[gmx-users] Grompp error in 4.x regarding dihedral multiplicity

[Mark Abraham]

Sarah Witzke wrote:
> Dear gromacs users,
> 
> I'm doing simulations of small molecules in lipid bilayers. Doing the simulations with a DMPC bilayer works fine for all 4.x versions of gromacs. But when I try to do an energy minimization of a POPC bilayer, grompp gives the following error:
> 
> /people/disk2/sarah/gromacs-4.0.3/bin/grompp -f em.mdp -c system.gro -p popcpalc32.top -o test.tpr
> 
> Ignoring obsolete mdp entry 'cpp'
> checking input for internal consistency...
> processing topology...
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/ffgmxnb.itp
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/ffgmxbon.itp
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/ff_dum.itp
> Generated 1502 of the 2346 non-bonded parameter combinations
> Generating 1-4 interactions: fudge = 1
> Generated 1068 of the 2346 1-4 parameter combinations
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/flexspc.itp
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/spc.itp
> Opening library file /people/disk2/sarah/gromacs-4.0.3/share/top/ions.itp
> Excluding 3 bonded neighbours molecule type 'PALC'
> turning all bonds into constraints...
> Excluding 3 bonded neighbours molecule type 'POPC'
> turning all bonds into constraints...
> Excluding 2 bonded neighbours molecule type 'SOL'
> turning all bonds into constraints...
> processing coordinates...
> double-checking input for internal consistency...
> renumbering atomtypes...
> converting bonded parameters...
> 
> -------------------------------------------------------
> Program grompp, VERSION 4.0.3
> Source code file: convparm.c, line: 68
> 
> Fatal error:
> Value of 'multiplicity' in Proper Dih. is 0, which is smaller than the minimum of 1
> -------------------------------------------------------
> 
> When looking through the lipid .itp file (popc.itp) I do see dihedrals with multiplicity of 0 (which is not the case in dmpc.itp) - I haven't made it myself but it has been used many times before.
> 
> When running grompp in version 3.3.3 I get no error message:

Yup - there was no test for validity of the input.

Since a proper dihedral with multiplicity of zero is just a constant 
(see manual eq 4.61), it does nothing at all to the forces, nothing 
meaningful to energies and nothing to energy differences. Other than a 
convenient way to "comment out" dihedrals (which would be better done 
with a real comment!) I can't see the purpose of such dihedrals of 
multiplicity zero.

Mark