[molpro-user] Molpro 2012.1.13 contest_lih2_D0D1

[Scott Brozell]

Hi,

With two different intel compiler versions and in serial and with two
different MVAPICH versions the contest_lih2_D0D1 error magnitude check failed.

contest_lih2_D0D1.errout
...
 ENERGY(1:2)      = [      -8.37852681    -8.37852648] HARTREE
 DMZ(1:2)         = [       1.60195582    -0.45286133] AU
 SETTING EOLD(1)        =        -8.37852640                                  
 SETTING DIPOLD(1)      =        -0.45286414                                  
 SETTING EOLD(2)        =        -8.37852639                                  
 SETTING DIPOLD(2)      =         1.60195639                                  
 SETTING DE(1)          =        -0.00000041  HARTREE                         
 SETTING DE(2)          =        -0.00000009  HARTREE                         
 SETTING DD(1)          =         2.05481996  AU                              
 SETTING DD(2)          =        -2.05481772  AU                              
 SETTING DEMAX          =         0.00000041  HARTREE                         
 SETTING DDMAX          =         2.05481996  AU                              
 *** ERRORS DETECTED in contest_lih2_D0D1.test

contest_lih2_D0D1.test
...
show,energy,dmz
eold(1)=-8.37852640
dipold(1)=-0.45286414
eold(2)=-8.37852639
dipold(2)=1.60195639
de(1)=energy(1)-eold(1)
de(2)=energy(2)-eold(2)
dd(1)=dmz(1)-dipold(1)
dd(2)=dmz(2)-dipold(2)
demax=max(abs(de(1)),abs(de(2)))
ddmax=max(abs(dd(1)),abs(dd(2)))

if(demax.lt.1.d-7.and.ddmax.lt.1.d-4) then
text,no errors in contest_lih2_D0D1.test
else
text,ERRORS DETECTED in contest_lih2_D0D1.test
endif


It looks like this check could/should be rewritten to avoid
this false positive.  I can provide more details if necessary.

thanks,
scott