\algnewcommand\Output{\item[\algorithmicoutput]}
\newtheorem{proposition}{Proposition}
-\newtheorem{proof}{Proof}
\begin{document}
%
% affiliations
\author{\IEEEauthorblockN{Rapha\"el Couturier\IEEEauthorrefmark{1}, Lilia Ziane Khodja\IEEEauthorrefmark{2}, and Christophe Guyeux\IEEEauthorrefmark{1}}
-\IEEEauthorblockA{\IEEEauthorrefmark{1} Femto-ST Institute, University of Franche Comte, France\\
+\IEEEauthorblockA{\IEEEauthorrefmark{1} Femto-ST Institute, University of Franche-Comt\'e, France\\
Email: \{raphael.couturier,christophe.guyeux\}@univ-fcomte.fr}
\IEEEauthorblockA{\IEEEauthorrefmark{2} INRIA Bordeaux Sud-Ouest, France\\
Email: lilia.ziane@inria.fr}
However, iterative methods suffer from scalability problems on parallel
computing platforms with many processors, due to their need of reduction
-operations, and to collective communications to achive matrix-vector
+operations, and to collective communications to achieve matrix-vector
multiplications. The communications on large clusters with thousands of cores
and large sizes of messages can significantly affect the performances of these
iterative methods. As a consequence, Krylov subspace iteration methods are often used
At each outer iteration, the sparse linear system $Ax=b$ is partially
solved using only $m$
iterations of an iterative method, this latter being initialized with the
-best known approximation previously obtained.
-GMRES method~\cite{Saad86}, or any of its variants, can be used for instance as an
+last obtained approximation.
+GMRES method~\cite{Saad86}, or any of its variants, can potentially be used as
inner solver. The current approximation of the Krylov method is then stored inside a matrix
$S$ composed by the successive solutions that are computed during inner iterations.
equals to the restart number of the GMRES-like method. Moreover, a tolerance
threshold must be specified for the solver. In practice, this threshold must be
much smaller than the convergence threshold of the TSIRM algorithm (\emph{i.e.}
-$\epsilon_{tsirm}$). Line~\ref{algo:store}, $S_{k~ mod~ s}=x^k$ consists in copying the
-solution $x_k$ into the column $k~ mod~ s$ of the matrix $S$. After the
+$\epsilon_{tsirm}$). Line~\ref{algo:store}, $S_{k \mod s}=x^k$ consists in copying the
+solution $x_k$ into the column $k \mod s$ of the matrix $S$, where $S$ is a matrix of size $n\times s$ whose column vector $i$ is denoted by $S_i$. After the
minimization, the matrix $S$ is reused with the new values of the residuals. To
solve the minimization problem, an iterative method is used. Two parameters are
required for that: the maximum number of iterations and the threshold to stop the
\end{itemize}
-The parallelisation of TSIRM relies on the parallelization of all its
+The parallelization of TSIRM relies on the parallelization of all its
parts. More precisely, except the least-squares step, all the other parts are
obvious to achieve out in parallel. In order to develop a parallel version of
our code, we have chosen to use PETSc~\cite{petsc-web-page}. For
line~\ref{algo:matrix_mul} the matrix-matrix multiplication is implemented and
efficient since the matrix $A$ is sparse and since the matrix $S$ contains few
-colums in practice. As explained previously, at least two methods seem to be
+columns in practice. As explained previously, at least two methods seem to be
interesting to solve the least-squares minimization, CGLS and LSQR.
In the following we remind the CGLS algorithm. The LSQR method follows more or
\label{sec:04}
Let us recall the following result, see~\cite{Saad86}.
\begin{proposition}
+\label{prop:saad}
Suppose that $A$ is a positive real matrix with symmetric part $M$. Then the residual norm provided at the $m$-th step of GMRES satisfies:
\begin{equation}
||r_m|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{m}{2}} ||r_0|| ,
the convergence of GMRES($m$) for all $m$ under that assumption regarding $A$.
\end{proposition}
+
We can now claim that,
\begin{proposition}
-If $A$ is a positive real matrix and GMRES($m$) is used as solver, then the TSIRM algorithm is convergent.
+If $A$ is a positive real matrix and GMRES($m$) is used as solver, then the TSIRM algorithm is convergent. Furthermore,
+let $r_k$ be the
+$k$-th residue of TSIRM, then
+we still have:
+\begin{equation}
+||r_k|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{km}{2}} ||r_0|| ,
+\end{equation}
+where $\alpha$ and $\beta$ are defined as in Proposition~\ref{prop:saad}.
\end{proposition}
\begin{proof}
-Let $r_k = b-Ax_k$, where $x_k$ is the approximation of the solution after the
-$k$-th iterate of TSIRM.
-We will prove that $r_k \rightarrow 0$ when $k \rightarrow +\infty$.
+We will prove by a mathematical induction that, for each $k \in \mathbb{N}^\ast$,
+$||r_k|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{mk}{2}} ||r_0||.$
+
+The base case is obvious, as for $k=1$, the TSIRM algorithm simply consists in applying GMRES($m$) once, leading to a new residual $r_1$ which follows the inductive hypothesis due to Proposition~\ref{prop:saad}.
-Each step of the TSIRM algorithm
+Suppose now that the claim holds for all $m=1, 2, \hdots, k-1$, that is, $\forall m \in \{1,2,\hdots, k-1\}$, $||r_m|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{km}{2}} ||r_0||$.
+We will show that the statement holds too for $r_k$. Two situations can occur:
+\begin{itemize}
+\item If $k \mod m \neq 0$, then the TSIRM algorithm consists in executing GMRES once. In that case, we obtain $||r_k|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{m}{2}} ||r_{k-1}||\leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{km}{2}} ||r_0||$ by the inductive hypothesis.
+\item Else, the TSIRM algorithm consists in two stages: a first GMRES($m$) execution leads to a temporary $x_k$ whose residue satisfies $||r_k|| \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{m}{2}} ||r_{k-1}||\leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{km}{2}} ||r_0||$, and a least squares resolution.
+Let $\operatorname{span}(S) = \left \{ {\sum_{i=1}^k \lambda_i v_i \Big| k \in \mathbb{N}, v_i \in S, \lambda _i \in \mathbb{R}} \right \}$ be the linear span of a set of real vectors $S$. So,\\
+$\min_{\alpha \in \mathbb{R}^s} ||b-R\alpha ||_2 = \min_{\alpha \in \mathbb{R}^s} ||b-AS\alpha ||_2$
+
+$\begin{array}{ll}
+& = \min_{x \in span\left(S_{k-s+1}, S_{k-s+2}, \hdots, S_{k} \right)} ||b-AS\alpha ||_2\\
+& = \min_{x \in span\left(x_{k-s+1}, x_{k-s}+2, \hdots, x_{k} \right)} ||b-AS\alpha ||_2\\
+& \leqslant \min_{x \in span\left( x_{k} \right)} ||b-Ax ||_2\\
+& \leqslant \min_{\lambda \in \mathbb{R}} ||b-\lambda Ax_{k} ||_2\\
+& \leqslant ||b-Ax_{k}||_2\\
+& = ||r_k||_2\\
+& \leqslant \left(1-\dfrac{\alpha}{\beta}\right)^{\frac{km}{2}} ||r_0||,
+\end{array}$
+\end{itemize}
+which concludes the induction and the proof.
\end{proof}
+We can remark that, at each iterate, the residue of the TSIRM algorithm is lower
+than the one of the GMRES method.
+
%%%*********************************************************
%%%*********************************************************
\section{Experiments using PETSc}
\hline
\end{tabular}
-\caption{Comparison of (F)GMRES and 2 stage (F)GMRES algorithms in sequential with some matrices, time is expressed in seconds.}
+\caption{Comparison of (F)GMRES and TSIRM with (F)GMRES in sequential with some matrices, time is expressed in seconds.}
\label{tab:02}
\end{center}
\end{table}
example ex15 of PETSc on the Juqueen architecture. Different numbers of cores
are studied ranging from 2,048 up-to 16,383. Two preconditioners have been
tested: {\it mg} and {\it sor}. For those experiments, the number of components (or unknowns of the
-problems) per processor is fixed to 25,000, also called weak scaling. This
+problems) per core is fixed to 25,000, also called weak scaling. This
number can seem relatively small. In fact, for some applications that need a lot
of memory, the number of components per processor requires sometimes to be
small.
In Figure~\ref{fig:01}, the number of iterations per second corresponding to
-Table~\ref{tab:01} is displayed. It can be noticed that the number of
-iterations per second of FMGRES is constant whereas it decrease with TSIRM with
-both preconditioner. This can be explained by the fact that when the number of
-core increases the time for the minimization step also increases but, generally,
+Table~\ref{tab:03} is displayed. It can be noticed that the number of
+iterations per second of FMGRES is constant whereas it decreases with TSIRM with
+both preconditioners. This can be explained by the fact that when the number of
+cores increases the time for the least-squares minimization step also increases but, generally,
when the number of cores increases, the number of iterations to reach the
threshold also increases, and, in that case, TSIRM is more efficient to reduce
the number of iterations. So, the overall benefit of using TSIRM is interesting.
\begin{tabular}{|r|r|r|r|r|r|r|r|r|}
\hline
- nb. cores & threshold & \multicolumn{2}{c|}{GMRES} & \multicolumn{2}{c|}{TSIRM CGLS} & \multicolumn{2}{c|}{TSIRM LSQR} & best gain \\
+ nb. cores & threshold & \multicolumn{2}{c|}{FGMRES} & \multicolumn{2}{c|}{TSIRM CGLS} & \multicolumn{2}{c|}{TSIRM LSQR} & best gain \\
\cline{3-8}
& & Time & \# Iter. & Time & \# Iter. & Time & \# Iter. & \\\hline \hline
2,048 & 8e-5 & 108.88 & 16,560 & 23.06 & 3,630 & 22.79 & 3,630 & 4.77 \\
\hline
\end{tabular}
-\caption{Comparison of FGMRES and 2 stage FGMRES algorithms for ex54 of Petsc (both with the MG preconditioner) with 25000 components per core on Curie (restart=30, s=12), time is expressed in seconds.}
+\caption{Comparison of FGMRES and TSIRM with FGMRES algorithms for ex54 of Petsc (both with the MG preconditioner) with 25,000 components per core on Curie (restart=30, s=12), time is expressed in seconds.}
\label{tab:04}
\end{center}
\end{table*}
\begin{tabular}{|r|r|r|r|r|r|r|r|r|r|r|}
\hline
- nb. cores & \multicolumn{2}{c|}{GMRES} & \multicolumn{2}{c|}{TSIRM CGLS} & \multicolumn{2}{c|}{TSIRM LSQR} & best gain & \multicolumn{3}{c|}{efficiency} \\
+ nb. cores & \multicolumn{2}{c|}{FGMRES} & \multicolumn{2}{c|}{TSIRM CGLS} & \multicolumn{2}{c|}{TSIRM LSQR} & best gain & \multicolumn{3}{c|}{efficiency} \\
\cline{2-7} \cline{9-11}
- & Time & \# Iter. & Time & \# Iter. & Time & \# Iter. & & GMRES & TS CGLS & TS LSQR\\\hline \hline
+ & Time & \# Iter. & Time & \# Iter. & Time & \# Iter. & & FGMRES & TS CGLS & TS LSQR\\\hline \hline
512 & 3,969.69 & 33,120 & 709.57 & 5,790 & 622.76 & 5,070 & 6.37 & 1 & 1 & 1 \\
1024 & 1,530.06 & 25,860 & 290.95 & 4,830 & 307.71 & 5,070 & 5.25 & 1.30 & 1.21 & 1.01 \\
2048 & 919.62 & 31,470 & 237.52 & 8,040 & 194.22 & 6,510 & 4.73 & 1.08 & .75 & .80\\
\hline
\end{tabular}
-\caption{Comparison of FGMRES and 2 stage FGMRES algorithms for ex54 of Petsc (both with the MG preconditioner) with 204,919,225 components on Curie with different number of cores (restart=30, s=12, threshol 5e-5), time is expressed in seconds.}
+\caption{Comparison of FGMRES and TSIRM with FGMRES for ex54 of Petsc (both with the MG preconditioner) with 204,919,225 components on Curie with different number of cores (restart=30, s=12, threshold 5e-5), time is expressed in seconds.}
\label{tab:05}
\end{center}
\end{table*}
%%%*********************************************************
%%%*********************************************************
+A novel two-stage iterative algorithm has been proposed in this article,
+in order to accelerate the convergence Krylov iterative methods.
+Our TSIRM proposal acts as a merger between Krylov based solvers and
+a least-squares minimization step.
+The convergence of the method has been proven in some situations, while
+experiments up to 16,394 cores have been led to verify that TSIRM runs
+5 or 7 times faster than GMRES.
+
-future plan : \\
-- study other kinds of matrices, problems, inner solvers\\
-- test the influence of all the parameters\\
-- adaptative number of outer iterations to minimize\\
-- other methods to minimize the residuals?\\
-- implement our solver inside PETSc
+For future work, the authors' intention is to investigate
+other kinds of matrices, problems, and inner solvers. The
+influence of all parameters must be tested too, while
+other methods to minimize the residuals must be regarded.
+The number of outer iterations to minimize should become
+adaptative to improve the overall performances of the proposal.
+Finally, this solver will be implemented inside PETSc.
% conference papers do not normally have an appendix
%%%*********************************************************
\section*{Acknowledgment}
This paper is partially funded by the Labex ACTION program (contract
-ANR-11-LABX-01-01). We acknowledge PRACE for awarding us access to resource
+ANR-11-LABX-01-01). We acknowledge PRACE for awarding us access to resources
Curie and Juqueen respectively based in France and Germany.
