Problem with the Curve-Fitting-Method-based consistency inspector

[julimox]

As suggested I move my post from Pymbar to here.

I am running pymbar 3.0 with the alchemical_analysis without problems except for the " -c " flag.

On the provided sample input data it is running fine but if I want to use it on my dhdl files I get the following mistake:

Plotting the CFM figure...
/opt/soft/anaconda-2.2.0/lib/python2.7/site-packages/matplotlib/axes/_base.py:2562: UserWarning: Attempting to set identical left==right results
in singular transformations; automatically expanding.
left=17.7338848684, right=17.7338848684
'left=%s, right=%s') % (left, right))
/opt/soft/anaconda-2.2.0/lib/python2.7/site-packages/matplotlib/axes/_base.py:2562: UserWarning: Attempting to set identical left==right results
in singular transformations; automatically expanding.
left=17.7624749232, right=17.7624749232
'left=%s, right=%s') % (left, right))

In case this is of interest: The dhdl-files were generated with Gromacs 5.0.4

My files are at

https://drive.google.com/folderview?id=0B2M9aqeJrxnYYTlsdU9jMnBQOGs&usp=sharing

Best regards,
Julian

[FEPanalysis]

Julian,

The attached files are of 1.0 ps, i.e. 6 snapshots, each which is not long enough to see what, if any, the problem is. What I can tell right away is that the "-c" analysis does not crash. Should they be of more reasonable length, I will be more specific.

[davidlmobley]

Yes, please send the input files to reproduce the problem.

Thanks.

On Thu, Oct 22, 2015 at 12:46 PM, FEPanalysis [email protected]
wrote:

Julian,

The attached files are of 1.0 ps, i.e. 6 snapshots, each which is not long
enough to see what, if any, the problem is. What I can tell right away is
that the "-c" analysis does not crash. Should they be of more reasonable
length, I will be more specific.

—
Reply to this email directly or view it on GitHub
#34 (comment)
.

David Mobley
Associate Professor
Department of Pharmaceutical Sciences
Department of Chemistry
3134B Natural Sciences I
University of California, Irvine
Irvine, CA 92697
[email protected]
work (949) 824-6383
cell (949) 385-2436

[julimox]

I had cut the files more than I wanted to - sorry for that.
The files updated to 100ps are now in the folder.

[julimox]

A kind reminder in case this issue is getting forgotten.

[davidlmobley]

Sorry about that. Pavel (@FEPanalysis) said he'll look at it next week. I
thought he'd put that on GitHub but I see now it was only to me.

Thanks and sorry for the delay.

On Wed, Oct 28, 2015 at 4:39 AM, julianzachmann [email protected]
wrote:

A kind reminder in case this issue is getting forgotten.

—
Reply to this email directly or view it on GitHub
#34 (comment)
.

David Mobley
[email protected]
949-385-2436

[FEPanalysis]

Julian,

After a quick look the problem seems to be in bad overlap -- not the actual one (the consistency between TI and BAR suggests there is no real problem) but as it is understood by the CFM function which produces empty boxes. Seems to be a bug that needs to be fixed. Thanks for the warning and test data. I'll be spending more time with it next week.