7
$\begingroup$

According to Wikipedia:

Functional forms and parameter sets have been defined by the developers of interatomic potentials and feature variable degrees of self-consistency and transferability. When functional forms of the potential terms vary, the parameters from one interatomic potential function can typically not be used together with another interatomic potential function.[19] In some cases, modifications can be made with minor effort, for example, between 9-6 Lennard-Jones potentials to 12-6 Lennard-Jones potentials.[9] Transfers from Buckingham potentials to harmonic potentials, or from Embedded Atom Models to harmonic potentials, on the contrary, would require many additional assumptions and may not be possible.

I haven't really understood this.

Can you give me a layman's definition of "Transferable" FF?

$\endgroup$
0

2 Answers 2

10
$\begingroup$

A forcefield defining the interactions of an atom is said to be transferrable if it can be successfully used in chemical simulations outside of where it was originally designed or fit.

For example: you are given interaction potential parameters for a particular model of water that recreates the correct experimental melting point and structure of water. The model is transferred to a simulation of a supersaturated solution of $\ce{NaCl}$ in water, with some interaction potential of $\ce{NaCl}$ defined using the same functional form (e.g. Lennard-Jones). The interaction parameters between water and the ions are then defined by some combining rules. If the simulation gives useful results, the water model is said to be transferrable to simulations of supersaturated $\ce{NaCl}$ in water.

Transferability basically refers to the ability of an interaction potential to be useful/accurate between different chemical environments.

$\endgroup$
4
$\begingroup$

I believe Hayden's answer is the correct one based on how the term transferability is typically used in this regard. However, the Wikipedia article you site seems to be describing a slightly different sense of the term in that paragraph. From what they describe, and the linked papers they cite, they seem to mean "transferability" as the ability to transfer parameters from one force field model to another. The example they give is taking the parameters used to fit a 9-6 Lennard-Jones potential and repurposing them to use in a 12-6 Lennard-Jones potential. So that paragraph seems to be more about whether the model parameters can be meaningfully converted between different models, rather than whether the force field is applicable for systems/properties beyond those for which the parameters were initially fit.

This is definitely nonstandard terminology, as neither of the papers Wikipedia cites use the term "transferability" to describe what they are doing.

$\endgroup$
2
  • $\begingroup$ I would say both are two sides of the same coin. You simply can't transfer a force field to a new chemical situation if there aren't well-defined mixing rules to generate all cross interactions. But at the same time, a force field that gives meaningless results in a new chemical situation can't really be said to be transferrable to that situation. $\endgroup$
    – Hayden S
    May 23 at 16:04
  • $\begingroup$ There are lots of well defined mixing rules for cross interactions, they just don't work well :S It would be interesting to train/fit all cross interactions explicitly and not use mixing rules. $\endgroup$
    – B. Kelly
    May 25 at 11:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.