GROMACS version:2021.4

GROMACS modification: No

Hi,

This is a little bit involved, please bear with me …

I was trying to emulate with gmx a kind of dual topology free-energy calculation,

using a simple ethane->methanol in vacuum test-case.

I first calculated single point energies for a particular conformer of each ethane, and methanol.

then did the same calculation for a hybrid structure with dual topology (CH3(-CH3)(-OH))

identical coordinates, free-energy = yes, delta-lambda = 0, and init-lambda-state

corresponding to either lamda=0 (ethane) or lambda=1 (methanol). So I have four cases:

- a normal simulation (single point energy) for ethane
- a normal simulation (single point energy) for methanol
- a simulation of the hybrid, free energy on, and lambda=0 (corresponding to ethane)
- a simulation of the hybrid, free energy on, and lambda=1 (corresponding to methanol)

Again, these are all only single point energies, and based on identical coordinates.

As expected the non-bonded energy terms are identical in cases 1 vs 3, as

well as in 2 vs 4. Also as expected, the bonded energy terms of case 3 and 4

are identical, and equal to the sum of the corresponding terms in case 1 and 2,

However the last statement is true only for the bond and the dihedral energies,

but not for the angle terms where the numbers clearly differ!

This came quite unexpected, I double and triple checked my input files, but found

nothing that would explain this …

I put the input files into an archive - which i cannot attach/upload here, therefore

I put the archive here: http://www.brunsteiner.net/angles.tgz from where it can be

downloaded - to reproduce the issue type:

# case1

gmx grompp -f md.mdp -c e.pdb -p e.top -o mde.tpr -maxwarn 1

gmx mdrun -v -deffnm mde -c mde-out.pdb

#case 2

gmx grompp -f md.mdp -c m.pdb -p m.top -o mdm.tpr -maxwarn 1

gmx mdrun -v -deffnm mdm -c mdm-out.pdb

#case 3

gmx grompp -f md0.mdp -c ems.pdb -p fe.top -o fe0s.tpr -maxwarn 1

gmx mdrun -v -deffnm fe0s -c fe0s-out.pdb

#case 4

gmx grompp -f md1.mdp -c ems.pdb -p fe.top -o fe1s.tpr -maxwarn 1

gmx mdrun -v -deffnm fe1s -c fe1s-out.pdb

I’d be grateful for any suggestions here!

cheers,

Michael