1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
|
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML3.2 EN">
<HTML>
<HEAD> <link rel="canonical" href="http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/KSP/MatCreateLMVMSymBadBrdn.html" />
<META NAME="GENERATOR" CONTENT="DOCTEXT">
<TITLE>MatCreateLMVMSymBadBrdn</TITLE>
</HEAD>
<BODY BGCOLOR="FFFFFF">
<div id="version" align=right><b>petsc-3.10.3 2018-12-18</b></div>
<div id="bugreport" align=right><a href="mailto:petsc-maint@mcs.anl.gov?subject=Typo or Error in Documentation &body=Please describe the typo or error in the documentation: petsc-3.10.3 v3.10.3 docs/manualpages/KSP/MatCreateLMVMSymBadBrdn.html "><small>Report Typos and Errors</small></a></div>
<A NAME="MatCreateLMVMSymBadBrdn"><H1>MatCreateLMVMSymBadBrdn</H1></A>
Creates a limited-memory Symmetric "Bad" Broyden-type matrix used for approximating Jacobians. L-SymBadBrdn is a convex combination of L-DFP and L-BFGS such that SymBrdn^{-1} = (1 - phi)*BFGS^{-1} + phi*DFP^{-1}. The combination factor phi is restricted to the range [0, 1], where the L-SymBadBrdn matrix is guaranteed to be symmetric positive-definite. Note that this combination is on the inverses and not on the forwards. For forward convex combinations, use the L-SymBrdn matrix.
<H3><FONT COLOR="#CC3333">Synopsis</FONT></H3>
<PRE>
#include "petscksp.h"
<A HREF="../Sys/PetscErrorCode.html#PetscErrorCode">PetscErrorCode</A> <A HREF="../KSP/MatCreateLMVMSymBadBrdn.html#MatCreateLMVMSymBadBrdn">MatCreateLMVMSymBadBrdn</A>(<A HREF="../Sys/MPI_Comm.html#MPI_Comm">MPI_Comm</A> comm, <A HREF="../Sys/PetscInt.html#PetscInt">PetscInt</A> n, <A HREF="../Sys/PetscInt.html#PetscInt">PetscInt</A> N, <A HREF="../Mat/Mat.html#Mat">Mat</A> *B)
</PRE>
The provided local and global sizes must match the solution and function vectors
used with <A HREF="../KSP/MatLMVMUpdate.html#MatLMVMUpdate">MatLMVMUpdate</A>() and <A HREF="../Mat/MatSolve.html#MatSolve">MatSolve</A>(). The resulting L-SymBrdn matrix will have
storage vectors allocated with <A HREF="../Vec/VecCreateSeq.html#VecCreateSeq">VecCreateSeq</A>() in serial and <A HREF="../Vec/VecCreateMPI.html#VecCreateMPI">VecCreateMPI</A>() in
parallel. To use the L-SymBrdn matrix with other vector types, the matrix must be
created using <A HREF="../Mat/MatCreate.html#MatCreate">MatCreate</A>() and <A HREF="../Mat/MatSetType.html#MatSetType">MatSetType</A>(), followed by <A HREF="../KSP/MatLMVMAllocate.html#MatLMVMAllocate">MatLMVMAllocate</A>().
This ensures that the internal storage and work vectors are duplicated from the
correct type of vector.
<P>
Collective on <A HREF="../Sys/MPI_Comm.html#MPI_Comm">MPI_Comm</A>
<P>
<H3><FONT COLOR="#CC3333">Input Parameters</FONT></H3>
<TABLE border="0" cellpadding="0" cellspacing="0">
<TR><TD WIDTH=40></TD><TD ALIGN=LEFT VALIGN=TOP><B>comm </B></TD><TD>- MPI communicator, set to <A HREF="../Sys/PETSC_COMM_SELF.html#PETSC_COMM_SELF">PETSC_COMM_SELF</A>
</TD></TR>
<TR><TD WIDTH=40></TD><TD ALIGN=LEFT VALIGN=TOP><B>n </B></TD><TD>- number of local rows for storage vectors
</TD></TR>
<TR><TD WIDTH=40></TD><TD ALIGN=LEFT VALIGN=TOP><B>N </B></TD><TD>- global size of the storage vectors
</TD></TR></TABLE>
<P>
<H3><FONT COLOR="#CC3333">Output Parameter</FONT></H3>
<DT><B>B </B> -the matrix
<br>
<P>
It is recommended that one use the <A HREF="../Mat/MatCreate.html#MatCreate">MatCreate</A>(), <A HREF="../Mat/MatSetType.html#MatSetType">MatSetType</A>() and/or <A HREF="../Mat/MatSetFromOptions.html#MatSetFromOptions">MatSetFromOptions</A>()
paradigm instead of this routine directly.
<P>
<H3><FONT COLOR="#CC3333">Options Database Keys</FONT></H3>
<DT><B>-mat_lmvm_num_vecs </B> -maximum number of correction vectors (i.e.: updates) stored
<br>
<DT><B>-mat_lmvm_phi </B> -(developer) convex ratio between BFGS and DFP components of the update
<br>
<DT><B>-mat_lmvm_scale_type </B> -(developer) type of scaling applied to J0 (none, scalar, diagonal)
<br>
<DT><B>-mat_lmvm_theta </B> -(developer) convex ratio between BFGS and DFP components of the diagonal J0 scaling
<br>
<DT><B>-mat_lmvm_rho </B> -(developer) update limiter for the J0 scaling
<br>
<DT><B>-mat_lmvm_alpha </B> -(developer) coefficient factor for the quadratic subproblem in J0 scaling
<br>
<DT><B>-mat_lmvm_beta </B> -(developer) exponential factor for the diagonal J0 scaling
<br>
<DT><B>-mat_lmvm_sigma_hist </B> -(developer) number of past updates to use in J0 scaling
<br>
<P>
<P>
<H3><FONT COLOR="#CC3333">See Also</FONT></H3>
<A HREF="../Mat/MatCreate.html#MatCreate">MatCreate</A>(), MATLMVM, MATLMVMSYMBRDN, <A HREF="../KSP/MatCreateLMVMDFP.html#MatCreateLMVMDFP">MatCreateLMVMDFP</A>(), <A HREF="../KSP/MatCreateLMVMSR1.html#MatCreateLMVMSR1">MatCreateLMVMSR1</A>(),
<BR><A HREF="../KSP/MatCreateLMVMBFGS.html#MatCreateLMVMBFGS">MatCreateLMVMBFGS</A>(), <A HREF="../KSP/MatCreateLMVMBrdn.html#MatCreateLMVMBrdn">MatCreateLMVMBrdn</A>(), <A HREF="../KSP/MatCreateLMVMBadBrdn.html#MatCreateLMVMBadBrdn">MatCreateLMVMBadBrdn</A>()
<P><B></B><H3><FONT COLOR="#CC3333">Level</FONT></H3>intermediate<BR>
<H3><FONT COLOR="#CC3333">Location</FONT></H3>
</B><A HREF="../../../src/ksp/ksp/utils/lmvm/symbrdn/symbadbrdn.c.html#MatCreateLMVMSymBadBrdn">src/ksp/ksp/utils/lmvm/symbrdn/symbadbrdn.c</A>
<BR><A HREF="./index.html">Index of all KSP routines</A>
<BR><A HREF="../../index.html">Table of Contents for all manual pages</A>
<BR><A HREF="../singleindex.html">Index of all manual pages</A>
</BODY></HTML>
|
