[CP2K-user] Regression test: how to check test result of wat_mode_sel_range.inp when np=1 or 2

[vitesse lin]

Dear all,


We run regression test two times with different number of processes under 
the same folder.

 

For the first time, np=1 and regress test reported wat_mode_sel_range.inp 
was a NEW one:

*wat_mode_sel_range.inp                                           
1736.718534    NEW (   9.61 sec)*

 

For second time, np=2 and regress test reported these:

*wat_mode_sel_range.inp                                              
7.162315 WRONG RESULT TEST 18*

*…*

*xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx*




* /home/mpich-ucx/TEST-local_cpu-psmp-2019-11-04_01-56- 
07/Fist/regtest-5-vib/wat_mode_sel_range.inp.out : MS| TRACKED FREQUENCY : 
old = 1736.718534 new = 7.162315 relative error : 2.41480055e+02 > 
numerical tolerance = 7e-08 
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx*

 

To investigate this WRONG issue, we run wat_mode_sel_range.inp 
independently.

When np=1, it reported:
*MS| TRACKED FREQUENCY (1) IS: 1736.718534 cm-1*
When np=2, it reported:

*MS| TRACKED FREQUENCY (1) IS: 1736.709744 cm-1*
* MS| TRACKED FREQUENCY (2) IS: 7.162315 cm-1*

So our questions are:
Did np=2 get reasonable results?

Is it reasonable to compare *TRACKED FREQUENCY (1) of np=1*  with *TRACKED 
FREQUENCY (2) of np=2* to check if wat_mode_sel_range.inp gets the right 
result or not? 


Thanks,

Vitesse.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191112/ebfbdb54/attachment.htm>