Efficient mapping of phase diagrams with conditional Boltzmann Generators

Machine Learning Science and Technology, Journal Year: 2024, Volume and Issue: 5(4), P. 045045 - 045045

Published: Oct. 8, 2024

Abstract The accurate prediction of phase diagrams is central importance for both the fundamental understanding materials as well technological applications in material sciences. However, computational relative stability between phases based on their free energy a daunting task, traditional estimators require large amount simulation data to obtain uncorrelated equilibrium samples over grid thermodynamic states. In this work, we develop deep generative machine learning models Boltzmann Generator approach entire diagrams, employing normalizing flows conditioned states, e.g. temperature and pressure, that they map to. By training single flow transform distribution sampled at only one reference state wide range target temperatures pressures, can efficiently generate across diagram. Using permutation-equivariant architecture allows us, thereby, treat solid liquid same footing. We demonstrate our by predicting solid–liquid coexistence line Lennard-Jones system excellent agreement with state-of-the-art methods while significantly reducing number evaluations needed.

Language: Английский

---

Machine learning in molecular biophysics: Protein allostery, multi-level free energy simulations, and lipid phase transitions

Biophysics Reviews, Journal Year: 2025, Volume and Issue: 6(1)

Published: Feb. 12, 2025

Machine learning (ML) techniques have been making major impacts on all areas of science and engineering, including biophysics. In this review, we discuss several applications ML to biophysical problems based our recent research. The topics include the use identify hotspot residues in allosteric proteins using deep mutational scanning data analyze how mutations these hotspots perturb co-operativity framework a statistical thermodynamic model, improve accuracy free energy simulations by integrating from different levels potential functions, determine phase transition temperature lipid membranes. Through examples, illustrate unique value extracting patterns or parameters complex sets, as well remaining limitations. By implementing approaches context physically motivated models computational frameworks, are able gain deeper mechanistic understanding better convergence numerical simulations. We conclude briefly discussing introduced can be further expanded tackle more problems.

Language: Английский

---

Statistical phase evaluation approach for defect phase diagrams

Physical review. B./Physical review. B, Journal Year: 2025, Volume and Issue: 111(5)

Published: Feb. 27, 2025

We propose an analytical thermodynamic model for describing defect phase transformations, which we term the statistical evaluation approach (SPEA). The SPEA assumes a Boltzmann distribution of finite-size fractions and calculates their average. To benchmark performance model, apply it to construct binary surface diagrams metal alloys. Two alloy systems are considered: Mg with Ca substitutions Ni Nb substitutions. firm basis against can be leveled, first perform Monte Carlo (MC) simulations coupled cluster expansion density functional theory dataset. then demonstrate that reproduces MC results accurately. Specifically, correctly predicts order-disorder transitions as well coexistence 1/3 ordered disordered phase. Finally, compare method sublattice commonly used in CALPHAD describe random solution phases transitions. proposed provides highly efficient modeling transformations. Published by American Physical Society 2025

Language: Английский

---

Accurate prediction of thermoresponsive phase behavior of disordered proteins

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: March 6, 2025

Protein responses to environmental stress, particularly temperature fluctuations, have long been a subject of investigation, with focus on how proteins maintain homeostasis and exhibit thermoresponsive properties. While UCST-type (upper critical solution temperature) phase behavior has studied extensively can now be predicted reliably using computational models, LCST-type (lower transitions remain less explored, lack models capable accurate prediction. This gap limits our ability probe fully undergo in response changes. Here, we introduce Mpipi-T, residue-level coarse-grained model designed predict proteins. Parametrized both atomistic simulations experimental data, Mpipi-T accounts for entropically driven protein separation that occurs upon heating. Accordingly, predicts temperature-driven quantitatively single- multi-chain systems. Beyond its predictive capabilities, demonstrate provides framework uncovering the molecular mechanisms underlying heat stress responses, offering new insights into sense adapt thermal changes biological

Language: Английский

---

A Triple Point on the Phase Diagram of a One-component System in the Van der Waals Approximation

Vestnik Moskovskogo Universiteta Seriya 3 Fizika Astronomiya, Journal Year: 2025, Volume and Issue: 80(№1, 2025)

Published: Jan. 1, 2025

In this work, a phase diagram of the neighborhood triple point one-component system in van der Waals approximation is constructed. It shown that makes it possible to describe corresponding coexistence three aggregate states matter – solid, liquid and gaseous. The possibility using for points other types discussed.

Language: Английский

---

Quantum Support Vector Classifier for Phase Diagram Prediction in Quinary Systems

Materials Horizons, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

This study represents a novel methodology utilizing quantum support vector classifier to predict phase diagrams in quinary systems which enhances predictive accuracy beyond classical methods.

Language: Английский

---

Ab initio phase diagrams of binary alloys in the low solute concentration limit

The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(18)

Published: May 8, 2025

Phase diagrams are crucial to the design of new materials, understand their phase stability and metastability under different thermodynamic conditions, such as composition, temperature, pressure. Here, we use an ab initio approach study diagram a binary alloy within low concentration limit solute. Using molecular dynamics calculations based on density functional theory, estimate solute partitioning ratios in solid–liquid equilibria. The chemical potential difference between solvent atoms both solid liquid phases is calculated using integration. As illustration techniques, have applied this method reproduce Al–Mg at zero We also compute coexistence curve pure Al by applying phase-coexistence with free energy correction technique. results close agreement experiment, demonstrating reliability models.

Language: Английский

---

Interpolation and differentiation of alchemical degrees of freedom in machine learning interatomic potentials

Nature Communications, Journal Year: 2025, Volume and Issue: 16(1)

Published: May 10, 2025

Language: Английский

---

A Triple Point on the Phase Diagram of a One-Component System in the van der Waals Approximation

Moscow University Physics Bulletin, Journal Year: 2025, Volume and Issue: 80(1), P. 60 - 65

Published: Feb. 1, 2025

Language: Английский

---

Monitoring Aqueous Sucrose Solutions Using Droplet Microfluidics: Ice Nucleation, Growth, Glass Transition, and Melting

Langmuir, Journal Year: 2024, Volume and Issue: 40(12), P. 6304 - 6316

Published: March 18, 2024

Freezing and freeze-drying processes are commonly used to extend the shelf life of drug products ensure their safety efficacy upon use. When designing a freezing process, it is beneficial characterize multiple physicochemical properties formulation, such as nucleation rate, crystal growth temperature concentration maximally freeze-concentrated solution, melting point. Differential scanning calorimetry has predominantly been in this context but does have practical limitations unable quantify kinetics nucleation. In work, we introduce microfluidic technique capable quantifying interest use investigate aqueous sucrose solutions varying concentration. Three freeze–thaw cycles were performed on droplets with 75-μm diameters at cooling warming rates 1 °C/min. During each cycle, visual appearance was optically monitored they experienced nucleation, growth, formation melting. Nucleation manifested increases droplet brightness during phase. Heating associated further increase solution approached. beyond point corresponded decrease brightness. Comparison literature confirmed accuracy new while offering data solution. Thus, presented here may serve complement differential freeze-drying. future, could be applied plethora mixtures that undergo processing, whether pharmaceutics, food production, or beyond.

Language: Английский

---