Compute fep pair style mliap param scale not supported

LAMMPS
1

Problem Description

I’m trying to use LAMMPS to compute hybrid/overlay potentials, but encountering an error when using compute fep. My input script combines mliap and lj/cut/soft pair styles, but I get an error when attempting free energy perturbation calculations with mliap.

Error Message

ERROR: Compute fep pair style mliap param scale not supported (src/FEP/compute_fep.cpp:215)

Input Script

variable        ONE equal 1
pair_style      hybrid/overlay mliap unified ./mace-mpa-0-medium.model-mliap_lammps.pt 0 lj/cut/soft 1.000000 0.500000 6.000000
pair_coeff      * * mliap Zn Cl O H
pair_coeff      1 1 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      1 2 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      1 3 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      1 4 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      2 2 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      2 3 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      2 4 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      3 3 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      3 4 lj/cut/soft 0.010000 3.000000 0.500000
pair_coeff      4 4 lj/cut/soft 0.010000 3.000000 0.500000
fix             tot_pot all adapt/fep 0 pair mliap scale * * v_LAMBDA
compute         e_diff all fep ${TEMP} pair mliap scale * * v_ONE
variable        e_diff equal c_e_diff[1]

Do you have any solutions to this problem?

2

Where in the LAMMPS documentation does it say that pair style mliap is compatible with compute fep or fix adapt/fep?

To make it compatible some C++ programming is required. How much depends on the individual pair style implementation.

3

Thank you for your reply. So, are you saying that this issue stems from the MLIAP interface not supporting FEP? Because I have consulted relevant materials, but I have not found any explanation of MLIAP supporting FEP, nor am I sure which software packages it can be used with.

4

The error message says that, too.

You didn’t do a good job at that. Both, the documentation for compute fep, and of fix compute adapt/fep contain a list of supported pair styles. mliap is not on it. It cannot get more definite than that.

P.S.: it is unrealistic to expect a code as flexible as LAMMPS to list all permutations of features that are NOT supposed to be combined. There are far too many of those and it is impossible to maintain and update this in a meaningful way given the rate at which new features are added to LAMMPS.