PIMD metadynamics failed due to no access to "myatoms"

[Qinghua Liao]

Hello, 

I am trying to run PIMD metadynamics using i-PI 3.16 and ASE/MACE. 

The normal PIMD works well, but the metadynamics crashed, as

it could not obtain number of atoms from the system, even though I provided a xyz file. 

The attached are my input and log files, the metadyamics part of the input was written 

take a reference from this post: 

https://groups.google.com/g/ipi-users/c/aeFZdb_jOJw/m/0YnzBXcqBwAJ

For the installation, py_plumed and plumed were installed with ASE/MACE, it works well for my metadynamcis with ASE/MACE/PLUMED. 

Any suggestions are welcome and appreciated. 

All the best,

Qinghua

[Qinghua Liao]

Hello, 

After adding mode='xyz', and now the simulation is running: 

  <ffplumed name='opes'>
    <file mode='xyz'> CH4_1hx323.xyz </file>
    <plumed_dat> plumed.dat </plumed_dat>
  </ffplumed>

However, there is no output from metadynamics. I will test it and try to figure it out, I read something from the online documentation. 

All the best,

Qinghua

[Qinghua Liao]

I found that if I put this second in the input file, pimd.xml: 

  <smotion mode="metad">
     <metad> <metaff> [ opes ] </metaff> </metad>
  </smotion>

I can get some output files, but the simulation crashed when the first Gaussian was added.

What I learned is that smotion is for i-PI built-in metadynamics, which has some conflicts with external PLUMED. 

#

  File "/gpfs/home/ub/ub487110/programs/i-PI_316/ipi/engine/smotion/metad.py", line 104, in step
    for fb in fc._forces:
              ^^^^^^^^^^
AttributeError: 'ScaledForceComponent' object has no attribute '_forces'
WARNING: internal inconsistency in allocated PlumedMain instances (1)

#

I would really appreciate if someone could give me some tips! 

All the best,

Qinghua

[Qinghua Liao]

I found the examples from "example/features/metadynamics", will study those examples there. 

(I will delete this post later)

All the best,

Qinghua

[Michele Ceriotti]

You can also check this out https://atomistic-cookbook.org/examples/pi-metad/pi-metad.html

[Qinghua Liao]

Thanks a lot Michele! Now I get it working. 

All the best,

Qinghua