Disablee PME tuning with the group scheme¶
PME tuning with the group cut-off scheme did not work, because a critical flag was not set. Thus, interactions between charge-group pairs at distances between rlist and rcoulomb were missing. The search cost does go up, so the PME tuning does not produce very large cut-offs and grid spacings. This limits the error, but also the chance of detection.
PME tuning is now completely disabled with the group scheme.
Fixed value of Ewald shift¶
In all the short-ranged kernel flavours, a very slightly wrong shift was used. They would often cancel in practice, and energy differences computed on neighbour lists of the same number of particles will have the error cancel. The difference doesn’t even show up in the regressiontests.
Fixed flat-bottomed position restraints with multiple ranks¶
Reallocation was never done for flat-bottomed restraints, so the indexing could go out of range, leading to segfaults.
Fixed handling of virials from previous steps¶
An appending restart of md-vv plus nose-hoover plus no pressure coupling truncated off a correct edr frame and wrote one with zero virial and wrong pressure. And in the same case, a no-append restart wrote a duplicate frame that does not agree with the one written before termination.
Clarified use of tau-p and pcoupltype¶
grompp used to permit “tau-p = 5 5” which is misleading. The recent fix for Issue 1893 leads to the user receiving a grompp warning, so this improves the docs to make clear that pressure coupling is not implemented like temperature coupling (where different time constants might be used).
Corrected kernel launch bounds for CUDA sm_60¶
The GP100 architecture has half the SM size (64 ALUs) compared to 5.x hence 64 threads/block already achieves maximum occupancy and the incorrectly tweaked launch configuration is not optimal. This change reverts a previous incorrect increase in block size.
Increased the default reset step to 1500 in gmx tune_pme¶
Fixed incorrect charge in HISE.¶
Another fix in 5.1.4 introduced an incorrect charge on HISE CG, which leads to a fractional charge on the residue. Any simulation with this charge will be wrong.
Fixed crash in gmx order¶
gmx order used a cumbersome floating point method to compute an index in a histogram leading to index -1. The present code is simpler and robust, in fact the old code was likely wrong.
Fixed gmx rmsf -q -oq.¶
The wrong coordinates from the -s file were being used instead of those from the -q file. gmx rmsf -oq was otherwise fine.
Made editconf B-factor attachment more useful in practice¶
B-factor values will be added to residues unless an index is larger than the number of residues or an option is specified. Protein residue indices can start from any number and, in case they start from a large number, there is no way to add B-factor values to residues.