COMBINE_COLVAR BUG: from Google Group (2017 and 2012)

[KanySUI]

Hello,

I have noticed a bug where COMBINE_COLVAR incorrectly reports COLVAR values.

For example, one would expect two COLVARs below to be the same:

   &COLVAR
    &COORDINATION
     ATOMS_FROM 1
     KINDS_TO H
     NN 9
     ND 14
     R0 [angstrom] 1.5
    &END COORDINATION
   &END COLVAR
 

and

   &COLVAR
    &COMBINE_COLVAR
     &COLVAR
      &COORDINATION
       ATOMS_FROM 1
       KINDS_TO H
       NN 9
       ND 14
       R0 [angstrom] 1.5
      &END COORDINATION
     &END COLVAR
 
      FUNCTION CV_combined
      VARIABLES CV_combined
    &END COMBINE_COLVAR
   &END COLVAR

However, metadynLog reports correct CV1, while CV2 (combined CV) is listed as 0.

I think this bug was reported earlier at https://groups.google.com/forum/#!searchin/cp2k/COMBINE_COLVAR%7Csort:relevance/cp2k/_uFtNhs8ptY/la3vSdiX7I4J but apparently it is not fixed yet. I would like to confirm that this bug is (at least) specific to COMBINE_COLVAR, where COORDINATION is used together with ATOMS_FROM and KINDS_TO.

I have attached an input file that would reproduce this result. It should take less than a minute to run on a single core.

Source:

colvar_test.txt https://groups.google.com/forum/#!searchin/cp2k/COMBINE_COLVAR|sort:date/cp2k/TaTgJuSrpmU/LaFsW5xHAwAJ https://groups.google.com/forum/#!searchin/cp2k/COMBINE_COLVAR%7Csort:relevance/cp2k/_uFtNhs8ptY/la3vSdiX7I4J