Hi Scott,
thanks for bringing this up. It's really a GB parameter update (not really
a "bug fix"), so it's hard to say if the old code should be changed or not.
In principle the new params are "better", but this is really an example of
problems arising from having force field info embedded into the actual code
base. GB has been that way for a long time, which leads to many other
problems (such as it being too easy to choose input flags like igb that are
inconsistent with what's in the prmtop). It might be good to have a broader
discussion about how to more cleanly separate force field parameters from
the code.
carlos
On Fri, May 26, 2023 at 2:03 PM Scott Brozell via AMBER-Developers <
amber-developers.ambermd.org> wrote:
> Hi,
>
> possible FAILURE: check mdout.trxox.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gb8_trx
> 45c45
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
>
> These are due to a released AmberTools23 with updated gb8 defaults in the
> code and the tests, but Amber22 pmemd does not have an update that corrects
> its defaults, so when the AmberTools23 test case is run with Amber22 pmemd
> failures occur.
>
> Shouldn't we release an update so the tests pass (see below) and to be
> consistent with the manual; it is probably just this:
>
> git diff c50f15d9ec80edc3e8e4fd2ee2b7186a3cc8ab57
> 6fdfc2427cbcab5e7e143015c6b083cf03271c83 src/pmemd/src/mdin_ctrl_dat.F90
>
>
> On Sat, May 06, 2023 at 05:11:13PM -0400, David A Case via AMBER wrote:
> > On Fri, May 05, 2023, Francesco Trozzi via AMBER wrote:
> > >
> > >I am currently installing Amber 22 on an Ubuntu machine.
> > >During testing of the serial CPU version and serial CUDA version I
> > >encountered the following possible failures (link to the diff and log
> > >files
> >
> > These are pretty innocuous errors. The ones marked "ignored" are known
> to
> > us, but we don't think they involve any real problems. The two others
> > involve a change in defaults (which somehow slipped by our checks) and a
> > roundoff error.
>
> make[2]: Leaving directory `/amber/gnu/11.2/mvapich2/2.3/22/test'
> 162 file comparisons passed
> 7 file comparisons failed (2 of which can be ignored)
> 0 tests experienced errors
>
> possible FAILURE: check mdout.trxox.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gb8_trx
> 45c45
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
> possible FAILURE: check mdout.trxox_md.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gb8_trx
> 47c47
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
> possible FAILURE: check mdout.trxox_md.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gb8_trx
> 47c47
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
> possible FAILURE: check mdout.1hji_md.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gbneck2nu/1hji
> 51c51
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
> possible FAILURE: check mdout.3p4a.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/gbneck2nu/modified_nu/3p4a
> 52c52
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
> possible FAILURE: (ignored) check mdout.dif
> /amber/gnu/11.2/mvapich2/2.3/22/test/phmd/implicit
> 139c139
> < gbalphaP = 0.41836, gbbetaP = 0.29005, gbgammaP =
> 0.10642
> ---
> > gbalphaP = 1.00000, gbbetaP = 0.80000, gbgammaP =
> 4.85000
> ---------------------------------------
>
> scott
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri May 26 2023 - 12:30:03 PDT
