[gmx-users] gmx spatial producing negative isodensities.

[Stefan Hansen]

Dear all,

I have been attempting to obtain SDFs using gmx spatial however with the
results of getting negative isodensities. The trajectory contains 25000
frames, and ~10000 atoms.

Running gmx spacial with: gmx spatial -s mol_packed.gro -f mol.xtc -bin 0.1
and choosing the co-solute of interest to generate the SDF for and choosing
the system for group to output coords, yielded  the following:

Counts per frame in all 160380 cubes divided by -7.881814e-02
Normalized data: average 1.000000e+00, min -0.000000e+00, max -2.103577e+00

This test was conducted on 2 different machines 1 using gromacs version
2018.2 and second 2016.3.

Further we found that conducting the very same analysis however using a bin
size of 0.14 yields reasonable results, here meaning postive isodensities.

So my questions are:
Has this behaviour been observed before? i.e. is this a bug?
Am I using the gmx spatial wrong?

Regards,
Stefan