CONSTANT pca_peig VALUES error

[mmfarrugia]

Hello,

I emailed the google group but I did not receive any replies or a confirmation of receipt, so I hope this is the correct place to raise this issue.

I did try to include the hash for the version, but it seems that the individual who installed it did so in an odd way because I get a 'dubious ownership' message, my apologies.
VERSION.txt indicates 2.11.0-dev only.

I am using PLUMED v2.11.0-dev, installed with AMBER 24.0. However, I am getting a rather odd error when running PCAVARS for a plumed metadynamics:

PLUMED: Action RMSD_VECTOR
PLUMED: with label pca_at
PLUMED: with arguments :
PLUMED: vector with label pca_at_pos
PLUMED: matrix with label pca_at_ref
PLUMED: added component to this action: pca_at.dist
PLUMED: added component to this action: pca_at.disp
PLUMED: calculating RMSD distance between 1 sets of 12899 atoms. Distance between vector pca_at_pos of atoms and matrix of configurations in pca_at_ref
PLUMED: method for alignment : OPTIMAL
PLUMED: chosen to use SQUARED option for MSD instead of RMSD
PLUMED: Action CONSTANT
PLUMED: with label pca_peig
PLUMED: reading in 2 by 38697 matrix
PLUMED: ERROR in input to action CONSTANT with label pca_peig : keyword VALUES could not be read correctly
PLUMED:
PLUMED:
PLUMED: ################################################################################
PLUMED:
PLUMED: (core/Action.cpp:309) void PLMD::Action::error(const string&) const
PLUMED: ERROR in input to action CONSTANT with label pca_peig : keyword VALUES could not be read correctly
PLUMED: ################################################################################

I can include the whole file, but line 309 of Action.cpp is just the error handling.

I narrowed down the origin via following the Action calls in the out file which are registered for the shortcut action PCAVARS.
PCAVARS, RMSD PDB2CONSTANT, CONSTANT, WHOLE_MOLECULES, POSITION, CONCATENATE, RMSD_VECTOR, CONSTANT, error

I am used to compiling languages, but I don't know C or C++ and am a first-time PLUMED user so thank you for your patience if this is not correct. However, I think I found the origin of the error in the PCAVars.cpp within the mapping namespace:

I would assume that it is either originating from pvec or coeff1/coeffs.

I am a PhD candidate in biophysics using an HPC so I don't have permission to modify or compile any of these resources. Thus, before I try installing and troubleshooting locally with an open access AMBER and some print statements on pvec and coeff1/coeffs, is this an issue which anyone else has encountered in this version? Additionally, does amber24 indeed come patched with plumed? This would help me convince the HPC to replace it with an official release.

Some background on the compute setup:
It was installed on the HPC at my university, which uses a module system, and I was told that plumed was pre-packaged with AMBER in the 2024 version. However, I can't seem to find documentation of AMBER coming pre-patched with plumed in the 2024 version and it seems unlikely to me that it would come pre-patched with a dev release of plumed so I think this may be incorrect. Normally I would just move to an official release of plumed but the HPC isn't keen on that option because their documentation indicates that it came with the amber24 installation.

Troubleshooting Context:
I already check that all the necessary files are present before running anything in my job queue setup. I did adapt this setup from gromacs to amber. I already had issues with environment variables (plumed kernel) and enabling of modules so we recompiled the code recently with all modules enabled and corrected the environment variables. Nothing popped on the make check according to HPC, but we did not run a make installcheck. When make-ing, I did notice that it warned it couldn't find the fftw library (hpc only loads the module within other modules but amber 24 should contain it).

Here is the text of my PLUMED input .dat file:
PCAVARS REFERENCE=plumed-mode-input.pdb TYPE=OPTIMAL LABEL=pca

METAD ...
LABEL=metad
ARG=pca.eig-1,pca.eig-2
PACE=500
HEIGHT=0.1
BIASFACTOR=10
SIGMA=0.001,0.001
FILE=plumed-mode-metadyn.hills
TEMP=300.0
... METAD

monitor the two variables and the metadynamics bias potential

PRINT ARG=pca.eig-1,pca.eig-2,pca.residual,metad.bias FILE=plumed-mode-metadyn.out STRIDE=500

I can also include the mode input pdb or link to a google drive folder where I put the whole directory if that would help.

Thank you for your help!

Mikaela Farrugia (she/her)
Biophysics PhD Candidate