From: Simon Dürr (
Date: Mon Sep 21 2015 - 13:39:30 CDT


I'm combining multiple volumetric maps calculated on different GPUs
for different segments of the trajectory of my system
>package require volutil #returns 1.3
>volutil -combinepmf gpu1.dx gpu2.dx gpu3.dx gpu4.dx -o combine1.dx

I want to have the arithmetic mean of all 3 runs I did with the same
system ( I have 3x combine*.dx files which differ a little bit)
I gave the same command a try but it gave a different result than
doing the calculation manually using Matlab.
>volutil -combinepmf combine1.dx combine2.dx combine3.dx -o final_volumetric_map.dx

After looking into the combine.dx files I found the last line to always be this:
"object "created by volutil, T = -1.00 Kelvin," class field"
I think it should be something like this:
object "ligand pmf [kT], 10000 frames, T = 300.00 Kelvin, maxenergy =
85, cutoff = 10.00A" class field

I had a look into molfile.C and binary_ops.C but as I'm not that
familiar with C l I couldn't really identify the cause of the problem.
After having added the "object "ligand ..." correctly at the end of
the DX object I get the same result from combinepmf and Matlab.
I guess this is because there is a variable called weight read in from
the last line of the DX which by default is set to one which accounts
for the difference of the two resulting DX maps.

Are you guys aware of this behaviour or is this a bug in volutil?