Print a warning when translating subscripted functions

[maxima.git] / share / sym / docsym-en.tex

\documentclass[11pt]{article}
\usepackage[letterpaper]{geometry}
\usepackage{amsmath,amssymb}
\usepackage{makeidx}
\usepackage{supertabular}
\usepackage[hypertex,colorlinks=true,linkcolor=blue,filecolor=webgreen,%
hyperindex=true,extension=dvi]{hyperref}


\title{The SYM Package for Maxima \\ User's Manual}
\author{Annick Valibouze\\
LITP (tour 45-55)\\
4, Place Jussieu\\
75252 Paris Cedex 05\\
Unit\'e associ\'ee au CNRS No 248\\
et\\
GDR DE CALCUL FORMEL MEDICIS\\
\small{e-mail: avb@sysal.ibp.fr}}
\date{July 19 1993}

\makeindex

\begin{document}
\maketitle

\tableofcontents 
\newpage
\noindent

\section{Introduction}

This documentation concerns a module for manipulating symmetric functions,
implemented in Common Lisp. This module, named \texttt{SYM}, is actually
presented as an extension of the symbolic calculation system \texttt{Maxima}.

Among its applications we note the calculation of resolvents.


\section{Definitions and notation}

We will consider a ring $\mathcal{A}$.

\subsection{Symmetric polynomials}

We are given an integer $n>0$.

\subsubsection{Group actions}

Let $S_n$ be the symmetric group of degree $n$ (each element is a permutation of
$\{1,\dots,n\}$).

Let $E$ be an arbitrary set. Denote by $E^n$ the set of $n$-tuples of elements
of $E$.  The \emph{action} of $S_n$ on any $n$-tuple $\underline
a=(a_1,\ldots,a_n)$ of $E^n$ is defined as
\begin{eqnarray*}
  S_n   \times E^n & \longrightarrow &  E^n\\
  \sigma \times {\underline a} & \longrightarrow &
  \sigma{\underline a}=(a_{\sigma (1)},\ldots,a_{\sigma (n)}).
\end{eqnarray*}
Let $\mathcal{A}[\underline x]$ be the ring of polynomials in the $n$ variables
$\underline x=(x_1,\ldots ,x_n)$, with coefficients in $\mathcal{A}$.  The
action of $S_n$ on $\mathcal{A}(\underline x)$ is defined as follows:
\begin{eqnarray*}
  S_n \times \mathcal{A}[\underline x] & \longrightarrow &
  \mathcal{A}[\underline x] \\
  \sigma \times f  & \longrightarrow &  \sigma f \;\; :\;
  \sigma f({\underline x}) = f(\sigma{\underline x}).
\end{eqnarray*}

\textbf{Remark.} The action of $S_n$ extends to the sets $E^c$ or
$\mathcal{A}[x_1,\ldots ,x_c]$ with $c \leq n$ simply by embedding these sets in
$E^n$ and $\mathcal{A}[\underline x]$ respectively.

Let $f$ be a function from $\mathcal{A}[\underline x]$ (resp. $\underline a$, a
$n$-tuple).  We will denote $S_nf$ (resp. $S_n \underline a$) the set $\{\sigma
f \;|\; \sigma \in S_n \}$ (resp. $\{\sigma\underline a \;|\; \sigma \in S_n
\}$), called the \textit{orbit} of $f$ (resp. $\underline a$) under the action
of $S_n$.


\subsubsection{Partitions, symmetric polynomials}

Let $m$ be a natural number.  A \textit{partition} of $m$ is a decreasing
sequence of natural numbers $i_1\geq i_2 \geq \ldots$ called \textit{parts}
whose sum is $m$. This sum is the \textit{weight} of the partition and its
\textit{length} is the number of its non-zero parts.

Let $p$ be a polynomial in $n$ variables.  This polynomial is \textit{symmetric}
if it remains invariant under the action of $S_n$, i.e. if $S_n p =\{p\}$.

A symmetric polynomial may be represented by partitions.  Let $I=(i_1,\ldots,
i_n)$ (i.e. $i_1\geq i_2\geq \ldots \geq i_n$) be a partition of length $\le n$
and let ${\underline x}^I$ be the monomial
\begin{equation*}
  {\underline x}^I = x_1^{i_1}x_2^{i_2}\ldots x_n^{i_n}.
\end{equation*}
A \textit{monomial form} $M_I({\underline x})$ on ${\underline x}$ indexed by
$I$ is the sum of the monomials of the orbit of ${\underline x}^I$ under the
action of $S_n$:
\begin{equation*}
  M_I({\underline x}) =\sum_{J \in S_nI}{\underline x}^J.
\end{equation*}
The monomial forms constitute naturally a basis of the vector space on the ring
of symmetric polynomials: any symmetric polynomial can be represented by a
finite linear combination of monomial forms.


\subsubsection{Contracted and partitioned representations}

Let $p$ be a symmetric polynomial on a ring $\mathcal{A}$, and suppose that it
is expressed in the basis of monomial forms.

In the \emph{contracted} representation of $p$ we replace every monomial form
$M_I(x)$ by $x^I$, or by every monomial of $S_nx^I$, its orbit.

In the \emph{partitioned} representation, we replace every monomial term
$cM_I(x)$ where $c$ is a coefficient in $\mathcal{A}$, by the list
$[c,i_1,\ldots ,i_n]$ and join the whole into one list.

\textbf{Example.} The contracted polynomial associated to $3x^4 + 3y^4 - 2xy^5 -
2x^5y$ is $3x^4 -2xy^5$, and the partitioned polynomial is [[3,4],[-2,5,1]].



\subsection{Multi-symmetric polynomials}

Here we generalize the symmetric case. We are given an integer $r>0$ and an
$r$-tuple of integers $D=(d_1, \ldots ,d_r)$.

\subsubsection{Group actions}

Let $E_1, E_2, \ldots ,E_r$ be $r$ arbitrary sets.  Denote by $D(E)$ the product
$E_1^{d_1}\times E_2^{d_2}\times \ldots \times E_r^{d_r}$.

The action of the product of the symmetric groups $S_D=S_{d_1}\times S_{d_2}
\times \ldots \times S_{d_r}$ on $D(E)$ is naturally defined as
\begin{eqnarray*}
  S_D \times D(E)  & \longrightarrow & D(E) \\
  \sigma \times A  & \longrightarrow &
  \sigma A=(\sigma_1{\underline a}_1,\ldots ,\sigma_r{\underline a}_r),
\end{eqnarray*}
where $\sigma = (\sigma_1,\ldots ,\sigma_r)$ with $\sigma_i \in S_{d_i}$.  The
orbit of $A$ under the action of $S_D$ will be denoted $S_DA$.

Similarly, if $X=({\underline x}_1, \ldots,{\underline x}_r)$ is an $r$-tuple
such that the $i$th element ${\underline x}_i$ is a $d_i$-tuple of variables, we
wil denote by $\mathcal{A}[X]$ the set of polynomials in the variables of $X$
and with coefficients in $\mathcal{A}$.  The action of $S_D$ on $\mathcal{A}[X]$
is defined as follows:
\begin{eqnarray*}
  S_D \times \mathcal{A}[X] & \longrightarrow & \mathcal{A}[X]\\
  \sigma \times f  & \longrightarrow & \sigma f \;\; :\;
  \sigma f(X)= f(\sigma X).
\end{eqnarray*}
The orbit of $f$ under the action of $S_D$ will be denoted $S_Df$.


\subsubsection{Multi-partitions, multi-symmetric polynomials}

A \textit{multi-partition}, $I$, of order $r$ is an $r$-tuple of partitions.  We
will call the \textit{multi-length} of $I$ the list of lengths of the partitions
that constitute $I$.

Let $p$ be a polynomial in $d_1+\cdots + d_r$ variables. This polynomial is
called \textit{multi-symmetric} if it is invariant by $S_D$, that is if $S_D p =
\{p\}$.

A multi-symmetric polynomial may be represented by multi-partitions.  Let $X$ be
as above and $I=(I_1,\ldots,I_r)$ be an $r$-tuple of $\mathbf{N}^{d_1}\times
\ldots \times \mathbf{N}^{d_r}$.  We will naturally denote by $X^I$ the monomial
\begin{equation*}
  X^I = {\underline x}_1^{I_1}\ldots{\underline x}_r^{I_r}.
\end{equation*}
If $I$ is a multi-partition of multi-length less than $D$, we will define the
\textit{monomial multi-form} $M_I(X)$ on $X$ indexed by $I$ as the sum of the
monomials of the orbit of $X^I$ under the action of $S_D$:
\begin{equation*}
  M_I(X) \;=\; \sum_{J \in S_DI} X^J.
\end{equation*}
It is natural to represent a multi-symmetric polynomial by a linear combination
of monomial multi-forms over the ring of coefficients of the polynomial.


\subsubsection{Contracted form}

Let us now consider a multi-symmetric polynomial with coefficients in a ring
$\mathcal{A}$ in the variables of $X$.  This polynomial is naturally expressed
as a finite sum of $cM_I(X)$ where $c \in \mathcal{A}$ and $M_I(X)$ is a
monomial multi-forme on $X$. A \textit{contracted form} associated to a
multi-symmetric polynomial consists in replacing in this polynomial each
$M_I(X)$ by the monomial $X^I$ or by an other monomial of its orbit $S_DX^I$.



\section{Initialization and usage}

The file \texttt{sym.mac} includes all the commands for auto-loading.  It is
ultimately in this file where one must change the path names if \texttt{SYM}
must be used from another directory than the one where it is installed.

The file \texttt{compile.lisp} includes the list of files to compile.  To execute
it one may load it under Lisp or under Maxima by the command
\texttt{load("sym/compile")}.  Once the the files of \texttt{SYM} are compiled
the module can be loaded under Maxima via the file \texttt{sym.mac}:
\begin{verbatim}
  load("sym.mac");
\end{verbatim}

If a function's argument is a list, the function will fill in missing values by
formal arguments: for the $i$th elementary symmetric function these will be
\texttt{ei}, for the $i$th power function they will be \texttt{pi}, and for the
$i$th complete (homogeneous) symmetric function they will be \texttt{hi}.

There are many evaluation modes for polynomials in Maxima: \texttt{meval},
\texttt{expand}, \texttt{rat}, \texttt{ratsimp}.  \texttt{SYM} allows a choice
of operation mode.  On every function call, \texttt{SYM} tests if the flag
\texttt{oper} has been changed.  If so, the operating mode is modified as
follows:
\begin{description}
  \item If \texttt{oper = meval} (default), the operations are carried out with
  \texttt{meval}.
  \item If \texttt{oper = expand}, they are done with \texttt{expand}.
  \item If \texttt{oper = rat}, they are done with \texttt{rat}.
  \item If \texttt{oper = ratsimp}, they are done with \texttt{ratsimp}.
\end{description}
The \texttt{meval} mode is advantageous in numerical calculations.  With formal
arguments, \texttt{rat} mode is often preferable.

To set the flag \texttt{oper} to \texttt{expand}, for example, it suffices to
write \texttt{oper\ :\ expand;}.



\section{Description of the available functions}

Take $q$ as the minimum of the degree of the polynomial \texttt{sym} and the
cardinal \texttt{card}.  \texttt{lvar} contains the variables of the polynomial
under consideration, thus distinguishing them from any eventual parameters.


\subsection{Combinatorics}

\begin{itemize}
  \item \texttt{arite(degree,arity,powers)} \index{arite(degree, arity, powers)}
    applies the ``arity theorem'' (A. Valibouze \textit{Sur l'arit\'e
    des fonctions}, European Journal of Combinatorics, 1992?).  This function
    allows to go from a power function of a resolvent in \texttt{arity} variables
    to a power function in \texttt{degree} variables.  It adds a binomial
    coefficient to each partition.  It is assumed that the power functions are
    given in the basis of monomial forms, in partitioned representation, in the
    list \texttt{powers}.
  \item \texttt{card\_orbit(part,n)} \index{card\_orbit(part, n)}
    $\longrightarrow$ \texttt{integer} \\
    \texttt{part} is a partition in the form $[a_1,m_1,...,a_q,m_q]$ where
    $m_i$ is the multiplicity of $a_i$ in the partition.  The function computes
    the cardinal of the orbit of the partition under the action of the symmetric
    group of degree \texttt{n}. 
  \item \texttt{multinomial(r,part)} \index{multinomial(r, part)}
    $\longrightarrow$ \texttt{integer} \\
    where \texttt{r} is the weight of the partition \texttt{part}. This function
    returns the associated multinomial coefficient: if the parts of partition
    \texttt{part} are $i_1, i_2, \dots, i_k$, the result is
    $r!/(i_1!i_2! \cdots i_k!)$.
  \item \texttt{card\_stab(L,eq)} \index{card\_stab(L, eq)}
    $\longrightarrow$ \texttt{integer} \\
    \texttt{L} is a list of ordered objects and \texttt{eq} is the equality test
    for them.  If the list \texttt{L} has length $n$, this function computes the
    cardinal of the stabilizer of \texttt{L} under the action of the symmetric
    group of order $n$.  (The \emph{stabilizer} of $L$ is the subgroup of
    permutations that maps $L$ to $L$, i.e. leaves it invariant.)
 \item \texttt{permut(L)} \index{permut(L)}
   $\longrightarrow$ \texttt{list} \\
   returns the list of permutations of the list \texttt{L}.
\end{itemize}
Examples:
\small
\begin{verbatim}
  card_orbit([5,2,1,3], 6);
                                 60
  card_stab([a, a, c, b, b], eq);
                                  4
  card_stab([1,1,2,3,3], "=");
                                  4
\end{verbatim}
Here the list of power functions is $[x^2y^4,x^5y^5 + x^2,x^2y^2+x^3]$, the
arity is 2, and the degree is 4. (This function is useful in the calculation of
resolvents.)
\begin{verbatim}
  arite(4,2,[[[1,2,4]],[[1,5,5],[1,2]],[[1,2,2],[1,3]]]);

       [[[1, 2, 4]], [[1, 5, 5], [3, 2]], [[1, 2, 2], [3, 3]]]
\end{verbatim}
\normalsize


\subsection{On polynomials in one variable}

\begin{itemize}
  \item \texttt{ele2polynome(cele,z)} \index{ele2polynome(cele,z)}
    $\longrightarrow$ \texttt{polynomial} \\
    returns the polynomial in \texttt{z} s.t. the elementary symmetric functions of
    its roots are in the list \texttt{cele}.  \texttt{cele}=$[n,e_1,\dots,e_n]$
    where $n$ is the degree of the polynomial and $e_i$ the $i$th elementary
    symmetric function.
  \item \texttt{polynome2ele(poly,z)} \index{polynome2ele(poly,z)}
    $\longrightarrow$ \texttt{cele} \\
    produces the list \texttt{cele}=$[n,e_1,\dots,e_n]$ where $n$ is the degree
    of the polynomial in the variable \texttt{z} and $e_i$ the $i$th elementary
    symmetric function of its roots.
\end{itemize}
Examples:
\small
\begin{verbatim}
 ele2polynome([2,e1,e2],z);
                                  2
                                 z  - e1 z + e2

 polynome2ele(x^7-14*x^5  + 56*x^3  - 56*x + 22, x);
 
                      [7, 0, - 14, 0, 56, 0, - 56, - 22]

 ele2polynome( [7, 0, - 14, 0, 56, 0, - 56, - 22], x);

                          7       5       3
                         x  - 14 x  + 56 x  - 56 x + 22
\end{verbatim}
\normalsize


\subsection{Changing representations}

A symmetric polynomial may be given in many forms: extended, contracted, or
partitioned. The functions described below-dessous allow passing from one form
to another. Some of them additionally test for symmetry.

In all cases the variables of the polynomial are contained in the list
\texttt{lvar}.
\begin{itemize}
   \item \texttt{tpartpol(poly,lvar)} \index{tpartpol(polynomial, lvar)}
   $\longrightarrow$ \texttt{ppart} \\ 
   \texttt{partpol(poly, lvar)} \index{partpol(poly,lvar)}
   $\longrightarrow$ \texttt{ppart} \\ 
   return, in increasing (resp. decreasing) lexicographic order, the partitioned
   polynomial associated to the polynomial given in extended form.  If the
   polynomial is not symmetric, the function \texttt{tpartpol} returns an error.
  \item \texttt{tcontract(poly,lvar)} \index{tcontract(poly, lvar)}
    $\longrightarrow$ \texttt{pc} \\ \texttt{contract(poly,lvar)}
    \index{contract(poly, lvar)} $\longrightarrow$ \texttt{pc} \\ act respectively
    like \texttt{tpartpol} and \texttt{partpol} in recreating the contracted
    represention.
  \item \texttt{cont2part(pc,lvar)} \index{cont2part(pc, lvar)}
    $\longrightarrow$ \texttt{ppart} \\ returns the partitioned form
    \texttt{ppart} of a symmetric polynomial given in its contracted form
    \texttt{pc}.
  \item \texttt{part2cont(ppart,lvar)} \index{part2cont(ppart, lvar)}
    $\longrightarrow$ \texttt{pc} \\ returns the contracted form \texttt{pc} of a
    symmetric polynomial given in its partitioned form \texttt{ppart}.
  \item \texttt{explose(pc,lvar)} \index{explose(pc, lvar)} $\longrightarrow$
    \texttt{psym} \\ returns the extended form \texttt{psym} of a symmetric
    polynomial given in its contracted form \texttt{pc}.
\end{itemize}
Examples:
\small
\begin{verbatim}
  tpartpol(x^3-y, [x,y]);

                         manque des monomes (no monomials)
\end{verbatim}
Consider the symmetric polynomial in $\mathbb{Z}[x,y,z]$ (the ring of
polynomials in $x,y,z$ with coefficients in $\mathbb{Z}$) of which one
contracted form is $2a^3bx^4y$. We will perform various changes of
representations on it:
\small
\begin{verbatim}
 lvar : [x,y,z]$
 pc : 2*a^3*b*x^4*y;  
                                3    4
                             2 a  b x  y
 psym : explose(pc, lvar);
                         3      4      3      4      3    4   
                      2 a  b y z  + 2 a  b x z  + 2 a  b y  z 
             3    4        3      4      3    4
        + 2 a  b x  z + 2 a  b x y  + 2 a  b x  y
\end{verbatim}
\normalsize And if we apply to it the function \texttt{contract} we recover a
contracted form:
\small
\begin{verbatim}
 contract(psym, lvar);
                                 3    4
                              2 a  b x  y
 tcontract(psym, lvar);
                                 3    4
                              2 a  b x  y
 tpartpol(psym, lvar);
                                  3
                             [[2 a  b, 4, 1]]
 partpol(psym, lvar);
                                  3
                             [[2 a  b, 4, 1]]
 ppart : cont2part(pc, lvar);
                                  3
                             [[2 a  b, 4, 1]]
 part2cont(ppart, lvar);
                                 3    4
                              2 a  b x  y
\end{verbatim}
\normalsize



\subsection{Functions related to partitions}

\begin{itemize}
  \item \texttt{kostka(part1,part2)} \index{kostka(part1,part2)}
    $\longrightarrow$ integer \\
    (written by P.Esperet) returns the Kostka number associated to the
    partitions \texttt{part1} and \texttt{part2}.
  \item \texttt{treinat(part)} \index{treinat(part)} 
    $\longrightarrow$ list of partitions 
    inferior in the natural order to the partition \texttt{part} and of the same
    weight.
  \item\texttt{treillis(n)} \index{treillis(n)} 
    $\longrightarrow$ list of partitions of weight $n$.
  \item \texttt{lgtreillis(n,m)} \index{lgtreillis(n,m)} 
    $\longrightarrow$ list of partitions of weight $n$ and length $m$.
  \item \texttt{ltreillis(n,m)} \index{ltreillis(n,m)}
    $\longrightarrow$ list of partitions of weight $n$ and length $\le m$.
\end{itemize}
\small
\begin{verbatim}
 kostka([3,3,3],[2,2,2,1,1,1]);
                                  6
 lgtreillis(4,2);
                           [[3, 1], [2, 2]]
 ltreillis(4,2);
                         [[4, 0], [3, 1], [2, 2]]
 treillis(4);
                [[4], [3, 1], [2, 2], [2, 1, 1], [1, 1, 1, 1]]
 treinat([5]);
                               [[5]]
 treinat([1,1,1,1,1]);
                        [[5], [4, 1], [3, 2], [3, 1, 1],
                    [2, 2, 1], [2, 1, 1, 1], [1, 1, 1, 1, 1]]
 treinat([3,2]);
                        [[5], [4, 1], [3, 2]]
\end{verbatim}
\normalsize



\subsection{Orbit calculations}

\begin{itemize}
  \item \texttt{orbit(poly,lvar)} \index{orbit(poly, lvar)}
    $\longrightarrow$ $S_n$(poly) \\
    produces the list of polynomials in the orbit of \texttt{poly}
    under the action of the symmetric group $S_n$. The $n$ variables of
    \texttt{poly} are in the list \texttt{lvar}.
  \item \texttt{multi\_orbit(polyl,[lvar$_{1}$,lvar$_{2}$,\ldots ,lvar$_{r}$])
      \index{multi\_orbit(poly,[lvar$_{1}$,lvar$_{2}$,\ldots ,lvar$_{r}$])}
      $\longrightarrow$ ${S_D}$(poly) }\\
    the variables of \texttt{poly} are in the lists
    \texttt{lvar$_1$,lvar$_{2}$,\ldots ,lvar$_{r}$} on which act, respectively,
    the symmetric groups $S_{d_1},S_{d_2},\ldots ,S_{d_r}$.  This function
    returns the orbit of \texttt{poly} under the action of the product $S_D$ of
    these symmetric groups.
\end{itemize}
\textbf{Examples:}
\small
\begin{verbatim}
 orbit(a*x+b*y,[x,y]);
                            [a y + b x, b y + a x]
 orbit(2*x+x**2,[x,y,z]);
                                       2         2         2
                                     [z  + 2 z, y  + 2 y, x  + 2 x]
 multi_orbit(a*x+b*y,[[x,y],[a,b]]);
                                        [b y + a x, a y + b x]
 multi_orbit(x+y+2*a,[[x,y],[a,b,c]]);
 
                  [y + x + 2 c, y + x + 2 b, y + x + 2 a]
\end{verbatim}
\normalsize


\subsection{The contracted product of two symmetric polynomials}

\begin{itemize}
\item\texttt{multsym(ppart1,ppart2,n)}\index{multsym(ppart1,ppart2,n)}
  $\longrightarrow$ \texttt{pc} \\
  calculates the product of two symmetric polynomials in \texttt{n} variables
  given in the partitioned forms \texttt{part1} and \texttt{part2}.
\end{itemize}
Given two symmetric polynomials \texttt{p1} and \texttt{p2}. We will
calculate their product by a classical method for the product of two
arbitrary polynomials, and then obtain this product by \texttt{multsym}. We will
work in $\mathbb{Z}[x,y]$.
\small
\begin{verbatim}
 p1 : x*y^2  + x^2*y$
 p2 : y+x$
 prod : expand(p1*p2);
                              3      2  2    3
                           x y  + 2 x  y  + x  y
\end{verbatim}
\normalsize
We verify that this is the extended form of the product obtained with
\texttt{multsym}:
\small
\begin{verbatim}
 partpol(prod,[x,y]);
                                [[1, 3, 1], [2, 2, 2]]
 ppart1 : partpol(p1,[x,y]);
                                   [[1, 2, 1]]
 ppart2 : partpol(p2,[x,y]);
                                   [[1, 1, 0]]
 multsym(ppart1, ppart2, 2);
                                [[1, 3, 1], [2, 2, 2]]
\end{verbatim}
\normalsize


\subsection{Basis changes}

In general, the list \texttt{lvar} will represent the list of variables of
polynomials.
\begin{itemize}
  \item \texttt{elem(cel,sym,lvar)} \index{elem(cel,sym,lvar)}
    $\longrightarrow$  $P(e_1,\dots,e_q)$ \\
    decomposes the symmetric polynomial \texttt{sym} into the elementary symmetric
    functions contained in the list \texttt{cel} (3 possible flags).
  \item \texttt{multi\_elem([cel$_{1}$,\dots,cel$_{r}$],multi\_pc,[lvar$_{1}$,
     \dots,lvar$_{r}$])}
     \index{multi\_elem([cel$_{1}$,\dots,cel$_{p}$],multi\_pc,[lvar$_{1}$,
        \dots,lvar$_{r}$])}
      $\longrightarrow$ P(cel$_{1}$,\dots,cel$_{r}$) \\
    gives a polynomial \texttt{multi\_pc} multi-symmetric under the action
    of $S_D$, in multi-contracted form.  It is decomposed successively into each
    of the groups \texttt{cel$_{j}$} of elementary symmetric functions of the set 
    ${\underline x}_j$.  The list of variables \texttt{lvar}$_j$ allows reading
    the polynomial.  The variables of ${\underline x}_j$ are found in the
    expression of the multi-contracted polynomial.
  \item \texttt{pui(cpui,sym,lvar)} \index{pui(cpui, sym, lvar)}
    $\longrightarrow$ $P(p_1,\dots,p_q)$ \\ decomposes a symmetric polynomial into
    power functions (3 flags possible).
  \item \texttt{multi\_pui([cpui$_{1}$,\ldots, cpui$_{p}$],multi\_pc, [lvar$_{1}$, \ldots,var$_{p}$])}
    \index{multi\_pui([cpui$_{1}$,\ldots,cpui$_{p}$], multi\_pc, [lvar$_{1}$,\ldots,lvar$_{p}$])} 
    $\longrightarrow$ P(cpui$_{1}$,\ldots,cpui$_{p}$) \\ 
    acts like \texttt{multi\_elem} in decomposing the multi-symmetric polynomial
    into power functions.
\end{itemize}
If the symmetric polynomial is in a contracted form, the flag \texttt{elem}
should be 1 (its default value).

Let us consider the symmetric polynomial $x^4+y^4+z^4+t^4 - 2(x(y+z+t) + y(z+t)
+ z t)$ of which a contracted form is $x^4 -2 y z$.
\small
\begin{verbatim}
  elem:1$

  elem([],x**4 - 2*y*z, [x,y,z]); 

                   4          2                 2
                 e1  - 4 e2 e1  + 4 e3 e1 + 2 e2  - 2 e2 - 4 e4
\end{verbatim}
\normalsize
Let us now suppose that the number of variables of the symmetric polynomial is 3
(i.e. $t=0$). This value 3 must be noted at the head of the list \texttt{cel} as
follows:
\small
\begin{verbatim}
elem([3],x^4 - 2*y*z,[x,y,z]);

                      4          2                 2
                    e1  - 4 e2 e1  + 4 e3 e1 + 2 e2  - 2 e2
\end{verbatim}
\normalsize
If, in addition, the first elementary symmetric function has a particular value,
say $e_1=7$, it should be in the second position of the list \texttt{cel}:
\small
\begin{verbatim}
  elem([3,7],x^4-2*x*y,[x,y]);

                               2
                   28 e3 + 2 e2  - 198 e2 + 2401
\end{verbatim}
\normalsize
The $(i+1)$th element of the list \texttt{cel} should be the $i$th elementary
symmetric function.  If the symmetric polynomial is in an extended form the flag
\texttt{elem} should be 2, and if it is in a partitioned form it should be 3:
\small
\begin{verbatim}
 elem :2$
 elem([3,f1,f2,f3],x^4+y^4+z^4 - 2*(x*y + x*z + y*z),[x,y,z]);

                      4          2                 2
                    f1  - 4 f2 f1  + 4 f3 f1 + 2 f2  - 2 f2

 elem:3$
 elem(([],[[1, 2, 1]],[]);
                             e1 e2 - 3 e3
\end{verbatim}
\normalsize
It is the same for the flag \texttt{pui} associated to the function \texttt{pui}.

For the function \texttt{pui}, if formal arguments need to be added to
\texttt{cpui} we remember the cardinal of the alphabet, if it is supplied, to
calculate the power functions as a function of the first.  We then use the
function \texttt{puireduc} (see later).
\small
\begin{verbatim}
  multi_elem([[2,e1,e2],[2,f1,f2]],a*x+a^2+x^3,[[x,y],[a,b]]);

                                2                       3
                     - 2 f2 + f1  + e1 f1 - 3 e1 e2 + e1

  multi_pui([[2,p1,p2],[2,t1,t2]],a*x+a^2+x^3,[[x,y],[a,b]]);
    
                                              3
                                       3 p1 p2   p1
                          t2 + p1 t1 + ------- - ---
                                          2       2
\end{verbatim}
\normalsize
\begin{itemize}
  \item \texttt{ele2pui(m,cel)} \index{ele2pui(m,cel)}
    $\longrightarrow$ \texttt{cpui} \\
    performs the passage from the elementary symmetric functions to the power
    functions $p_1,\ldots,p_m$.
  \item \texttt{pui2ele(n,cpui)} \index{pui2ele(n,cpui)}
      $\longrightarrow$ \texttt{cel} \\
      performs the passage from the power functions to the elementary symmetric
      functions.  If the flag \texttt{pui2ele} is set to \texttt{girard}, one gets
      the list of the first \texttt{n} elementary symmetric functions
      $e_1,\dots,e_n$, and if it is set to \texttt{close}, one gets the $n$th
      elementary symmetric function.
\end{itemize}
Below, we are looking for the first 3 elementary symmetric functions.  We are
not supplying values for the first 3 power functions, so \texttt{pui2ele} adds
formal parameters \texttt{p1}, \texttt{p2}, \texttt{p3}.
\small
\begin{verbatim}
  pui2ele(3,[]);
                           2                      3
                         p1    p2  p3   p1 p2   p1
                 [3, p1, --- - --, -- - ----- + ---]
                          2    2   3      2      6
\end{verbatim}
\normalsize
Below we are looking for the first 4 elementary symmetric functions in terms of
the power functions such that $p_1=2$.  The cardinal of the alphabet being 3,
the fourth elementary symmetric function is thus null.  Subsequently the
function \texttt{ele2pui} calculates the first 3 power functions in terms of the
first 3 elementary symmetric functions.
\small
\begin{verbatim}
 pui2ele(4,[3,2]);
                         4 - p2  p3 - 3 p2 + 4
                 [3, 2 , ------, -------------, 0]
                           2        3   
 ele2pui(3,[]);
                             2                            3
                   [3, e1, e1  - 2 e2, 3 e3 - 3 e1 e2 + e1 ]
\end{verbatim}
\normalsize
Here, as the cardinal is 2, the 3d elementary symmetric function is null:
\small
\begin{verbatim}
 ele2pui(3,[2]);
                                2           3
                      [2, e1, e1  - 2 e2, e1  - 3 e1 e2]
\end{verbatim}
\normalsize
\begin{itemize}
  \item \texttt{puireduc(n,cpui)} \index{puireduc(n,cpui)}
      $\longrightarrow$ [\texttt{card}, $p_1,p_2,p_3,\dots,p_n$] \\ 
    allows finding the power functions up to order \texttt{n} by knowing those
    of order up to \texttt{m}.  The cardinal (number of variables) is specified
    in \texttt{cpui}.
\end{itemize}
If the cardinal $n$ of the alphabet is given, one may express all the power
functions in terms of the first $n$.  We take, for example, 2 as the cardinal
and ask for the first 3 power functions:
\small
\begin{verbatim}
  puireduc(3,[2]);
                                       3
                           3 p1 p2   p1
               [2, p1, p2, ------- - ---]
                             2        2
\end{verbatim}
\normalsize
\begin{itemize}
  \item \texttt{ele2comp(m,cel)} \index{ele2comp(m,cel)}
    $\longrightarrow$ \texttt{ccomp} \\
    effects the passage from the elementary symmetric functions to the complete
    (homogeneous) symmetric functions $h_1,\ldots,h_m$.
  \item \texttt{pui2comp(n,cpui)} \index{pui2comp(n,cpui)}
    $\longrightarrow$ \texttt{ccomp} \\
    effects the passage from the power functions to the complete symmetric
    functions $h_1,\ldots,h_n$.
  \item \texttt{comp2ele(n,ccomp)} \index{comp2ele(n,ccomp)}
    $\longrightarrow$ \texttt{cel} \\
    effects the passage from the complete symmetric functions to the elementary
    symmetric functions $e_1,\ldots,e_n$.
  \item \texttt{comp2pui(n,ccomp)} \index{comp2pui(n,ccomp)}
    $\longrightarrow$ \texttt{cpui} \\
    effects the passage from the complete symmetric functions to the power
    functions $p_1,\ldots,p_n$.
  \item \texttt{mon2schur(part)} \index{mon2schur(part)}
    $\longrightarrow$ \texttt{pc} \\
    effects the passage from the monomial forms to the Schur functions.  The
    result \texttt{pc} is thus a symmetric polynomial in a contracted form.
  \item \texttt{schur2comp(P,[$h_{i_1}$,\dots,$h_{i_q}$])}
    \index{schur2comp(P,L)}
    $\longrightarrow$ \texttt{list of lists} \\
    effects the passage from the Schur functions, denoted $S_I$, to the complete
    functions.  The polynomial \texttt{P} is a polynomial in the complete
    functions $h_{i_k}$.  \emph{It is necessary} to denote these complete
    functions by an \texttt{h} concatenated with an integer.
\end{itemize}

The function \texttt{mon2schur} allows writing a Schur function in the basis of
monomial forms represented in their contracted form.  The Schur function is
given by a partition\footnote{A \emph{Schur function} is a certain kind of
  symmetric polynomial.  Schur polynomials are connected with Vandermonde
  determinants, and with Young tableaux, among other things. For some
  background, see, e.g. D.E. Knuth, Vol. 3, 2nd Ed., \S5.1.4, and Exercise 33
  there.}.
We will first verify that the Schur function associated with the partition
$(1^3)$ is equal to the 3d elementary symmetric function and that the one
associated with the partition $(3)$ is equal to the 3d complete symmetric
function (this follows from a general result).
\small
\begin{verbatim}
 mon2schur([1,1,1]);
                       x1 x2 x3
 mon2schur([3]);
                            2        3
               x1 x2 x3 + x1  x2 + x1

 mon2schur([1,2]);
                                   2
                    2 x1 x2 x3 + x1  x2
\end{verbatim}
\normalsize
Let us see with an example how, by a circular sequence of changes of basis, we
indeed recover what is given initially:
\small
\begin{verbatim}
  a1 :  pui2comp(3,[3]);
                                2                 3
                         p2   p1   p3   p1 p2   p1
                 [3, p1, -- + ---, -- + ----- + ---]
                          2    2   3      2      6
  a2 : comp2ele(3, a1);
                      2                      3
                    p1    p2  p3   p1 p2   p1
            [3, p1, --- - --, -- - ----- + ---]
                     2    2   3      2      6
  a3 : ele2pui(3,a2);
                       [3, p1, p2, p3]

  a4 : comp2pui(3,[]);
                          2                     3
         [3, h1, 2 h2 - h1 , 3 h3 - 3 h1 h2 + h1 ]

  a5 : pui2ele(3,a4);
                     2                        3
           [3, h1, h1  - h2, h3 - 2 h1 h2 + h1 ]

  a6 : ele2comp(3,a5);
                       [3, h1, h2, h3]
\end{verbatim}
\normalsize
Below we show how to express a Schur function in the basis of monomial forms (in
\texttt{c48}), of complete functions (in \texttt{c50}), elementary symmetric
functions (in \texttt{c51}), and power functions (in \texttt{c52}).
\small
\begin{verbatim}
(c48) mon2schur([1,2]);

                           2
(d48)       2 x1 x2 x3 + x1  x2

(c49) comp2ele(3, []);

                     2                        3
(d49)      [3, h1, h1  - h2, h3 - 2 h1 h2 + h1 ]

(c50) elem(d49, d48, [x1,x2,x3]);

(d50)             h1 h2 - h3

(c51) elem([], d48, [x1,x2,x3]);

(d51)             e1 e2 - e3

(c52) pui([], d48, [x1,x2,x3]);

                3
              p1    p3
(d52)         --- - --
              3    3

(c53) schur2comp(h1*h2-h3, [h1,h2,h3]);


(d53)                               s     
                                     1, 2

(c54) schur2comp(a*h3,[h3]);

(d54)                                 s  a
                                       3

\end{verbatim}
\normalsize
With the last commands we decomposed $h_1 h_2 - h_3$ and $h_3$ in the basis of
Schur functions.



\subsection{Resolvents}

Let $p$ be a polynomial of one variable $x$ and of degree $n$ on a ring $A$.
Let $f \in A[x_1,x_2,\ldots,x_n]$ be a transformation function and $S_nf$ its
orbit under the action of the symmetric group $S_n$.  Then the \emph{resolvent}
of $p$ by $f$, denoted $f_*(p)$, is the unitary polynomial:
\begin{equation*}
  f_*(p)(y) = \prod_{h\in S_nf} (y - h(\alpha_1,\ldots ,\alpha_n)),
\end{equation*}
where $\alpha_1,\ldots,\alpha_n$ are the roots of $p$.

If the transformation function is
\begin{enumerate}
  \item a polynomial in a single variable ,
  \item a linear polynomial,
  \item a linear polynomial whose non-zero coefficients alternate in sign,
  \item a linear polynomial with coefficients equal to 1,
  \item a polynomial symmetric in its variables,
  \item a monomial whose coefficient is 1,
  \item the function of the Cayley resolvent,
  \item of the Lagrange resolvent,
  \item an arbitrary polynomial,
\end{enumerate}
then each associated resolvent is respectively called
\begin{enumerate}
  \item \textit{unitary},
  \item \textit{linear},
  \item \textit{alternating},
  \item \textit{sum},
  \item \textit{symmetric},
  \item \textit{product},
  \item \textit{Cayley},
  \item \textit{Lagrange},
  \item \textit{general}.
\end{enumerate}
As we will see later, there are also other resolvents (dihedral, Klein, \dots).

We define the \textit{arity} of a function as the number of variables that
appear in its expression.  In general if a function is of arity $k$ we use
$f(x_1,\ldots,x_k)$ in place of $f(x_1,\ldots ,x_k ,\ldots ,x_n)$. Let us give
examples for each of these cases:
\begin{enumerate}
  \item $f(x)=x^7-x+1$ of arity 1,
  \item $f(x_1,x_2) = x_1+3x_2$ of arity 2,
  \item $f(x_1,x_2,x_3,x_4) = x_1 -x_2 + 3x_3-3x_4$ of arity 4,
  \item $f(x_1,x_2,x_3) = x_1+x_2+x_3$ of arity 3,
  \item $f(x_1,x_2,x_3) = 3x_1x_2 + 3x_2x_3 +3x_1x_3$ of arity 3,
  \item $f(x_1,x_2) =x_1x_2$ of arity 2,
  \item $f(x_1,x_2,x_3,x_4,x_5)=(x_1x2+x_2x_3+x_3x_4+x_4x_5+x_5x_1 -
  (x_1x_3+x_3x_5+x_5x_2+x_2x_4+x_4x_1))^2$
  \item $f(x_1,x_2,x_3) = \epsilon x_1 + \epsilon^2 x_2 + \epsilon^3 x_3$, where
  $\epsilon$ is a third root of unity,
  \item $f(x_1,x_2,x_3) = x_1 +2x_2x_3$ of arity 3.
\end{enumerate}
It is clear that a sum or product resolvant is also symmetric, and that a sum or
alternating resolvent is also linear.

The calculations of these resolvents are done in two ways.  Either by the
function \texttt{resolvante} or by a specific name preceded by
\texttt{resolvante}.  The list of possible functions is thus:
\begin{itemize}
  \item \texttt{resolvante\_produit\_sym(p, x)}
    \index{resolvante\_produit\_sym(p, x)} which calculates the list of all the
    product resolvents of the polynomial \texttt{p(x)}.
  \item \texttt{resolvante\_unitaire(p,q,x)} \index{resolvante\_unitaire(p, q, x)}
    which calculates the resolvant of the polynomial $p(x)$ by the polynomial
    $q(x)$.  That is, $\prod_{p(\alpha)=0}(y-q(\alpha))$.
  \item \texttt{resolvante\_alternee1(p,x)} \index{resolvante\_alternee1(p, x)}
    which calculates the transformation of $p(x)$ of degree $n$ by the function
    $\prod_{1\leq i<j\leq n-1} (x_i-x_j)$.
  \item \texttt{resolvante\_klein(p,x)} \index{resolvante\_klein(p, x)} which
    calculates the transformation of $p(x)$ by the function $x_1x_2+x_3$, and is
    so named because it is associated to the Klein group.
  \item \texttt{resolvante\_klein3(p,x)} \index{resolvante\_klein3(p, x)} which
    calculates the transformation of $p(x)$ by the function $x_1x_2x_4+x_4$.
  \item \texttt{resolvante\_vierer(p,x)} \index{resolvante\_vierer(p, x)} which
    calculates the transformation of $p(x)$ by the function $x_1x_2-x_3x_4$.
  \item \texttt{resolvante\_diedrale(p,x)} \index{resolvante\_diedrale(p, x)}
    which calculates the transformation of $p(x)$ by the function $x_1x_2+x_3x_4$.
  \item \texttt{resolvante\_bipartite(p,x)} \index{resolvante\_bipartite(p, x)}
    which calculates the transformation of $p(x)$ by the function $x_1x_2\ldots
    x_{n/2}+x_{n/2+1}\ldots x_n$.  The degree of $p$ must necessarily be even.
  \item \texttt{resolvante(p, x, f,[x1,x2,\ldots,xk])}
    \index{resolvante(p,x,f,[x1,x2,\ldots,xk])} which calculates the transformation
    of $p(x)$ by the function $f$ of arity $k$ in the variables
    $x_1,x_2,\ldots,x_k$.
\end{itemize}

It is essential for the efficiency of the computations to include in the list of
variables of $f$ only those which appear effectively in its expression.  Before
calling the function \texttt{resolvante}, according to the type of resolvent
calculated, one may set the flag \texttt{resolvante} so as to use the most
suitable algorithm.  According to the kinds of resolvent cited above, the flag
\texttt{resolvante} should be set respectively to
\begin{enumerate}
  \item unitaire,
  \item lineaire,
  \item alternee,
  \item somme,
  \item symetrique,
  \item produit,
  \item Cayley
  \item Lagrange,
  \item generale.
\end{enumerate}
The notion of resolvent can be gneralized by the transformation of $r$
polynomials by a function depending on $r$ blocks of variables:
\begin{itemize}
\item \texttt{direct} \index{direct}
  ([$P_1,P_2,\ldots,P_r$],y,f,[$lvar_1,lvar_2,\ldots ,lvar_r$])
  $\longrightarrow$ $f_*(P_1,P_2, \ldots, P_r)(y)$  \\
  which, given the $r$ polynomials $P_1,\ldots,P_r$ in the variable $y$, of
  degrees $d_1,\ldots ,d_r$, returns the product polynomial of $(y - h(a^{(1)},
  \ldots, a^{(p)}))$ where $a^{(i)}$ is the $d_i$-tuple of roots of $P_i$ for
  $i=1,\dots,r$ and where $h$ runs through the whole orbit of the function $f$
  under the action of the product of the symmetric groups $S_{d_1}\times \cdots
  \times S_{d_r}$.  We put in $lvar_i$ the variables of $f$ which appear
  effectively in its expression.  That is, $lvar_i$ may include fewer than $d_i$
  variables.  The function \texttt{direct} will deduce the action of the
  symmetric group $S_{d_i}$.
\end{itemize}
\textbf{Examples}.
\small
\begin{verbatim}
  resolvante:unitaire;
  resolvante(x^7-14*x^5+56*x^3-56*x+22, x, x^3-1, [x]);

        7      6        5         4          3   
      y  + 7 y  - 539 y  - 1841 y  + 51443 y  

                                  2
                        + 315133 y  + 376999 y + 125253  

 resolvante : lineaire$
 resolvante(x^4-1, x, x1+2*x2+3*x3, [x1,x2,x3]);
    24       20         16            12       
   y   + 80 y   + 7520 y   + 1107200 y   
                               8              4
                   + 49475840 y  + 344489984 y + 655360000
                                                      
 resolvante : generale$
 resolvante(x^4-1, x, x1+2*x2+3*x3, [x1,x2,x3]);
    24       20         16            12       
   y   + 80 y   + 7520 y   + 1107200 y   
                               8              4
                   + 49475840 y  + 344489984 y + 655360000

 resolvante(x^4-1, x, x1+2*x2+3*x3, [x1,x2,x3,x4]);
    24       20         16            12       
   y   + 80 y   + 7520 y   + 1107200 y   
                               8              4
                   + 49475840 y  + 344489984 y + 655360000

 direct([x^4-1], x, x1+2*x2+3*x3, [[x1,x2,x3]]);

    24       20         16            12       
   y   + 80 y   + 7520 y   + 1107200 y   
                               8              4
                   + 49475840 y  + 344489984 y + 655360000

 resolvante_diedrale(x^5-3*x^4+1, x);
   15       12       11       10        9         8         7        6
  x   - 21 x   - 81 x   - 21 x   + 207 x  + 1134 x  + 2331 x  - 945 x

                     5          4          3          2
             - 4970 x  - 18333 x  - 29079 x  - 20745 x  - 25326 x - 697
\end{verbatim}
\normalsize
We verify that the function \texttt{direct} is a generalisation of the function
\texttt{resolvante}:
\small
\begin{verbatim} 
 resolvante : lineaire$
 resolvante(x^4-1, x, x1+2*x2, [x1,x2]);
                           12       8        4
                          y   + 13 y  + 611 y  - 625
 direct([x^4-1], x, x1+2*x2, [[x1,x2,x3]]);
                           12       8        4
                          y   + 13 y  + 611 y  - 625
\end{verbatim}
\normalsize
The result of this last calculation was the same as
\begin{verbatim} 
  direct([x^4-1], x, x1+2*x2, [[x1,x2]]);
\end{verbatim}
As it is more efficient, it is advisable not to include variables that do not
appear in the expression of the transformation function.
\small
\begin{verbatim} 
 resolvante:lineaire$
 resolvante(x^4-1, x, x1+x2+x3, [x1,x2,x3]);
                                 4
                                y  - 1
 resolvante:symetrique$

 resolvante(x^4-1, x, x1+x2+x3, [x1,x2,x3]);
                                 4
                                y  - 1
 resolvante:lineaire$
 resolvante(x^4+x+1, x, x1-x2, [x1,x2]);
               12      8       6        4        2
              y   + 8 y  + 26 y  - 112 y  + 216 y  + 229
 resolvante:alternee$
 resolvante(x^4+x+1, x, x1-x2, [x1,x2]);
                12      8       6        4        2
              y   + 8 y  + 26 y  - 112 y  + 216 y  + 229
 resolvante:generale$
 resolvante(x^4+x+1, x, x1-x2, [x1,x2]);
               12      8       6        4        2
              y   + 8 y  + 26 y  - 112 y  + 216 y  + 229 
\end{verbatim}
\normalsize
We calculate below a direct image in two different ways.  The first displays the
intermediate calculations done by the function \texttt{direct}.  One may change
the flag \texttt{direct}.  The default is \texttt{puissances}, which means that
the function \texttt{multi\_pui} is used.  If we set \texttt{direct :
elementaire}, the function \texttt{direct} uses the function
\texttt{multi\_elem}, generally of lower performance.
\small
\begin{verbatim}
 l : pui_direct(multi_orbit(a*x+b*y, [[x,y],[a,b]]), [[x,y],[a,b]], [2,2]);

                                    2  2
                 [a x, 4 a b x y + a  x ]

 m: multi_elem([[2,e1,e2],[2,f1,f2]], l[1], [[x,y],[a,b]]);

                           e1 f1

 n: multi_elem([[2,e1,e2],[2,f1,f2]], l[2], [[x,y],[a,b]]);

                            2             2     2   2
              8 e2 f2 - 2 e1  f2 - 2 e2 f1  + e1  f1

 pui2ele(2, [2,m,n]);

                                       2           2
              [2, e1 f1, - 4 e2 f2 + e1  f2 + e2 f1 ]
 
 ele2polynome(%, y);

                    2                         2           2
                   y  - e1 f1 y - 4 e2 f2 + e1  f2 + e2 f1

 direct([z^2-e1*z+e2, z^2-f1*z+f2], z, b*v+a*u, [[u, v],[a, b]]);

                    2                         2           2
                   y  - e1 f1 y - 4 e2 f2 + e1  f2 + e2 f1
\end{verbatim}
\normalsize
\begin{itemize}
  \item\texttt{pui\_direct($[f_1,\ldots,f_q]$, [$lvar_1,\ldots,lvar_p$],
    [$d_1,d_2,\ldots,d_p$])} \\ 
    \index{pui\_direct} 
    Let $f$ be a polynomial in $r$ blocks of variables $lvar_1,\ldots,lvar_r$.
    Let $c_i$ be the number of variables in $lvar_i$ and let $S_C$ be the product
    of the $r$ symmetric groups $S_{c_i}$ of degrees $c_1,\dots,c_r$.  This group
    acts naturally on $f$.  The list \texttt{orbite} is the orbit, denoted $S_Cf$,
    of the function $f$ under the action of $S_C$.  (This list may be obtained via
    the function \texttt{multi\_orbit}).  The $d_i$ are integers such that
    $c_1\leq d_1, c_2 \leq d_2,\ldots, c_p\leq d_p$.  We denote by $S_D$ the
    product of the symmetric groups $S_{d_1} \times S_{d_2} \times \ldots \times
    S_{d_p}$.

    The function \texttt{pui\_direct} returns the first $N$ power functions of
    the orbit $S_Df$ pf the function $f$ deduced from the power functions of the
    orbit $S_Cf$ where $N$ is the cardinal of $S_Df$.

    The result is returned in multi-contracted form relative to $S_D$ (i.e. only
    one element is kept per orbit under the action of $S_D$).
\end{itemize}
\small
\begin{verbatim}
  L : [[x,y],[a,b]]$

  pui_direct([b*y + a*x, a*y + b*x], L, [2,2]);

                                2  2
             [a x, 4 a b x y + a  x ]

  pui_direct([b*y + a*x, a*y + b*x], L, [3,2]);

                         2  2     2    2        3  3
  [2 a x, 4 a b x y + 2 a  x , 3 a  b x  y + 2 a  x ,

        2  2  2  2      3    3        4  4
    12 a  b  x  y  + 4 a  b x  y + 2 a  x ,

        3  2  3  2      4    4        5  5
    10 a  b  x  y  + 5 a  b x  y + 2 a  x ,

        3  3  3  3       4  2  4  2      5    5        6  6
    40 a  b  x  y  + 15 a  b  x  y  + 6 a  b x  y + 2 a  x ]

 pui_direct ([y+x+2*c, y+x+2*b, y+x+2*a],[[x,y],[a,b,c]],[2,3]);

                             2              2
      [3 x + 2 a, 6 x y + 3 x  + 4 a x + 4 a , 

              2                   3        2       2        3
           9 x  y + 12 a x y + 3 x  + 6 a x  + 12 a  x + 8 a ]
\end{verbatim}
\normalsize 
Evidently, we find the same result by
\small
\begin{verbatim}
  pui_direct([y+x+2*a], [[x,y],[a]], [2,3]);

                             2              2
      [3 x + 2 a, 6 x y + 3 x  + 4 a x + 4 a , 

              2                   3        2       2        3
           9 x  y + 12 a x y + 3 x  + 6 a x  + 12 a  x + 8 a ]
\end{verbatim}
\normalsize


\newpage
\section{Meaning of objects}

\begin{flushleft}
\texttt{card} : the cardinal of the set of variables with which we are working.

\texttt{$e_k$} : $k$-th elementary symmetric function $\sum_{i_1<i_2<\cdots<
i_k} x_{i_1} x_{i_2} \cdots x_{i_k}$.

\texttt{$p_k$} : $k$-th power function $\sum_i x_i^k$.

\texttt{$h_k$} : $k$-th complete (homogeneous) symmetric function $\sum_{i_1\le
i_2\le \cdots\le i_k} x_{i_1} x_{i_2} \cdots x_{i_k}$.

\texttt{ele} = [$e_1,e_2,\ldots,e_n$], $n$ given in the function call.

\texttt{cele} = [card, $e_1,e_2,\ldots,e_n$]

\texttt{pui} = [$p_1,p_2,p_3,\ldots,p_m$], $m$ given in the function call.

\texttt{cpui} = [card, $p_{1},p_{2},p_{3},\ldots,p_{m}$].

\texttt{ccomp} = [card, $h_{1},h_{2},h_{3},\ldots,h_{m}$].

\texttt{sym} : symmetric polynomial of unspecified representation.

\texttt{fmc} : contracted monomial form.

\texttt{part} : partition.

\texttt{tc} : contracted term.

\texttt{tpart} : partitioned term.

\texttt{psym} : symmetric polynomial in extended form.

\texttt{pc} : symmetric polynomial in contracted form.

\texttt{multi\_pc} : multi-symmetric polynomial in multi-contracted form under
$S_D$.

\texttt{ppart} : symmetric polynomial in partitioned form.

\texttt{P($x_1,\ldots,x_q$)} : polynomial in $x_1,\ldots,x_q$.

\texttt{lvar} : list of variables.

$[lvar_1, \ldots,lvar_r]$ : list of lists of variables.
\end{flushleft}


\addcontentsline{toc}{section}{Index}
% makeindex docsym
\printindex


\end{document}