Closed JohannesKarwou closed 1 year ago
Hi,
We use $\unicode{xC5}$, eV, and $\text{eV/}\unicode{xC5}$, as shown at https://docs.deepmodeling.com/projects/deepmd/en/master/data/system.html.
openmm_deepmd_plugin
should be able to read it, but I am unsure if it can combine with DFTB3. I will contact @dingye18 for clarification.
Thanks for the information and the fast reply!
Currently, I was thinking about doing something like described here: https://github.com/JingHuangLab/openmm_deepmd_plugin/blob/master/python/tests/test_deepmd_npt.py, there I would not need a combination with DFTB3 or am I misunderstanding something...
GitHubContribute to JingHuangLab/openmm_deepmd_plugin development by creating an account on GitHub.
Hello, as far as I understand, the qdpi model is intended to be used in conjunction with DFTB3 as it helps bridge the accuracy gap between DFTB and ωB97X/6-31G*. Consequently, if you wish to incorporate the qdpi model into your simulation, DFTB3 is required.
Alternatively, are you specifically interested in running a simulation solely using the trained NNP model? If you want a simulation with NNP only, the Deepmd Plugin is enough to use.
Thanks for the answer! I would have two use cases in mind, in both cases I would want to simulate a box of water with one small solute (e.g. methanol) solvated in it.
a) Simulate the methanol with the qdpi model and the rest with a normal force field as described in the test_deepmd_fixed_dp_particles_reference
(https://github.com/JingHuangLab/openmm_deepmd_plugin/blob/master/python/tests/test_deepmd_dp_region.py) But if I understand you correctlyt this would be only possible by integrating DFTB3?
b) Simulate the same system solely with the NNP model as described in this test: test_deepmd_npt_reference
(https://github.com/JingHuangLab/openmm_deepmd_plugin/blob/master/python/tests/test_deepmd_npt.py). This might be a case which should be possible using the qdpi model?
GitHubContribute to JingHuangLab/openmm_deepmd_plugin development by creating an account on GitHub.
GitHubContribute to JingHuangLab/openmm_deepmd_plugin development by creating an account on GitHub.
DFTB3 should always be used with $\text{QD}\pi$. DFTB3 QM/MM should be used as the base in your system, and $\text{QD}\pi$ is applied to the QM region. If OpenMM cannot do QM/MM or DFTB3, I will recommend AMBER.
To running DFTB3/MM with OpenMM, you can refer to the psi4-openmm interface through this link. Besides, you can also try QM/MM with Q-Chem by following this repo.
However, I haven’t try the Integration of these QM/MM scripts with OpenMM Deepmd Plugin before. If you have any problems when using these scripts or Deepmd plugin, please let me know.
Thanks for the answer! I will have a look at the psi4-openmm interface and try if I can use it with the qdpi model.
I am not sure if this is the right place to pose my question, but if not you might guide me to the correct place.
I would like to use the
qdpi
model to try to run some dynamics usingopenmm
. I found theopenmm_deepmd_plugin
(https://github.com/JingHuangLab/openmm_deepmd_plugin) which, if I understand it correctly, would be able to read in adp_model_file
like the one provided for qdpi?For correctly setting up a system there, I need to provide these variables:
as stated in the README file of
openmm_deepmd_plugin
. Thecoord_coefficient
I can find out on my own, but I am not so sure what should be theforce_coefficient
and theenergy_coefficient
in this case, since I do not know which units are used for theqdpi
model. Since the original paper is giving everything inkcal/mol
andkcal/(molA)
I was wondering if these are the units also used in theqdpi.pb
model?