Added TSSetConvergedReason() and TS_CONVERGED_USER to allow TSSolve() to be stopped on request by user.

Hg-commit: 2d77f6a8715713cc6c0d9af9aafe6bd83291366b

Changed ts_exact_final_time from boolean to enum.
TSAdapt can now modify the final time step to stop exactly at
the given final time.

Hg-commit: 2bbe8e4fc9d649fbb4ec4bc19259912d0788d4c2

PetscDrawBoxedString() returns height and width

Hg-commit: d8250334932c8842de68cfc0077c9b184db5f077

tsview for draw

Hg-commit: 3879d449010e169c4a1fd844921f113a463fb812

DMTS is now attached to DM through ->dmts and not with PetscObjectCompose()

Hg-commit: 73e222393cbc1bfccc3195abf364ab09b1d64b9b

made DMTS a regular PetscObject instead of using a Container

Hg-commit: f37ce877caff68186ff449a07f72967013d39c1a

fixed naming of KSPDM, SNESDM, TSDM and child objects, fixed DMXXYYYContextZZZ() to DMGet/CopyDMXXX()

Hg-commit: a8931f032682db8d679ab51da5c3819b9c32b632

Remove unused variables

Hg-commit: 0cabef41af4dc9e87b11f3e99891357220ec8403

TS should try really hard to generate the matrix if not explicitly provided

Hg-commit: 991ee090b6e6528c19ef0b4f4722c2b77d7ef61f

TSSolve() did not properly handle null u argument
more work on TSView() binary and TSLoad()

Hg-commit: 85c2341b3eb28d6cdaa697c3b1e99963abb9edbf

added stubs for TS/SNES/KSP/PC Load and binary view, no working impls yet

Hg-commit: 7dfdadcaf4660e2aa05346cf920006cd46b990b6

Prototype for TSGetSolveTime() and fix some man page refs to TSGetConvergedReason

Hg-commit: 44b6a6cd91796d32b18d6c64337bbc9828475062

removed ftime from TSSolve() and added TSGetSolveTime()

Hg-commit: dd5658b5e51af8b3e1732c3368cf2c47376c68a6

added to TSStep() documentation that it may overstep final time set with TSSetDuration()

Hg-commit: ef7e41477b24a19ad09c58587c7789770a92c7aa

Factor ts->time_step out of "shift" field and switch -ts_monitor_sp_eig

TS implementations should not store the entire shift internally, and
instead should use ts->time_step directly. TSMonitorSPEig

Factor ts->time_step out of "shift" field and switch -ts_monitor_sp_eig

TS implementations should not store the entire shift internally, and
instead should use ts->time_step directly. TSMonitorSPEig() computes the
"steady-state" Jacobian by manipulating ts->time_step before calling
SNESComputeJacobian().

Hg-commit: 28c7df5560d100a222655100368151915451540f

show more ...

documentation for -1 step in TSMonitor

Hg-commit: 0cc987c6440b0a0b89061c56f216f0e7314531fd

Document change to -ts_monitor_solution, update examples

Hg-commit: 227f175daa1a06a113ecef4a7ab224aa1fd2dcd7

remove silly shift in tseig

Hg-commit: 5c2c8f943bbe46f0a240f6d30c6166208ea5c9f7

SNESTSFormJacobian_ARKIMEX() always passed imex of true, now pass ark->imex
fixed bug that ARKIMEX didn't work for fully implicit because of above

Hg-commit: 4d3f2a85abbbbf18fa09dfab2dc3da9d18abb322

-ts_monitor_sp_eig now predicts instability baed on eigenvalues

Hg-commit: 8fa1d1000762d968a2bd809d5b6862d79ec92aaf

added TSComputeLinearStability() for Euler and theta, use it in graphics
cleaned up formating in rosw and arkimex files

Hg-commit: 553e2c2ebaaa83a3a28c43188c2ab6f232784856

-ts_monitor_sp_eig plots stability region of a method (hardcoded for test case)

Hg-commit: dfd38f6f6e51b1601f5c6ead5946f7bb9d06b03d

commit after merge

Hg-commit: d0f613618ca66a5aeb1aac97d210d2cc29755d3d

work on eig plotting

Hg-commit: 012a81e69eae35930c4e6020c99cc2cb92fc7fb0

reemove OBJSC and OBJSF from invididual makefiles and auto-define in conf/variables

Hg-commit: 2f01c17f3c19dc4b89aaf306cfd60a12312c270d