On Wed, Oct 25, 2017 at 4:16 PM, David Cerutti <dscerutti.gmail.com> wrote:
> I'm also looking into what we can do to remedy the problem of so many false
> positives for failures in the SPFP test suite. I ran both test suites
> before committing--what happened is that there is a baseline of 40+ test
> cases that "fail" in the SPFP mode, and in skimming those diffs I didn't
Here is what I use to quickly ascertain if something is going wrong
with the SPFP tests:
grep "###" <test diff file> | sort -g -k 9 | tail
If the largest error is on the order of 0.5 kcal/mol I usually assume
all is well. With the broken code the output of the above command is
(on a tesla k20m):
### Maximum relative error in matching lines = 3.45e-03 at line 106 field 3
### Maximum absolute error in matching lines = 4.20e-03 at line 147 field 3
### Maximum relative error in matching lines = 1.14e-02 at line 2046 field 4
### Maximum relative error in matching lines = 1.66e-02 at line 194 field 3
### Maximum absolute error in matching lines = 2.47e-02 at line 144 field 3
### Maximum absolute error in matching lines = 1.00e-01 at line 362 field 12
### Maximum absolute error in matching lines = 5.86e-01 at line 104 field 3
### Maximum absolute error in matching lines = 1.10e+00 at line 180 field 3
### Maximum relative error in matching lines = 4.79e+01 at line 263 field 3
### Maximum absolute error in matching lines = 4.09e+03 at line 245 field 3
This is a quick-and-dirty way to make sure things aren't falling
apart. We clearly need a better mechanism in the long run though.
-Dan
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Oct 26 2017 - 05:30:02 PDT
