METHOD, APPARATUS AND COMPUTER PROGRAM FOR MOLECULAR SIMULATION

Patent №

US 7,908,129

Granted

2011-03-15

Filed 2008

Owner

FUJITSU LIMITED

Lab

AI components

4

ml · planning · evo · hardware

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

12222382

A method of simulating behaviour of a molecular system with m degrees of freedom over time comprising a partial momentum refreshment step and a molecular dynamics step, wherein the partial momentum refreshment step comprises: given a starting position q and a starting momentum p of the molecular system, partially refreshing the momentum to define refreshed momentum p′ evaluating the shadow Hamiltonian 331 Δt at position q and momentum p′; and accepting or rejecting the refreshed momentum p′ according to a Metropolis-type function and if p′ is accepted using p′ as the resulting momentum p and starting position q as the resulting position q or if it is rejected, using p as the resulting momentum p and starting position q as the resulting position; and wherein the molecular dynamics step comprises: given a starting position q and starting momentum p of the molecular system, running a molecular dynamics simulation over a fixed number of iterations and obtaining new position q′ and new momentum p′; evaluating the shadow Hamiltonian 332 Δt at position q′ and momentum p′ after the molecular dynamics simulation; and accepting or rejecting the new system configuration produced by the molecular dynamics simulation according to a Metropolis-type function and, if the new system configuration is accepted, using q′ as the resulting position q and p′ as the resulting momentum p or, if it is rejected, using the original starting position q as the resulting position q and negating the original starting momentum p to give the resulting momentum p; wherein either the partial momentum refreshment or the molecular dynamics step is the first step of the method, and the resulting position and resulting momentum of the first step provides the starting position q and starting momentum p for the next step.

AI classification

AI hardware0.96
Machine learning0.73
Planning0.66
Evolutionary computation0.60
Knowledge representation0.39
Natural language0.08
Vision0.01
Speech0.00

Ownership

FUJITSU LIMITED

assignment · 219170643

Assignors

AKHMATSKAYA, ELENA VITALIEVNA, NOBES, ROSS HOWARD, REICH, SEBASTIAN

On an employer assignment, the assignors are typically the inventors.

© 2026 NYSGPT2525 LLC