q_1,$ that is it depends only on $q_2$, i.e. $R=R(q_2)$. This vanishes on the sphere of $b$ i.e. it is a multiple of $(q_2-b)^s.$\\
% Note that $P(q_1,q_2)$ is {\bf irreducible}.
%% We can assume that $\alpha=1$. Suppose that there exist $c,d\in \mathbb{H}$ such that $q_1q_2+q_1\beta+q_2\gamma+\delta=(q_1-c)*(q_2-d)$ we get that $d=-\beta$ and $c=-\gamma$ and thus $c$ and $d$ are both real, this implies that the *-product is pointwise. But $(q_1-c)\cdot(q_2-d)$ cannot vanish on the spherical set, a contradiction. \\
%%
%Moreover it is impossible to find two ideals $I_1$ and $I_2$ such that $\mathcal{V}(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2)$, that is $V$ is {\em irreducible}. Indeed suppose, by contradiction, that there exist at least two points $(a',b')$ and $(a'',b'')\in \mathcal{V}(I_1)\subset \mathcal{V}(I)$ thus, since they belong to the same spherical set, the entire sphere through $(a',b')$ belongs to $\mathcal{V}(I_1)$, thus $\mathcal{V}(I)=\mathbb{S}_{(a,b)}=\mathbb{S}_{(a',b')}\subset \mathcal{V}(I_1)$. \\
% This procedure can be applied to a spherical set $S_{(a_1,a_2,\cdots,a_n)}\in\mathbb{H}^n$ taking as generators of $I$ the polynomials with real coefficients introduced in Proposition \ref{GeneratoriSfere} $P_{\ell}(q_{\ell},q_{{\ell}+1})$ (with $\ell:1\cdots ,n-1)$ and $(q_n-a_n)^s$.
% We have thus proved that
%\begin{example}\label{radnoprime} The ideal $I$ in $\mathbb{H}[q]$ generated by the polynomial $q^2+1$ is radical. Indeed $$\sqrt{I}=\mathcal{J}(\mathcal{V}(I))=\mathcal{J}(\mathbb{S}_i)=I.$$ This is an example of a radical ideal which is not completely prime.
%\end{example}
%
%
%
%\section{Open problems}
%\begin{enumerate}
% \item Is there a more constructive definition of the radical of an ideal? Yes: Aryapoor. In particular, if $I$ is two-sided, then the description of the radical is the same of the complex case.
%For example, if $\mathcal V_c(I)=\mathcal V_c(\mathcal S(I))$, then $\sqrt{I}$ is two-sided.
%
% \item If $I$ is quasi prime, then is $\sqrt{I}$ quasi prime? And $\sqrt{\mathcal S(\sqrt{I})}$?
% \item If $I$ is quasi prime, then it is radical?
% \item We know that $I$ quasi prime, does not imply $I$ irreducible (see Example \ref{qp_not_irred}). Is the converse true ($I$ irreducible $\implies$ $I$ quasi prime)? \\
% Is the equivalence true when $I$ is a completely prime ideal?
%%\item Can we say something more for principal ideals? Is it true that the variety associated with a principal ideal generated by an irreducile polynomial is irreducible?
%%\item Are $\mathbb S_{\mathcal V_c(I)}$ always slice algebraic sets?
%%Or, even more, there exists an ideal $I'$ such that $\mathbb S_{\mathcal V_c(I)}=\mathcal V_c(I')$?
%%
%% Yes, because $\mathbb S_{\mathcal V_c(I)}$ contains only points with commuting components.
%% Thus $\mathcal J(\mathbb S_{\mathcal V_c(I)})=:I'$ is an ideal (by definition) and coincides with the set of polynomials vanishingn on $\mathbb S_{\mathcal V_c(I)}$ (thanks to the expression of the $*$-product on points with commuting components).
%% Hence $\mathbb S_{\mathcal V_c(I)}=\mathcal V (I')$. NO
%\item Controllare ma forse si (vedi le ultime cose in rosso).In the principal case, does it hold the equivalence $I$ quasi prime and radical $\iff \mathcal V_c(I)$ irreducible?
%$\Leftarrow$ holds.
%
%To prove the equivalence it suffices to show that $\langle H\rangle$ quasi prime and radical $\implies \sqrt{\langle H^s \rangle}$ is quasi prime or $ \mathcal V_c(\langle H^s \rangle)=\mathbb S_{\mathcal V_c(\langle H\rangle )}$ is irreducible.
%\item {\color{red} Vale in generale: If $I$ is a completely prime/quasi prime right ideal then $\mathcal J(\mathbb S_{\mathcal V_c(I)})=\sqrt{\mathcal S(I)}$ is quasi prime? }
%
%\end{enumerate}
%
%\begin{center}CONTROLLATO FINO QUI
%\end{center}
%
% Before proving
% Conjecture 2 we need a Lemma
% \noindent The one-to-one correspondence holds also for zeros with non commuting components.
% [$I^s$ non risulta ben definito, comunque non e' un ideale]
% \begin{lemma} \label{unionesfere}
% Taking $I$ an ideal $$\mathcal{V}_c(I^s)=\bigcup_{a\in
% \mathcal{V}_c(I)} \mathbb{S}_{\sigma_{\ell}(a)}$$ where
% $\ell=1,\ldots, 2^{n-1}$, and $\sigma_{\ell}(a)$ denotes all the
% elements in $\mathbb{H}^n$ of the type
% $$(a_1,g_2^{-1}a_2g_2,\cdots,g_n^{-1}a_ng_n)$$ where $g_j\in
% \mathbb{S},$ is such that $$g_j^{-1}a_jg_j,\cdot a_1=a_1\cdot
% g_j^{-1}a_jg_j$$ for all $j=2,\ldots,n$.
% \end{lemma}
% \begin{proof} Observe that the condition $g_j^{-1}a_jg_j,\cdot a_1=a_1\cdot g_j^{-1}a_jg_j$ if $a_1\in \mathbb{C}_i$ implies that $g_j$ is equal to $\pm i$ or it is an element in the sphere $\mathbb{S}$ orthogonal to $i.$
% \end{proof}
% \begin{proof}[Dimostrazione teorema \ref{cpirr}]
% Suppose by contradiction that $\mathcal{V}_c(I)$ is reducible that
% is $$\mathcal{V}_c(I)=\mathcal{V}_c(I_1)\cup \mathcal{V}_c(I_2).$$
% Take $I_1=\langle H_1,H_2,\ldots H_{\ell}\rangle$ and $I_2=\langle
% K_1,K_2,\ldots K_r\rangle.$ Suppose $I_1\not\subseteq I$, that is
% there exists $H_j\in I_1\setminus I$. Take $H_j*K_m^s$ for all
% $m=1,\ldots, r$. If $\textbf{\em a}\in \mathcal{V}_c(I)$, then $\textbf{\em a}$ belongs
% either to $\mathcal{V}_c(I_1)$ or to $\mathcal{V}_c(I_2).$ Now
% recall that $H_j*K_m^s(\textbf{\em a})$ is zero, if $H_j(\textbf{\em a})=0,$ or if
% $K_m^s(H_j^{-1}(\textbf{\em a})\textbf{\em a} H_j(\textbf{\em a}))=0.$ Thus if $\textbf{\em a}\in \mathcal{V}_c(I_1)$
% then $H_j(\textbf{\em a})=0$ and $H_j*K_m^s(\textbf{\em a})=0.$ If, instead $\textbf{\em a}\in
% \mathcal{V}_c(I_2)$, then for every $m=1,\ldots, r$ it turns out that $K_m(\textbf{\em a})=0,$ thus by
% Proposition \ref{corrsimm}, $K_m^s(\mathbb{S}_a)=0,$ and again
% $H_j*K_m^s(\textbf{\em a})=0.$ That is $$H_j*K_m^s \in
% \mathcal{I}(\mathcal{V}_c(I))=\sqrt{I}=I.$$ Indeed $I$ is
% completely prime and therefore radical. Now note that
% $$H_j*K_m^s\in I$$
% $$K^s_m*I\subseteq I$$ therefore $H_j\in I$ a contradiction, or
% $K_m^s\in I$ and, by hypothesis $K_m^s\in \mathcal{S}{(I)}.$ Thus
% $ \mathcal{S}(I_2)\subseteq \mathcal{S}(I)$ and $$\mathcal{V}_c( \mathcal{S}(I))\subseteq \mathcal{V}_c( \mathcal{S}(I_2)).$$
% Using Lemma \ref{unionesfere} we argue that the irreducibility of
% $\mathcal{V}_c(I^s)$ descends from the irreducibility of
% $\mathcal{V}_c(I))$ that is $\mathcal{V}_c(I_1^s)=\varnothing$
% implies $\mathcal{V}_c(I_1)=\varnothing$ and we get the claim.
% \end{proof}
% {\em Note that case $(2)$ can occur, that is the set
% $\mathcal{V}_c(I)$ is reducible, while $\mathcal{V}(I)$ is
% irreducible, for example, whenever $\mathcal{V}(I)$ is a balloon
% with a
% tail $$\mathbb{S}_{(\tilde{a}_1,\tilde{a}_2,\cdots,\tilde{a}_k)}\times{(a_{k+1},\cdots,a_n)}$$
% with $\tilde{a}_i\tilde{a}_j=\tilde{a}_j\tilde{a}_i$ (for
% $i,j:1,\cdots, k$) and $a_ia_j=a_ja_i$ (for $i,j:k+1,\cdots,
% n$). Indeed in this case $\mathcal{V}_c(I)$ splits as the union of
% the disjoint sets containing the two conjugate shadows of
% $(\tilde{a}_1,\tilde{a}_2,\cdots,\tilde{a}_k)$ in the same slice of
% $(a_{k+1},\cdots,a_n)$.We have that
% $\mathbb{S}_{(\tilde{a}_1,\tilde{a}_2,\cdots,\tilde{a}_k)}\times{(a_{k+1},\cdots,a_n)}=\mathcal{V}(J)$
% for an ideal $J$. Starting from a generic polynomial $p$ that
% generates $J$, after subsequent quotients of $p$ by $(q_i-a_{i})$
% with $i:k+1\cdots n$ we get a rest $R$ that vanishes on the
% spherical set and that does not depend by $q_i$ with $i:k+1\cdots
% n$ (indeed it has degree less than $1$ in each of these
% coordinates), $p$ is therefore generated by the polynomials in
% example \ref{sferacinese1}, which are all irreducible except the
% last one which is the symmetrisation of $(q_k-a_k).$ \\ Note that
% in this case the Conjecture holds and that it {\bf is not possible}
% to find a unique generator $h$ for the ideal
% $J$.\\ {\color{blue}The same situation occurs when
% $\mathcal{V}(I)=\mathbb{S}\times\mathbb{S}$, in this case
% $\mathcal{V}_c(I)=\mathbb{S}_{(i,i)}\cup \mathbb{S}_{(i,-i)}$}}
%\begin{enumerate}
%\item Case (2) can occur if $I$ is generated by a unique polynomial?
%\end{enumerate}
%
%% Forse. Da qui seguirebbe che i generatori del $\sqrt{I}$ li devo cercare tra quelli di $\sqrt{J}=J$ che sono con le propriet\`a richieste, ovvero irriducibili o simmetrizzati di irriducibili}
%\noindent {\bf{Conjecture 3}} An algebraic set $\mathcal{V}(I)$ is reducible as $$\mathcal{V}(I)=\mathcal{V}(I_1)\cup \mathcal{V}(I_2)$$ if and only if $I$ admits as generators the $*$-product of generators of $I_1$ and $I_2$.
\begin{thebibliography}{99}
%\bibitem{Nul1} A. Gori, G. Sarfatti, F. Vlacci
%\bibitem{cileni} A. E. Almendras Valdebenito--J. A. Briones Donoso--A. L. Tironi ``Resultants of skew polynomials over division rings'',
% {\tt arXiv:2109.13092v1}, (2021)
\bibitem{israeliani-1} G. Alon--E. Paran ``A central quaternionic Nullstellensatz'', {\sc J. Algebra}, {\bf 574}, p. 252--261, (2021)
\bibitem{israeliani-3} G. Alon--E. Paran ``On the geometry of zero sets of central quaternionic polynomials"
{\sc J. Algebra}, {\bf 659}, p. 780--788, (2024)
%Arxiv: {\tt{https://doi.org/10.48550/arXiv.2402.01378}}
%\bibitem{BR} Z. Bao--Z. Reichstein ``A non commutative Nullstellensatz'', {\sc J. Algebra Appl.}, {\bf 22}, (4) p. 1--9, (2023)
%\bibitem{israeliani} G. Alon--E. Paran ``A quaternionic Nullstellensatz'', {\sc J. Pure Appl. Algebra}, {\bf 225}, (4) p. 1--9, (2021)
%\bibitem{Ash} H. Aslaksen ``Quaternionic Determinants'' {\sc The Mathematical Intelligencer} {\bf 18}, p. 57--65, (1996)
%\bibitem{BR} Z. Bao--Z. Reichstein ``A non commutative Nullstellensatz'', {\sc J. Algebra Appl.}, {\bf 22}, (4) p. 1--9, (2023)
%\bibitem{bren} J.L. Brenner ``Applications of the Dieudonn\'e Determinant'', {\sc Linear Algebra and its Applications}, {\bf 1}, p. 511--536, (1968)
%\bibitem{Cas} B. Casselman ``Essays on the structure of reductive groups -- Determinants '', notes last revised in 2021
%\bibitem{Cohn} P.M. Cohn {\bf Skew fields --
%Theory of general division rings}
%Encyclopedia of Mathematics and its Applications,
%{\bf 57} (2008)
%\bibitem{AD} C. D'Andrea--A. Dickenstein ``Explicit formulas for the multivariate resultant'' {\sc J. Pure Appl. Algebra}, {\bf 164}, p. 59--86, (2001)
%\bibitem{erik} A. Lj. Eri\'c ``The resultant of non-commutative polynomials'' {\sc Matematiqki Vesnik} {\bf 60}, p. 3--8,(2008)
\bibitem{libroGSS} G. Gentili--C. Stoppato--D.C. Struppa {\bf Regular Functions of a Quaternionic Variable}, Springer Monographs in Mathematics, (2022)
%\bibitem{GS} G. Gentili--D.C Struppa ``A new theory of regular functions of a quaternionic variable'', {\sc Adv. Math.},
%{\bf 216}, (1), p. 279--301, (2007)
%\bibitem{GSV} G. Gentili--D.C. Struppa--F. Vlacci, ``The fundamental theorem of algebra for Hamilton and Cayley numbers.'' {\sc Math. Z.} {\bf 259}, p. 895--902, (2008)
\bibitem{severalvariables} R. Ghiloni -- A. Perotti ``Slice regular functions in several variables'', {\sc Math. Z} {\bf 302}, (1) p. 295--351, (2022)
\bibitem{Nul1} A. Gori--G. Sarfatti--F. Vlacci ``Zero sets and Nullstellensatz type theorems for slice regular quaternionic polynomials'',
{\sc Linear Algebra Appl.} {\bf 685}, p. 162--181, (2024)
\bibitem{Nul2} A. Gori--G. Sarfatti--F. Vlacci ``A Strong Version of the Hilbert Nullstellensatz for slice regular polynomials in several quaternionic variables", accepted for publication in {\sc J. Algebra}, preprint available at {\tt arXiv:2402.16784 }
%Published online in Linear Algebra Appl. https://doi.org/10.1016/j.laa.2024.01.005 (2024)
%\bibitem{reyes} M. Reyes ``A one-sided prime ideal principle for non commutative rings'', {\sc Journal of Algebra and Its Applications} p. 877--919 (2010)
%\bibitem{lang} S. Lang, {\bf Algebra}, Graduate Text in Mathematics {\bf 211}, (2002)
%\bibitem{manetti} M. Manetti, ``Nullstellensatz per tutti'',Miscellanea Franco Conti, Centro di Ricerca Matematica ``Ennio De Giorgi'', (2006)
%\bibitem{PV} J. Prezelj--F. Vlacci ``Divergence zero quaternionic vector fields and
%Hamming graphs", {\sc Ars Mathematica Contemporanea} {\bf 19}, p. 189--208, (2020)
%\bibitem{Rasheed} R. Rasheed ``Resultant-based Elimination for Skew Polynomials'', Proceedings of the 23rd International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC '21), pp. 11-18. IEEE, (2021)
\bibitem{reyes} M. Reyes ``A one-sided prime ideal principle for non commutative rings'', {\sc J. Algebra Appl.} {\bf 9}, (6) p. 877--919, (2010)
%\bibitem{singularities} C. Stoppato ``Singularities of slice regular functions'' {\sc Math. Nach.} {\bf 285}, (2012)
%\bibitem{ZZ}X. Zhao-- Y. Zhang ``Resultants of quaternionic polynomials''
%{\sc Hacet. J. Math. Stat.} {\bf 48}, (5), p. 1304--1311, (2019)
\end{thebibliography}
\end{document}