Skip to content

Commit

Permalink
new sparsity-structure packages csgrp, csgrshp & cchprodsp added
Browse files Browse the repository at this point in the history
git-svn-id: https://ccpforge.cse.rl.ac.uk/svn/cutest/cutest/trunk@326 ca3beb79-f7c6-489d-80df-314e45fd556d
  • Loading branch information
nimg committed Mar 30, 2017
1 parent a05f2af commit d1507be
Show file tree
Hide file tree
Showing 23 changed files with 1,152 additions and 185 deletions.
Binary file modified doc/pdf/cutest.pdf
Binary file not shown.
Binary file added doc/pdf/cutest_cchprodsp.pdf
Binary file not shown.
Binary file added doc/pdf/cutest_csgrp.pdf
Binary file not shown.
Binary file added doc/pdf/cutest_csgrshp.pdf
Binary file not shown.
Binary file modified doc/pdf/cutest_cshp.pdf
Binary file not shown.
12 changes: 12 additions & 0 deletions man/man3/cutest.3
Original file line number Diff line number Diff line change
Expand Up @@ -186,6 +186,10 @@ evaluate both the value and gradient of the objective function.
.B cutest_cofsg \fP(unthreaded) and \fBcutest_cofsg_threaded \fP(threaded)
evaluate both the value and sparse gradient of the objective function.
.TP
.B cutest_csgrp \fP(both unthreaded and threaded)
evaluate the sparsity pattern of the gradients of the objective function
and constraints
.TP
.B cutest_csgr \fP(unthreaded) and \fBcutest_csgr_threaded \fP(threaded)
evaluate the sparse gradients of the objective function and constraints.
.TP
Expand Down Expand Up @@ -233,6 +237,10 @@ including the objective as a sparse matrix.
evaluate the Hessian of the objective function or
an individual constraint as a sparse matrix.
.TP
.B cutest_csgrshp \fP(both unthreaded and threaded)
evaluate the sparsity pattern of the gradients of the objective function
and constraints and the Hessian of the Lagrangian function
.TP
.B cutest_csgrsh \fP(unthreaded) and \fBcutest_csgrsh_threaded \fP(threaded)
evaluate the constraint Jacobian and Hessian of the Lagrangian function
as sparse matrices.
Expand Down Expand Up @@ -266,6 +274,10 @@ evaluate the product of the constraint Jacobian or its transpose with a vector.
evaluate the product of the constraint Jacobian or its transpose with a
sparse vector.
.TP
.B cutest_cchprodsp \fP(both unthreaded and threaded)
evaluate the sparsity pattern needed when computing the products of the
constraint Hessians with a vector.
.TP
.B cutest_cchprods \fP(unthreaded) and \fBcutest_cchprods_threaded \fP(threaded)
evaluate the products of the constraint Hessians with a vector.
.TP
Expand Down
2 changes: 1 addition & 1 deletion man/man3/cutest_cchprods.3
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ CALL CUTEST_cchprods( status, n, m, goth, X, Y, VECTOR,
lchp, CHP_val, CHP_ind, CHP_ptr )
.SH DESCRIPTION
The CUTEST_cchprods subroutine forms the product of a vector with each of
the Hessian matrix of the constraint functions
the Hessian matrices of the constraint functions
.EQ
c(x)
.EN
Expand Down
114 changes: 114 additions & 0 deletions man/man3/cutest_cchprodsp.3
Original file line number Diff line number Diff line change
@@ -0,0 +1,114 @@
'\" e @(#)cutest_cchprodsp v1.5 03/2017;
.TH cutest_cchprodsp 3M "30 Mar 2017" "CUTEst user documentation" "CUTEst user documentation"
.SH NAME
CUTEST_cchprodsp \- CUTEst tool to detetrmine the sparsity structure used when
forming the matrix-vector products of a vector
with each of the Hessian matrices of the constraint functions.
.SH SYNOPSIS
.HP 1i
CALL CUTEST_cchprodsp( status, n, m, lchp, CHP_ind, CHP_ptr )
.SH DESCRIPTION
The CUTEST_cchprodsp subroutine obtins the sparsity structure used when
forming the product of a vector with each of
the Hessian matrices of the constraint functions
.EQ
c(x)
.EN
corresponding to the problem decoded from a SIF file by the script
\fIsifdecoder\fP at the point
.EQ
x =
.EN
X.

The problem under consideration
is to minimize or maximize an objective function
.EQ
f(x)
.EN
over all
.EQ
x
.EN
\(mo
.EQ
R sup n
.EN
subject to
general equations
.EQ
c sub i (x) ~=~ 0,
.EN
.EQ
~(i
.EN
\(mo
.EQ
{ 1 ,..., m sub E } ),
.EN
general inequalities
.EQ
c sub i sup l ~<=~ c sub i (x) ~<=~ c sub i sup u,
.EN
.EQ
~(i
.EN
\(mo
.EQ
{ m sub E + 1 ,..., m }),
.EN
and simple bounds
.EQ
x sup l ~<=~ x ~<=~ x sup u.
.EN
The objective function is group-partially separable
and all constraint functions are partially separable.
.LP
.SH ARGUMENTS
The arguments of CUTEST_cchprodsp are as follows:
.TP 5
.B status \fP[out] - integer
the outputr status: 0 for a succesful call, 1 for an array
allocation/deallocation error, 2 for an array bound error,
3 for an evaluation error,
.TP
.B n \fP[in] - integer
the number of variables for the problem,
.TP
.B m \fP[in] - integer
the total number of general constraints,
.TP
.B lchp \fP[in] - integer
a variable that specifies the declared lengths of CHP_val and CHP_ind.
The precise length required may be found by calling \fICUTEST_cdimchp\fP prior
to \fICUTEST_cchprodsp\fP,
.TP
.B CHP_ind \fP[inout] - integer
an array that gives the indices of the nonzeros in the result obtained by
multiplying the constraint Hessians by VECTOR. The indices for the i-th
constraint are stored in CHP_ind(CHP_ptr(i):CHP_ptr(i+1)-1), and will
match the values stored in CHP_val from a cutest_cchprods(3M) call.
.TP
.B CHP_ptr \fP[inout] - integer
an array of length m+1 that gives pointers to the starting positions in
CHP_ind for the indices of the nonzeros for the product with each Hessian.
CHP_ptr(m+1)-1 gives the total space required by CHP_ind.
.LP
.SH AUTHORS
I. Bongartz, A.R. Conn, N.I.M. Gould, D. Orban and Ph.L. Toint
.SH "SEE ALSO"
\fICUTEst: a Constrained and Unconstrained Testing
Environment with safe threads for mathematical optimization\fP,
N.I.M. Gould, D. Orban and Ph.L. Toint,
Computational Optimization and Applications \fB60\fP:3, pp.545-557, 2014.

\fICUTEr (and SifDec): A Constrained and Unconstrained Testing
Environment, revisited\fP,
N.I.M. Gould, D. Orban and Ph.L. Toint,
ACM TOMS, \fB29\fP:4, pp.373-394, 2003.

\fICUTE: Constrained and Unconstrained Testing Environment\fP,
I. Bongartz, A.R. Conn, N.I.M. Gould and Ph.L. Toint,
ACM TOMS, \fB21\fP:1, pp.123-160, 1995.

cutest_cdimchp(3M), cutest_cchprods(3M), sifdecoder(1).
19 changes: 7 additions & 12 deletions man/man3/cutest_csgr.3
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@ CUTEST_csgr \- CUTEst tool to evaluate constraints gradients and gradient of
objective/Lagrangian function.
.SH SYNOPSIS
.HP 1i
CALL CUTEST_csgr( status, n, m, X, Y, grlagf,
CALL CUTEST_csgr( status, n, m, X, Y, grlagf,
nnzj, lj, J_val, J_var, J_fun )
.SH DESCRIPTION
The CUTEST_csgr subroutine evaluates the gradients of the general
Expand All @@ -16,14 +16,9 @@ l(x,y) = f(x) + y sup T c(x)
corresponding to the problem decoded from a SIF file by the script
\fIsifdecoder\fP at the point
.EQ
(x,y) =
(x,y) =
.EN
(X,Y).
It also evaluates the Hessian matrix
of the Lagrangian function at
.EQ
(x,y).
.EN
The gradients are stored in a sparse format.

The problem under consideration
Expand Down Expand Up @@ -66,14 +61,14 @@ and simple bounds
.EQ
x sup l ~<=~ x ~<=~ x sup u.
.EN
The objective function is group-partially separable and
The objective function is group-partially separable and
all constraint functions are partially separable.
.LP
.LP
.SH ARGUMENTS
The arguments of CUTEST_csgr are as follows
.TP 5
.B status \fP[out] - integer
the outputr status: 0 for a succesful call, 1 for an array
the outputr status: 0 for a succesful call, 1 for an array
allocation/deallocation error, 2 for an array bound error,
3 for an evaluation error,
.TP
Expand Down Expand Up @@ -122,7 +117,7 @@ j-th general constraint function.
.SH AUTHORS
I. Bongartz, A.R. Conn, N.I.M. Gould, D. Orban and Ph.L. Toint
.SH "SEE ALSO"
\fICUTEst: a Constrained and Unconstrained Testing
\fICUTEst: a Constrained and Unconstrained Testing
Environment with safe threads\fP,
N.I.M. Gould, D. Orban and Ph.L. Toint,
Computational Optimization and Applications \fB60\fP:3, pp.545-557, 2014.
Expand All @@ -133,7 +128,7 @@ Environment, revisited\fP,
ACM TOMS, \fB29\fP:4, pp.373-394, 2003.

\fICUTE: Constrained and Unconstrained Testing Environment\fP,
I. Bongartz, A.R. Conn, N.I.M. Gould and Ph.L. Toint,
I. Bongartz, A.R. Conn, N.I.M. Gould and Ph.L. Toint,
ACM TOMS, \fB21\fP:1, pp.123-160, 1995.

cutest_ugr(3M), cutest_cgr(3M), sifdecoder(1).
106 changes: 106 additions & 0 deletions man/man3/cutest_csgrp.3
Original file line number Diff line number Diff line change
@@ -0,0 +1,106 @@
'\" e @(#)cutest_csgrp v1.5 03/2017;
.TH cutest_csgrp 3M "30 Mar 2017" "CUTEst user documentation" "CUTEst user documentation"
.SH NAME
CUTEST_csgrp \- CUTEst tool to evaluate the sparsity pattern of the
constraints gradients and gradient of objective/Lagrangian function.
.SH SYNOPSIS
.HP 1i
CALL CUTEST_csgrp( status, n, nnzj, lj, J_var, J_fun )
.SH DESCRIPTION
The CUTEST_csgrp subroutine evaluates sparsity pattern used when storing the
gradients of the general constraints and of either the objective function
or the Lagrangian function
.EQ
l(x,y) = f(x) + y sup T c(x)
.EN
corresponding to the problem decoded from a SIF file by the script
\fIsifdecoder\fP.

The problem under consideration
is to minimize or maximize an objective function
.EQ
f(x)
.EN
over all
.EQ
x
.EN
\(mo
.EQ
R sup n
.EN
subject to
general equations
.EQ
c sub i (x) ~=~ 0,
.EN
.EQ
~(i
.EN
\(mo
.EQ
{ 1 ,..., m sub E } ),
.EN
general inequalities
.EQ
c sub i sup l ~<=~ c sub i (x) ~<=~ c sub i sup u,
.EN
.EQ
~(i
.EN
\(mo
.EQ
{ m sub E + 1 ,..., m }),
.EN
and simple bounds
.EQ
x sup l ~<=~ x ~<=~ x sup u.
.EN
The objective function is group-partially separable and
all constraint functions are partially separable.
.LP
.SH ARGUMENTS
The arguments of CUTEST_csgrp are as follows
.TP 5
.B status \fP[out] - integer
the outputr status: 0 for a succesful call, 1 for an array
allocation/deallocation error, 2 for an array bound error,
3 for an evaluation error,
.TP
.B n \fP[in] - integer
the number of variables for the problem,
.TP
.B nnzj \fP[out] - integer
the number of nonzeros in J_var and J_fun,
.TP
.B lj \fP[in] - integer
the actual declared dimensions of J_var and J_fun,
.TP
.B J_var \fP[out] - integer
an array whose i-th component is the index of the variable with
respect to which the derivative is taken,
.TP
.B J_fun \fP[out] - integer
an array whose i-th component is the index of the problem function
whose derivative is taken. J_fun(i) = 0 indicates the
objective or Lagrangian function, while J_fun(i) = j > 0 indicates the
j-th general constraint function.
.LP
.SH AUTHORS
I. Bongartz, A.R. Conn, N.I.M. Gould, D. Orban and Ph.L. Toint
.SH "SEE ALSO"
\fICUTEst: a Constrained and Unconstrained Testing
Environment with safe threads\fP,
N.I.M. Gould, D. Orban and Ph.L. Toint,
Computational Optimization and Applications \fB60\fP:3, pp.545-557, 2014.

\fICUTEr (and SifDec): A Constrained and Unconstrained Testing
Environment, revisited\fP,
N.I.M. Gould, D. Orban and Ph.L. Toint,
ACM TOMS, \fB29\fP:4, pp.373-394, 2003.

\fICUTE: Constrained and Unconstrained Testing Environment\fP,
I. Bongartz, A.R. Conn, N.I.M. Gould and Ph.L. Toint,
ACM TOMS, \fB21\fP:1, pp.123-160, 1995.

cutest_csgr(3M), sifdecoder(1).
Loading

0 comments on commit d1507be

Please sign in to comment.