Use the new API instead of a private function in adjoint RK

Funded-by:
Project: FASTMath adjoint
Time: 0.5 hours
Reported-by:
Thanks-to: Barry Smith <bsmith@mcs.anl.gov>

Use FD for the adjoint RK

- Works for both -snes_fd and -snes_fd_color.
- Querying for "SNESMatFDColoring" does not work for explicit methods since SNES is not used in the forward run; so check the

Use FD for the adjoint RK

- Works for both -snes_fd and -snes_fd_color.
- Querying for "SNESMatFDColoring" does not work for explicit methods since SNES is not used in the forward run; so check the Jacobian compute function instead.

Funded-by:
Project: FASTMath adjoint
Time: 2.0 hours
Reported-by:
Thanks-to: Barry Smith <bsmith@mcs.anl.gov>

show more ...

Add PetscArraycmp(), PetscArraycpy(), PetscArrayzero(), PetscArraymove(), update code to use new forms

Replace a couple multi-mallocs with a single malloc
Replace backwards loops in MatSetValues_*()

Add PetscArraycmp(), PetscArraycpy(), PetscArrayzero(), PetscArraymove(), update code to use new forms

Replace a couple multi-mallocs with a single malloc
Replace backwards loops in MatSetValues_*() with PetscArraymove()

Commit-type: style-fix, feature

show more ...

Man pages: remove .keywords: fields

This field is recognized by doctext from Sowing (http://wgropp.cs.illinois.edu/projects/software/sowing/doctext/node20.htm#Node20),
but it doesn't seem to be unif

Man pages: remove .keywords: fields

This field is recognized by doctext from Sowing (http://wgropp.cs.illinois.edu/projects/software/sowing/doctext/node20.htm#Node20),
but it doesn't seem to be uniformly used or maintained.

Thus, remove all .keywords: fields, and a following blank line, if present.

This is accomplished with GNU sed (gsed on OS X), with the following commands.
*Warning* that this type of command can corrupt a .git directory,
so be cautious in reusing or modifying these commands. They first look
for and delete matching lines with a following line consisting of only whitespace,
and then delete any remaining matching lines.

find src/ -type f -exec gsed -i '/keywords:/ {N; /\n\s*$/d}' {} +
find src/ -type f -exec gsed -i '/keywords:/d' {} +
find include/ -type f -exec gsed -i '/keywords:/ {N; /\n\s*$/d}' {} +
find include/ -type f -exec gsed -i '/keywords:/d' {} +

Hints on the sed command obtained from: https://unix.stackexchange.com/questions/100754/how-to-delete-a-specific-line-and-the-following-blank-line-using-gnu-sed

show more ...

use floating point coefficients for Verner's methods

added 6th, 7th, and 8th order robust Verner RK methods

Change names for Hessian functions

Funded-by:
Project: FASTMath adjoint
Time: 0.5 hours
Reported-by:
Thanks-to:

RK: remove check for MatDestroy()

Add TSAdjointSetForward and TSAdjointResetForward

AdjointRK: add support for integral in 2nd-order adjoint

Also fix the problem in calculating parametric 2nd-order adjoint sensitivity.

Funded-by:
Project: PETSc for Theta
Time: 8.0 hours
Reported-b

AdjointRK: add support for integral in 2nd-order adjoint

Also fix the problem in calculating parametric 2nd-order adjoint sensitivity.

Funded-by:
Project: PETSc for Theta
Time: 8.0 hours
Reported-by:
Thanks-to:

show more ...

Refactor quadrature

- Use a separate TS for quadrature
- Initially focus on Theta and Beuler
- Tested on ex3opt and ex9opt
- RK support is not complete
- Fixed memory leak. log_view for ex3sa shows

Refactor quadrature

- Use a separate TS for quadrature
- Initially focus on Theta and Beuler
- Tested on ex3opt and ex9opt
- RK support is not complete
- Fixed memory leak. log_view for ex3sa shows that there is one matrix that is not destroyed.

Funded-by:
Project: FASTMath adjoint
Time: 9.0 hours
Reported-by:
Thanks-to:

show more ...

Add second-order adjoint for ERK

Funded-by:
Project: FASTMath adjoint
Time: 4.0 hours
Reported-by:
Thanks-to:

Add TLM ERK

Funded-by:
Project: FASTMATH adjoint
Time: 4.0 hours
Reported-by:
Thanks-to:

AdjointRK: bug fix for RKs with zeros in b

- One of the problems was caused by FASL
- The other problem was caused by zeros in the b vector. The stage sensitivity vector cannot be simply set to zero

AdjointRK: bug fix for RKs with zeros in b

- One of the problems was caused by FASL
- The other problem was caused by zeros in the b vector. The stage sensitivity vector cannot be simply set to zero when b[i] is zero.
- Before this fix, the adjoint of RK works for only some methods such as 2a,3,4; now it works for all RK methods

Funded-by:
Project: FASTMath adjoint
Time: 2.0 hours
Reported-by:
Thanks-to:

show more ...

Optimize adjoint ERK and prepare for 2nd-order adjoint

Funded-by:
Project: PETSc for Theta
Time: 2.0 hours
Reported-by:
Thanks-to:

Replace deprecated adjoint functions with new ones

Change y to U for consistency

Funded-by:
Project: FASTMATH adjoint
Time: 1.0 hours
Reported-by:
Thanks-to:

Add missing ;

RK: add missing PetscFunctionBegin

Funded-by:
Project: PETSc time integration
Time:
Reported-by:
Thanks-to:

Fix warnings from BSD build

Funded-by:
Project: PETSc time integration
Time:
Reported-by: nightly test
Thanks-to: Satish Balay

Add missing CHKERRQ()

Funded-by:
Project: PETSc time integration
Time:
Reported-by: nightly test
Thanks-to:

RK: fix warnings caused by commit 474dd77

Rename MRK to RK_Multirate

Funded-by:
Project: PETSc time integration
Time: 0.5 hours
Reported-by:
Thanks-to: Barry Smith

MRK: put in PetscTryMethod

Funded-by:
Project: PETSc time integration
Time: 1.0 hours
Reported-by:
Thanks-to:

Add option -ts_use_splitrhsfunction

This option is added at TS level so that all the multirate methods can use it.

Funded-by:
Project: PETSc time integration
Time: 8.0 hours
Reported-by:
Thanks-to:

MRK: bug fix for more than 2 partitions

- The implementation has been improved significantly (much easier to understand)
- At each level, the time integration of the fast components is viewed as a r

MRK: bug fix for more than 2 partitions

- The implementation has been improved significantly (much easier to understand)
- At each level, the time integration of the fast components is viewed as a refinement to the solution at previous level
- The refinement is done recursively

Funded-by:
Project:
Time: 30.0 hours
Reported-by:
Thanks-to:

show more ...