Showing 1–9 of 9 results for author: Schutski, R

BindGPT: A Scalable Framework for 3D Molecular Design via Language Modeling and Reinforcement Learning

Authors: Artem Zholus, Maksim Kuznetsov, Roman Schutski, Rim Shayakhmetov, Daniil Polykovskiy, Sarath Chandar, Alex Zhavoronkov

Abstract: Generating novel active molecules for a given protein is an extremely challenging task for generative models that requires an understanding of the complex physical interactions between the molecule and its environment. In this paper, we present a novel generative model, BindGPT which uses a conceptually simple but powerful approach to create 3D molecules within the protein's binding site. Our mode… ▽ More Generating novel active molecules for a given protein is an extremely challenging task for generative models that requires an understanding of the complex physical interactions between the molecule and its environment. In this paper, we present a novel generative model, BindGPT which uses a conceptually simple but powerful approach to create 3D molecules within the protein's binding site. Our model produces molecular graphs and conformations jointly, eliminating the need for an extra graph reconstruction step. We pretrain BindGPT on a large-scale dataset and fine-tune it with reinforcement learning using scores from external simulation software. We demonstrate how a single pretrained language model can serve at the same time as a 3D molecular generative model, conformer generator conditioned on the molecular graph, and a pocket-conditioned 3D molecule generator. Notably, the model does not make any representational equivariance assumptions about the domain of generation. We show how such simple conceptual approach combined with pretraining and scaling can perform on par or better than the current best specialized diffusion models, language models, and graph neural networks while being two orders of magnitude cheaper to sample. △ Less

Submitted 5 June, 2024; originally announced June 2024.

TTOpt: A Maximum Volume Quantized Tensor Train-based Optimization and its Application to Reinforcement Learning

Authors: Konstantin Sozykin, Andrei Chertkov, Roman Schutski, Anh-Huy Phan, Andrzej Cichocki, Ivan Oseledets

Abstract: We present a novel procedure for optimization based on the combination of efficient quantized tensor train representation and a generalized maximum matrix volume principle. We demonstrate the applicability of the new Tensor Train Optimizer (TTOpt) method for various tasks, ranging from minimization of multidimensional functions to reinforcement learning. Our algorithm compares favorably to popular… ▽ More We present a novel procedure for optimization based on the combination of efficient quantized tensor train representation and a generalized maximum matrix volume principle. We demonstrate the applicability of the new Tensor Train Optimizer (TTOpt) method for various tasks, ranging from minimization of multidimensional functions to reinforcement learning. Our algorithm compares favorably to popular evolutionary-based methods and outperforms them by the number of function evaluations or execution time, often by a significant margin. △ Less

Submitted 28 September, 2022; v1 submitted 30 April, 2022; originally announced May 2022.

Comments: 26 pages, 8 figures, accepted to Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS 2022). Pre camera-ready version

Tensor Network Quantum Simulator With Step-Dependent Parallelization

Authors: Danylo Lykov, Roman Schutski, Alexey Galda, Valerii Vinokur, Yuri Alexeev

Abstract: In this work, we present a new large-scale quantum circuit simulator. It is based on the tensor network contraction technique to represent quantum circuits. We propose a novel parallelization algorithm based on \stepslice . In this paper, we push the requirement on the size of a quantum computer that will be needed to demonstrate the advantage of quantum computation with Quantum Approximate Optimi… ▽ More In this work, we present a new large-scale quantum circuit simulator. It is based on the tensor network contraction technique to represent quantum circuits. We propose a novel parallelization algorithm based on \stepslice . In this paper, we push the requirement on the size of a quantum computer that will be needed to demonstrate the advantage of quantum computation with Quantum Approximate Optimization Algorithm (QAOA). We computed 210 qubit QAOA circuits with 1,785 gates on 1,024 nodes of the the Cray XC 40 supercomputer Theta. To the best of our knowledge, this constitutes the largest QAOA quantum circuit simulations reported to this date. △ Less

Submitted 20 April, 2022; v1 submitted 4 December, 2020; originally announced December 2020.

Simple heuristics for efficient parallel tensor contraction and quantum circuit simulation

Authors: Roman Schutski, Dmitry Kolmakov, Taras Khakhulin, Ivan Oseledets

Abstract: Tensor networks are the main building blocks in a wide variety of computational sciences, ranging from many-body theory and quantum computing to probability and machine learning. Here we propose a parallel algorithm for the contraction of tensor networks using probabilistic graphical models. Our approach is based on the heuristic solution of the $μ$-treewidth deletion problem in graph theory. We a… ▽ More Tensor networks are the main building blocks in a wide variety of computational sciences, ranging from many-body theory and quantum computing to probability and machine learning. Here we propose a parallel algorithm for the contraction of tensor networks using probabilistic graphical models. Our approach is based on the heuristic solution of the $μ$-treewidth deletion problem in graph theory. We apply the resulting algorithm to the simulation of random quantum circuits and discuss the extensions for general tensor network contractions. △ Less

Submitted 1 May, 2020; v1 submitted 22 April, 2020; originally announced April 2020.

Comments: 11 pages, 12 figures

Journal ref: Phys. Rev. A 102, 062614 (2020)

An adaptive algorithm for quantum circuit simulation

Authors: Roman Schutski, Danil Lykov, Ivan Oseledets

Abstract: Efficient simulation of quantum computers is essential for the development and validation of near-term quantum devices and the research on quantum algorithms. Up to date, two main approaches to simulation were in use, based on either full state or single amplitude evaluation. We propose an algorithm that efficiently interpolates between these two possibilities. Our approach elucidates the connecti… ▽ More Efficient simulation of quantum computers is essential for the development and validation of near-term quantum devices and the research on quantum algorithms. Up to date, two main approaches to simulation were in use, based on either full state or single amplitude evaluation. We propose an algorithm that efficiently interpolates between these two possibilities. Our approach elucidates the connection between quantum circuit simulation and partial evaluation of expressions in tensor algebra. △ Less

Submitted 10 December, 2019; v1 submitted 27 November, 2019; originally announced November 2019.

Comments: 10 pages, 11 figures

Journal ref: Phys. Rev. A 101, 042335 (2020)

Graph Convolutional Policy for Solving Tree Decomposition via Reinforcement Learning Heuristics

Authors: Taras Khakhulin, Roman Schutski, Ivan Oseledets

Abstract: We propose a Reinforcement Learning based approach to approximately solve the Tree Decomposition (TD) problem. TD is a combinatorial problem, which is central to the analysis of graph minor structure and computational complexity, as well as in the algorithms of probabilistic inference, register allocation, and other practical tasks. Recently, it has been shown that combinatorial problems can be su… ▽ More We propose a Reinforcement Learning based approach to approximately solve the Tree Decomposition (TD) problem. TD is a combinatorial problem, which is central to the analysis of graph minor structure and computational complexity, as well as in the algorithms of probabilistic inference, register allocation, and other practical tasks. Recently, it has been shown that combinatorial problems can be successively solved by learned heuristics. However, the majority of existing works do not address the question of the generalization of learning-based solutions. Our model is based on the graph convolution neural network (GCN) for learning graph representations. We show that the agent builton GCN and trained on a single graph using an Actor-Critic method can efficiently generalize to real-world TD problem instances. We establish that our method successfully generalizes from small graphs, where TD can be found by exact algorithms, to large instances of practical interest, while still having very low time-to-solution. On the other hand, the agent-based approach surpasses all greedy heuristics by the quality of the solution. △ Less

Submitted 20 February, 2020; v1 submitted 18 October, 2019; originally announced October 2019.

Comments: 8 pages, 7 figures

Journal ref: NeurIPS 2020 Learning Meets Combinatorial Algorithms Workshop

Tensor-decomposition techniques for ab initio nuclear structure calculations. From chiral nuclear potentials to ground-state energies

Authors: Alexander Tichai, Roman Schutski, Gustavo E. Scuseria, Thomas Duguet

Abstract: The impact of applying state-of-the-art tensor factorization techniques to modern nuclear Hamiltonians derived from chiral effective field theory is investigated. Subsequently, the error induced by the tensor decomposition of the input Hamiltonian on ground-state energies of closed-shell nuclei calculated via second-order many-body perturbation theory is benchmarked. With the aid of the factorized… ▽ More The impact of applying state-of-the-art tensor factorization techniques to modern nuclear Hamiltonians derived from chiral effective field theory is investigated. Subsequently, the error induced by the tensor decomposition of the input Hamiltonian on ground-state energies of closed-shell nuclei calculated via second-order many-body perturbation theory is benchmarked. With the aid of the factorized Hamiltonian, the second-order perturbative correction to ground-state energies is decomposed and the scaling properties of the underlying tensor network are discussed. The employed tensor formats are found to lead to an efficient data compression of two-body matrix elements of the nuclear Hamiltonian. In particular, the sophisticated \emph{tensor hypercontraction} (THC) scheme yields low tensor ranks with respect to both harmonic-oscillator and Hartree-Fock single-particle bases. It is found that the tensor rank depends on the two-body total angular momentum $J$ for which one performs the decomposition, which is itself directly related to the sparsity the corresponding tensor. Furthermore, including normal-ordered two-body contributions originating from three-body interactions does not compromise the efficient data compression. Ultimately, the use of factorized matrix elements authorizes controlled approximations of the exact second-order ground-state energy corrections. In particular, a small enough error is obtained from low-rank factorizations in $^{4}$He, $^{16}$O and $^{40}$Ca. △ Less

Submitted 19 October, 2018; originally announced October 2018.

Comments: 16 pages, 13 figures, 1 table

Journal ref: Phys. Rev. C 99, 034320 (2019)

Tensor-Structured Coupled Cluster Theory

Authors: Roman Schutski, **mo Zhao, Thomas M. Henderson, Gustavo E. Scuseria

Abstract: We derive and implement a new way of solving coupled cluster equations with lower computational scaling. Our method is based on decomposition of both amplitudes and two electron integrals, using a combination of tensor hypercontraction and canonical polyadic decomposition. While the original theory scales as $O(N^6)$ with respect to the number of basis functions, we demonstrate numerically that we… ▽ More We derive and implement a new way of solving coupled cluster equations with lower computational scaling. Our method is based on decomposition of both amplitudes and two electron integrals, using a combination of tensor hypercontraction and canonical polyadic decomposition. While the original theory scales as $O(N^6)$ with respect to the number of basis functions, we demonstrate numerically that we achieve sub-millihartree difference from the original theory with $O(N^4)$ scaling. This is accomplished by solving directly for the factors that decompose the cluster operator. The proposed scheme is quite general and can be easily extended to other many-body methods. △ Less

Submitted 8 August, 2017; originally announced August 2017.

Journal ref: J. Chem. Phys. 147, 184113 (2017)

Multi-reference symmetry-projected variational approaches for ground and excited states of the one-dimensional Hubbard model

Authors: R. Rodríguez-Guzmán, Carlos A. Jiménez-Hoyos, R. Schutski, Gustavo E. Scuseria

Abstract: We present a multi-reference configuration mixing scheme for describing ground and excited states, with well defined spin and space group symmetry quantum numbers, of the one-dimensional Hubbard model with nearest-neighbor hop** and periodic boundary conditions. Within this scheme, each state is expanded in terms of non-orthogonal and variationally determined symmetry-projected configurations. T… ▽ More We present a multi-reference configuration mixing scheme for describing ground and excited states, with well defined spin and space group symmetry quantum numbers, of the one-dimensional Hubbard model with nearest-neighbor hop** and periodic boundary conditions. Within this scheme, each state is expanded in terms of non-orthogonal and variationally determined symmetry-projected configurations. The results for lattices up to 30 and 50 sites compare well with the exact Lieb-Wu solutions as well as with results from other state-of-the-art approximations. In addition to spin-spin correlation functions in real space and magnetic structure factors, we present results for spectral functions and density of states computed with an ansatz whose quality can be well-controlled by the number of symmetry-projected configurations used to approximate the systems with $N_{e}$ and $N_{e} \pm 1$ electrons. The intrinsic symmetry-broken determinants resulting from the variational calculations have rich structures in terms of defects that can be regarded as basic units of quantum fluctuations. Given the quality of the results here reported, as well as the parallelization properties of the considered scheme, we believe that symmetry-projection techniques, which have found ample applications in nuclear structure physics, deserve further attention in the study of low-dimensional correlated many-electron systems. △ Less

Submitted 12 June, 2013; v1 submitted 15 April, 2013; originally announced April 2013.

Comments: 17 pages, 10 figures

Journal ref: Phys. Rev. B 87, 235129 (2013)