BNTR trust radius now projects the radius and prediction onto the bounds
New tests added for BNLS and BNTR
Bounded NTR added and working for unconstrained cases.
TaoComputeBounds fixed so that bounded algorithms can work on unconstrained problems with default infinity bounds
Bounded NLS working on plate2
Box constraint projections added to BNLS, and it still works for unconstrained problems. Bounded problems are not yet tested.
trust radius update kernel isolated
Unconstrained NLS transferred to bounded folder and structured to support code-reuse with other Newton-Krylov methods
Man pages: add newlines after "Notes:"This allows for proper formatting from sowing.On OS X (using gsed, not the default BSD sed), from the PETSc root directory: find src include -type f \(
Man pages: add newlines after "Notes:"This allows for proper formatting from sowing.On OS X (using gsed, not the default BSD sed), from the PETSc root directory: find src include -type f \( -name "*.c" -or -name "*.h" -or -name "*.cxx" \) | xargs gsed -i 's/Notes\s*:\s*\(\w.*\)/Notes:\n \1/'This adds a newline and 4 spaces whenever "Notes:" is followed by any "word" character, in any .c, .h, or .cxx file in src/ or include/
show more ...
added newline to header in order to resolve solaris compile warning
BNCG algorithm added and tested
projected gradient code fixed to obey recent convergence test changed
norm update bug fixed in GPGC, test output reverted
fix propagated throughout, convergence tolerance bug discovered and fixed in toy problem
Convergence check separated from TaoMonitor. Changed propagated to unconstrained algorithms. Untested.
Added newline at the end of file to squash a solaris build warning
projected gradient descent algorithm added and tested (squashed)
Wrapping up for a pull request
manual merge because our work flow for petsc/git SUCKSSSSSCommit-type: i.e. bug-fix, portability-fix, testing-fix,style-fix, feature, documentation, exampleFunded-by:Project:Time: hoursRepo
manual merge because our work flow for petsc/git SUCKSSSSSCommit-type: i.e. bug-fix, portability-fix, testing-fix,style-fix, feature, documentation, exampleFunded-by:Project:Time: hoursReported-by:Thanks-to:Development Tools: Vim, Emacs, Eclipse
Merge remote-tracking branch 'origin/stefano_zampini/fix-missing-locks-tao'
TAOBQPIP: inquire before calling MatGetDiagonalthe actual diagonal is not needed, since the algorithm uses MatDiagonalSet with ADD_VALUES
TAO: added matrix-free finite difference option for Hessian evaluation and VecLock{Push|Pop} around user callsAdded test for MFFD hessianFix several examples that improperly used VecGetArrayFix m
TAO: added matrix-free finite difference option for Hessian evaluation and VecLock{Push|Pop} around user callsAdded test for MFFD hessianFix several examples that improperly used VecGetArrayFix man pages and some coding style in tao code
Supply quad precision version of cofficients for RK TS schemeFixes ex9opt lack of convergence instantly.Presumably we should do this ASAP for all TS types
XXXDestroy() already checks if the pointer is NULL so no reason to check the pointer before calling the destroyCommit-type: style-fixFunded-by: ECPProject: ECP/Tao
Remove the use and definition of __FUNCT__ throughout the codeSince all modern C/C++ compilers provide this functionality we no longer need to provide it manually in PETScTime: 1.5 hoursThanks-t
Remove the use and definition of __FUNCT__ throughout the codeSince all modern C/C++ compilers provide this functionality we no longer need to provide it manually in PETScTime: 1.5 hoursThanks-to: Andreas Mang <andreas@ices.utexas.edu>
12345678910>>...13