mattwthompson / ilforcefields

Library of ionic liquid forcefields for use in molecular simulation

ilforcefields

Library of ionic liquid forcefields and builders for use in molecular simulation

Installation

Clone this repo, nagivate into it, and pip install -e .

Run tests with

py.test -v

ilforcefields's Issues

Dataframe of Ionic Liquid Simulations

Since @Argon1999 started adding more IL parameters, I've started thinking about keeping dataframe or spreadsheet that keeps track of the ionic liquid simulations that have been run and their physical properties.

The way I've though about handling this, is to have a Signac project, where the state points would be something like:

• forcefield
• cation
• anion
• T
• P

The process would be running the simulations and collecting simple physical properties such as density and diffusion coefficients. That way, we can easily write the results to a Pandas Dataframe and users can check that the FF XMLs have been implemented properly. Curious if you have any thoughts on this.

Roadmap to publication

We should aim to make this repo public and also produce some sort of associated paper. This issue serves to plan and document the steps needed to accomplish that.

• Journal (#15)

• Support more force fields

• Support more models (as in atomistic vs. CG. vs. UA, etc.)

• Automated tests

• Documentation (for users and contributors)

• Validation of force fields via comparing physical properties to values reported in papers

What else should we think about? Feel free to edit this issue if GitHub allows.

Update terms with published correction

A correction was issues to a few angle and torsion force constants: doi:10.1021/jp0476996

This is unlikely to affect densities and fix #1 but should be done nonetheless

Padua's polarizable FF

https://pubs.acs.org/doi/10.1021/acs.jctc.9b00689

Apple&P Support

Uses the Buckingham potential so would require waiting for it to be supported in foyer or some OpenMM-based workaround

Journal

Not sure where this paper should go. The work is probably not novel enough for our typical journals, and the software is arguably not novel either (in the sense that no new libraries are developed) so I'm not sure about JOSS and similar journals. Somewhere in between seems appropriate to me, although I do not know of any such journals.

Something is off with lopes.xml

Some validation runs of EMIM-Tf2N have produced densities lower than the authors reported. The non-bonded parameters are probably most important but various bonded terms probably also have an impact.

Currently I see no better way to verify parameters than elbow grease.

This also points to the general issue: we probably need a better way to validate from-scratch xml files vs reported properties in relevant paper(s).

Tests

Need to add some simple tests, in order of importance (and ease/ability to test)

Force fields

• Atoms are typed according to our expectations

• Ions have net integer charge

• Individual parameters are consistent with values reported in the papers

Other

• Helper(s) like GetIL do sane things

Mixing Rules

At some point, we need to check and confirm the appropriate mixing rules for each force field. CL&P uses Lorentz Berthelot, so I believe KPL and NGKPL should as well. But this is worth checking out.