Sundials upgrade v2

[ergoregion]

No description provided.

[rs028]

@ergoregion Looking at the differences in the test output. In several files, the numerical differeces are very small but there seems to be quite a bit of additional whitespace in the new files. Is this a change in sundials/cvode or something you introduced (accidentally or on purpose)?

See for example: firstorder, secondorder, static in model_tests. These are not chemical models, just simple ODE systems, so should be easy to understand.

[rs028]

Another note. the test using SPGMR without the Banded Preconditioner crashes with the following error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x7f8e254f22ed in ???
#1  0x7f8e254f1503 in ???
#2  0x7f8e25124f0f in ???
#3  0x7f8e24ecb770 in cvLsInitialize
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode_ls.c:1439
#4  0x7f8e24ec3164 in cvInitialSetup
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode.c:2078
#5  0x7f8e24ec3164 in CVode
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode.c:1125
#6  0x7f8e25ef3976 in __fcvode_mod_MOD_fcvode
	at AtChem-lib/sundials-7.5.0/src/cvode/fmod_int64/fcvode_mod.f90:2105
#7  0x55e89a6c25e8 in ???
#8  0x55e89a69b4ce in ???
#9  0x7f8e25107c86 in ???
#10  0x55e89a69b4f9 in ???
#11  0xffffffffffffffff in ???

[ergoregion]

Another note. the test using SPGMR without the Banded Preconditioner crashes with the following error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x7f8e254f22ed in ???
#1  0x7f8e254f1503 in ???
#2  0x7f8e25124f0f in ???
#3  0x7f8e24ecb770 in cvLsInitialize
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode_ls.c:1439
#4  0x7f8e24ec3164 in cvInitialSetup
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode.c:2078
#5  0x7f8e24ec3164 in CVode
	at AtChem-lib/sundials-7.5.0/src/cvode/cvode.c:1125
#6  0x7f8e25ef3976 in __fcvode_mod_MOD_fcvode
	at AtChem-lib/sundials-7.5.0/src/cvode/fmod_int64/fcvode_mod.f90:2105
#7  0x55e89a6c25e8 in ???
#8  0x55e89a69b4ce in ???
#9  0x7f8e25107c86 in ???
#10  0x55e89a69b4f9 in ???
#11  0xffffffffffffffff in ???

This failure is erratic, it doesn't always happen.
try changing Lookback to lookback (capital letter change) in /short_no_pre/model/configuration/solver.parameters.
If that works maybe check the case sensitivity in other parameter files.

EDIT: I was wrong, only the order matters, not the text. Sorry

[rs028]

Closing this, as all changes have been ported to PR #549 now.