Re: [AMBER] 【Error on testing benz_water_md.reaxff and openMPI】

From: David A Case <david.case.rutgers.edu>
Date: Wed, 29 Dec 2021 09:58:44 -0500

On Wed, Dec 29, 2021, Shang-Wei Li (TMU) wrote:

>CMake Error at
>/home/ikki/Documents/amber20_src/cmake/FixCondaShebang/FixCondaShebang.cmake:26
>(file): file failed to open for reading (No such file or directory):
> /home/ikki/Documents/amber20//bin/amber.pip
>
>I searched some articles from the mail list. Someone stated that it can be
>ignored.

That is correct: it can be ignored. [Does any volunteer have time to
cherry-pick 5bab12ccce92926c over to amber21-with-patches?]

>So, I tried to run the make test.parallel with export DO_PARALLEL="mpirun
>-np 8".
>However, I got lots of errors: failed to start the process. All the tests
>failed... Because I installed the openMPI version, will this be the issue?

Be sure that "which mpirun" points to the correct place. How did you
install openMPI? Can you run a simple parallel test program?
>
>Summary of AmberTools serial tests:
>2733 file comparisons passed
>3 file comparisons failed (2 of which can be ignored)
>1 tests experienced errors

This means that you have two errors, one non-ignored comparison and one
program error. Look for "Possible FAILURE" and "Program error" in the logs.

>The most important error is the failure on Reaxff...
>
>Program received signal SIGSEGV: Segmentation fault - invalid memory
>reference.
>Segmentation fault (core dumped)
> ./Run.benz_water_md.reaxff: Program error
>make[2]: *** [Makefile:795: test.sander.EXTERN] Error 1

I'll leave this one for the ReaxFF folks. They may want to know more
information about your OS and compilers.

>Creating the bad prmtop test.parm7
>Testing proper treatment of a bad prmtop; an error message is expected:
>ERROR: Flag "ATOM_NAME" not found in PARM file
>PASSED

You can see from the comments ("an error message is expected") that this is
not a test error.
>
>Can these errors be ignored? I really want to use Reaxff to simulate the
>bond forming and breaking between two small molecules. Additionally, can
>you give me some advice on how to analyze and visualize the results of
>Reaxff MD?

You probably want to pay attention to the ReaxFF test error above, although
you could try.

>Can QM/MM produce the trajectory file showing the bond forming and breaking
>in the trajectory file?

Yes.

....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Dec 29 2021 - 07:00:02 PST
Custom Search