Coarse‐graining With the Relative Entropy

Shell, M. Scott

doi:10.1002/9781119290971.ch5

Cited by 70 publications

(63 citation statements)

References 81 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Whereas the choice of force-matching approach does not guarantee the recovery of individual pair correlation functions derived from full atomistic trajectories [7,27], but we have observed that with simple functional forms most of the error arises from the lack of complex terms in the potential of mean force, and the use of neural po-tentials recovers . To include the learning of structural cross-correlations, our method can optimized to incorporate iterative force matching [27] and relative entropy [32].…”

Section: Discussionmentioning

confidence: 99%

Coarse-graining auto-encoders for molecular dynamics

2019

View full text Add to dashboard Cite

Molecular dynamics simulations provide theoretical insight into the microscopic behavior of materials in condensed phase and, as a predictive tool, enable computational design of new compounds. However, because of the large temporal and spatial scales involved in thermodynamic and kinetic phenomena in materials, atomistic simulations are often computationally unfeasible. Coarsegraining methods allow simulating larger systems, by reducing the dimensionality of the simulation, and propagating longer timesteps, by averaging out fast motions. Coarse-graining involves two coupled learning problems; defining the mapping from an all-atom to a reduced representation, and the parametrization of a Hamiltonian over coarse-grained coordinates. Multiple statistical mechanics approaches have addressed the latter, but the former is generally a hand-tuned process based on chemical intuition. Here we present Autograin, an optimization framework based on autoencoders to learn both tasks simultaneously. Autograin is trained to learn the optimal mapping between all-atom and reduced representation, using the reconstruction loss to facilitate the learning of coarse-grained variables. In addition, a force-matching method is applied to variationally determine the coarse-grained potential energy function. This procedure is tested on a number of model systems including single-molecule and bulk-phase periodic simulations.

show abstract

Section: Discussionmentioning

confidence: 99%

Coarse-graining auto-encoders for molecular dynamics

2019

View full text Add to dashboard Cite

show abstract

“…While many multiscale strategies exist that hybridize quantum-classical or discrete-continuum models, the emphasis of our work is on algorithms that combine FG and CG molecular models of Newtonian particles. We continue by discussing the two main routes for implementing this multiscale approach, which are the in-serial [9][10][11][12][13][14][15][16] and in-parallel [17][18][19][20][21][22][23][24] formats.…”

Section: Introductionmentioning

confidence: 99%

Relative Resolution: A multipole approximation at appropriate distances

Chaimovich

Kremer

Peter

2019

Phys. Rev. Research

View full text Add to dashboard Cite

Recently, we introduced Relative Resolution (RelRes) as a hybrid formalism for fluid mixtures [Chaimovich et al., J. Chem. Phys. 143, 243107 (2015)]. The essence of this approach is that it switches molecular resolution in terms of relative separation: While nearest neighbors are characterized by a detailed fine-grained model, other neighbors are characterized by a simplified coarse-grained model. Once the two models are analytically connected with each other via energy conservation, RelRes can capture the structural and thermal behavior of various multicomponent and multiphase systems across state space. This current work is a natural continuation of our original communication. Most importantly, we present the comprehensive mathematics of RelRes, casting it as a multipole approximation at appropriate distances; the current set of equations technically applies for any arbitrary system in soft matter (e.g., water). Besides, we continue examining the capability of this multiscale approach in molecular simulations of various (nonpolar) uniform liquids, specifically examining a 2:1 mapping for dumbbell-like molecules, as well as a 6:1 mapping and a 6:2 mapping for butterflylike molecules. In turn, we exhaustively show that RelRes can successfully retrieve for these systems their static and dynamic behavior, given that the fine-grained and coarse-grained potentials are switched at the boundary between the first and second coordination shells, the location at which orientational correlations vanish. We finally conclude by discussing how RelRes is the inherent variant of the "cell-multipole" approach for molecular simulations and, thus, this multiscale framework is especially promising for studying biological systems.

show abstract

“…An interesting question is how far this goes: Can the method demonstrated in this paper be used more generally for identifying a relevant coarse-graining leading to pseudoisomorphs in the phase diagram? If yes, how does this connect to the current approaches to coarse graining for which important progress has recently been made [62][63][64][65][66][67]?…”

mentioning

confidence: 99%

Communication: Pseudoisomorphs in liquids with intramolecular degrees of freedom

Olsen

Dyre

Schrøder

2016

The Journal of Chemical Physics

View full text Add to dashboard Cite

Computer simulations show that liquids of molecules with harmonic intramolecular bonds may have "pseudoisomorphic" lines of approximately invariant dynamics in the thermodynamic phase diagram. We demonstrate that these lines can be identified by requiring scale invariance of the inherent-structure reduced-unit low-frequency vibrational spectrum evaluated for a single equilibrium configuration. This rationalizes why generalized excess-entropy scaling, density scaling, and isochronal superposition apply for many liquids with internal degrees of freedom.

show abstract

Coarse‐graining With the Relative Entropy

Cited by 70 publications

References 81 publications

Coarse-graining auto-encoders for molecular dynamics

Coarse-graining auto-encoders for molecular dynamics

Relative Resolution: A multipole approximation at appropriate distances

Communication: Pseudoisomorphs in liquids with intramolecular degrees of freedom

Contact Info

Product

Resources

About