FMatch gets different results in development
JensWehner opened this issue · comments
For votca/votca#175
@JensWehner can you have a look at this?
Sure, no worries.
So the fmatch discrepancies come from the different accuracy of the read in trajectories.
The old trajectories had 9 digits for forces
The new ones have 6
Due to the unit conversion I could not exactly reproduce that. But with 9 digits precision in the forces I got changes of the same magnitude as shown before.
I cannot completely rule out, that fmatch was changed, but I think that the difference is due to the file parser and not some internal changes.
Can we update our test case to use 9 digits for the forces?
Fixed