[gmx-developers] bug in h-bond constraints for shake?

[Michael Shirts]

I've found a bug with h-bond constraints using shake with any
integrator -- compiling on one processor OS X 10.6, gcc, double or
single precision, I get output from mdrun like what is given at the
end of this email.

If this ends up being a specific problem with my input files, rather
than a general problem with the combination of the two options, I will
provide specific input files!

Best,
Michael

Sample output:
**************************************************
Double sids (0, 1) for atom 12
Double sids (0, 1) for atom 13
Double sids (0, 1) for atom 14
Double sids (0, 1) for atom 15
Double sids (0, 1) for atom 16
Double sids (0, 1) for atom 17
Double sids (0, 1) for atom 18
Double sids (0, 1) for atom 19
Double sids (0, 1) for atom 20
Double sids (0, 1) for atom 21
Double sids (0, 1) for atom 22
Double sids (0, 1) for atom 23
Double sids (0, 1) for atom 24
Double sids (0, 1) for atom 25
Double sids (0, 1) for atom 26
Double sids (0, 1) for atom 27
Double sids (0, 1) for atom 28
Double sids (0, 1) for atom 29
Double sids (0, 1) for atom 30
Double sids (0, 1) for atom 31
Double sids (0, 1) for atom 32
Double sids (0, 1) for atom 33
Double sids (0, 1) for atom 34

-------------------------------------------------------
Program mdrun, VERSION 4.0.99_development_20090927
Source code file: invblock.c, line: 79

Fatal error:
Double entries in block structure. Item 35 is in blocks 1 and 0
 Cannot make an unambiguous inverse block.
For more information and tips for trouble shooting please check the
GROMACS website at
http://www.gromacs.org/Documentation/Errors
-------------------------------------------------------