Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview

Journal of Chemical Information and Modeling, Journal Year: 2024, Volume and Issue: 64(3), P. 597 - 620

Published: Jan. 29, 2024

Artificial Neural Networks (ANNs) are transforming how we understand chemical mixtures, providing an expressive view of the space and multiscale processes. Their hybridization with physical knowledge can bridge gap between predictivity understanding underlying This overview explores recent progress in ANNs, particularly their potential 'recomposition' mixtures. Graph-based representations reveal patterns among mixture components, deep learning models excel capturing complexity symmetries when compared to traditional Quantitative Structure–Property Relationship models. Key such as Hamiltonian networks convolution operations, play a central role representing The integration ANNs Chemical Reaction Physics-Informed for inverse kinetic problems is also examined. combination sensors shows promise optical biomimetic applications. A common ground identified context statistical physics, where ANN-based methods iteratively adapt by blending initial states training data. concept recomposition unveils reciprocal inspiration reactive highlighting behaviors influenced environment.

Language: Английский

---

k-means clustering for persistent homology

Advances in Data Analysis and Classification, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 31, 2024

Abstract Persistent homology is a methodology central to topological data analysis that extracts and summarizes the features within dataset as persistence diagram. It has recently gained much popularity from its myriad successful applications many domains, however, algebraic construction induces metric space of diagrams with highly complex geometry. In this paper, we prove convergence k -means clustering algorithm on diagram establish theoretical properties solution optimization problem in Karush–Kuhn–Tucker framework. Additionally, perform numerical experiments both simulated real various representations persistent homology, including embeddings well themselves their generalizations measures. We find performance directly measures outperform vectorized representations.

Language: Английский

---

Reaction Networks Resemble Low-Dimensional Regular Lattices

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 5, 2024

The computational exploration, manipulation, and design of complex chemical reactions face fundamental challenges related to the high-dimensional nature potential energy surfaces (PESs) that govern reactivity. Accurately modeling is crucial for understanding processes involved in, example, organocatalysis, autocatalytic cycles, one-pot molecular assembly. Our prior research demonstrated discretizing PESs using heuristics based on bond breaking formation produces a reaction network representation with low-dimensional structure (metric space). We now find these stoichiometry-preserving networks possess additional, though approximate, resemble regular lattices small amount random edge rewiring. heuristics-based discretization thus generates nonlinear dimensionality reduction by factor 10 an

Language: Английский

---

“Quantum-Chemoinformatics” for Design and Discovery of New Molecules and Reactions

Published: March 8, 2024

We give an overview of the role “quantum-chemoinformatics” in drug development. Quantum- chemoinformatics is a data-driven chemistry using descriptors on basis theoretical chemistry, especially quantum (QC) and ab initio molecular dynamics (MD) simulations. focus quantum-chemoinformatics for chemical reaction design prediction, which one important processes basic research start with brief historical then introduces two projects quantum-cheminformatics. The RMap project uses QC-based route networks discovery new molecules reactions. other related to environmental pollution by molecules, property should be taken into account evaluation. last section describes our recent attempt accelerate QC-data acquisition utilizing limited amount experimental data machine learning (ML) technology.

Language: Английский

---

Acceleration of Reaction Space Projector Analysis Using Combinatorial Optimization: Application to Organic Chemical Reactions

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 9, 2024

In recent years, automated reaction path search methods have established the concept of a route network. The Reaction Space Projector (ReSPer) visualizes potential energy hypersurface into lower-dimensional subspace using principal coordinates. main time-consuming process in ReSPer is calculating structural distance matrix, making it impractical for complex organic networks. We implemented Alternate Optimization (AO) algorithm, one combinatorial optimizations, to reduce computational costs. Evaluations gold clusters and Au

Language: Английский

---

Reproducing Reaction Route Map on the Shape Space from its Quotient by Complete Nuclear Permutation-Inversion group

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

This study develops an algorithm to reproduce reaction route maps (RRMs) in shape space from the outputs of potential search algorithms. To demonstrate this, GRRM is utilized as a but proposed should work with other algorithms principle. The does not require any encoding molecular configurations and thus applicable complicated realistic molecules for which efficient readily available. We show subgraphs RRM mapped each by action symmetry group are isomorphic also provide compute set feasible transformations sense Longuet--Higgins. toy models more molecules. Finally, we remark on absolute rate theory our perspective.

Language: Английский

---

Reproducing the Reaction Route Map on the Shape Space from Its Quotient by the Complete Nuclear Permutation-Inversion Group

Journal of Chemical Theory and Computation, Journal Year: 2023, Volume and Issue: 19(17), P. 5886 - 5896

Published: Aug. 29, 2023

This study develops an algorithm to reproduce reaction route maps (RRMs) in the shape space from outputs of potential search algorithms. To demonstrate algorithm, global mapping is utilized as a but proposed should work with other algorithms principle. The does not require any encoding molecular configurations and thus applicable complicated realistic molecules for which efficient readily available. We show that subgraphs RRM mapped each by action symmetry group are isomorphic also provide compute set feasible transformations sense Longuet-Higgins. toy models more molecules. Finally, we remark on absolute rate theory our perspective.

Language: Английский

---

Novel Descriptor of Potential Energy Surface: Persistent Homology of Reaction Route Map

Journal of Computer Chemistry Japan, Journal Year: 2024, Volume and Issue: 23(1), P. 33 - 36

Published: Jan. 1, 2024

A reaction route map (RRM), which is a collection of elementary pathways, contracts the potential energy surface (PES) with 3N − 6 variables (N: number atoms) into weighted graph representation. Although automated construction RRMs has greatly contributed to accurate understanding chemical mechanisms, only small fraction networks low activation energies are relevant actual reactions, and thus studies focusing on entire RRM have not been conducted. In this letter, we summarize our recent approach applying persistent homology (PH) analysis structure an RRM.

Language: Английский

---

“Quantum-Chemoinformatics” for Design and Discovery of New Molecules and Reactions

Published: Jan. 1, 2024

Language: Английский

---