-
Towards Improved Quantum Machine Learning for Molecular Force Fields
Authors:
Yannick Couzinié,
Shunsuke Daimon,
Hirofumi Nishi,
Natsuki Ito,
Yusuke Harazono,
Yu-ichiro Matsushita
Abstract:
This study explores the use of equivariant quantum neural networks (QNN) for generating molecular force fields, focusing on the rMD17 dataset. We consider a QNN architecture based on previous research and point out shortcomings in the parametrization of the atomic environments, that limits its expressivity as an interatomic potential and precludes transferability between molecules. We propose a re…
▽ More
This study explores the use of equivariant quantum neural networks (QNN) for generating molecular force fields, focusing on the rMD17 dataset. We consider a QNN architecture based on previous research and point out shortcomings in the parametrization of the atomic environments, that limits its expressivity as an interatomic potential and precludes transferability between molecules. We propose a revised QNN architecture that addresses these shortcomings. While both QNNs show promise in force prediction, with the revised architecture showing improved accuracy, they struggle with energy prediction. Further, both QNNs architectures fail to demonstrate a meaningful scaling law of decreasing errors with increasing training data. These findings highlight the challenges of scaling QNNs for complex molecular systems and emphasize the need for improved encoding strategies, regularization techniques, and hybrid quantum-classical approaches.
△ Less
Submitted 6 May, 2025;
originally announced May 2025.
-
Machine learning supported annealing for prediction of grand canonical crystal structures
Authors:
Yannick Couzinie,
Yuya Seki,
Yusuke Nishiya,
Hirofumi Nishi,
Taichi Kosugi,
Shu Tanaka,
Yu-ichiro Matsushita
Abstract:
This study investigates the application of Factorization Machines with Quantum Annealing (FMQA) to address the crystal structure problem (CSP) in materials science. FMQA is a black-box optimization algorithm that combines machine learning with annealing machines to find samples to a black-box function that minimize a given loss. The CSP involves determining the optimal arrangement of atoms in a ma…
▽ More
This study investigates the application of Factorization Machines with Quantum Annealing (FMQA) to address the crystal structure problem (CSP) in materials science. FMQA is a black-box optimization algorithm that combines machine learning with annealing machines to find samples to a black-box function that minimize a given loss. The CSP involves determining the optimal arrangement of atoms in a material based on its chemical composition, a critical challenge in materials science. We explore FMQA's ability to efficiently sample optimal crystal configurations by setting the loss function to the energy of the crystal configuration as given by a predefined interatomic potential. Further we investigate how well the energies of the various metastable configurations, or local minima of the potential, are learned by the algorithm. Our investigation reveals FMQA's potential in quick ground state sampling and in recovering relational order between local minima.
△ Less
Submitted 7 August, 2024;
originally announced August 2024.
-
First-quantized adiabatic time evolution for the ground state of a many-electron system and the optimal nuclear configuration
Authors:
Yusuke Nishiya,
Hirofumi Nishi,
Yannick Couzinié,
Taichi Kosugi,
Yu-ichiro Matsushita
Abstract:
We propose a novel adiabatic time evolution (ATE) method for obtaining the ground state of a quantum many-electron system on a quantum circuit based on first quantization. As a striking feature of the ATE method, it consists of only unitary operations representing real-time evolution, which means that it does not require any ancillary qubits, nor controlled real-time evolution operators. Especiall…
▽ More
We propose a novel adiabatic time evolution (ATE) method for obtaining the ground state of a quantum many-electron system on a quantum circuit based on first quantization. As a striking feature of the ATE method, it consists of only unitary operations representing real-time evolution, which means that it does not require any ancillary qubits, nor controlled real-time evolution operators. Especially, we explored the first-quantized formalism of ATE method in this study, since the implementation of first-quantized real-time evolution on quantum circuits is known to be efficient. However, when realizing the ATE quantum circuit in first-quantization formalism, obstacles are how to set the adiabatic Hamiltonian and how to prepare the corresponding initial ground state. We provide a way to prepare an antisymmetrized and non-degenerate initial ground state that is suitable as an input to an ATE circuit, which allows our ATE method to be applied to systems with any number of electrons. In addition, by considering a first-quantized Hamiltonian for quantum-mechanical electron system and classical nuclear system, we design a quantum circuit for optimal structure search based on ATE. Numerical simulations are demonstrated for simple systems, and it is confirmed that the ground state of the electronic system and optimal structure can be obtained by our method.
△ Less
Submitted 27 December, 2023; v1 submitted 7 September, 2023;
originally announced September 2023.
-
Annealing for prediction of grand canonical crystal structures: Efficient implementation of n-body atomic interactions
Authors:
Yannick Couzinie,
Yusuke Nishiya,
Hirofumi Nishi,
Taichi Kosugi,
Hidetoshi Nishimori,
Yu-ichiro Matsushita
Abstract:
We propose an annealing scheme usable on modern Ising machines for crystal structures prediction (CSP) by taking into account the general n-body atomic interactions, and in particular three-body interactions which are necessary to simulate covalent bonds. The crystal structure is represented by discretizing a unit cell and placing binary variables which express the existence or non-existence of an…
▽ More
We propose an annealing scheme usable on modern Ising machines for crystal structures prediction (CSP) by taking into account the general n-body atomic interactions, and in particular three-body interactions which are necessary to simulate covalent bonds. The crystal structure is represented by discretizing a unit cell and placing binary variables which express the existence or non-existence of an atom on every grid point. The resulting quadratic unconstrained binary optimization (QUBO) or higher-order unconstrained binary optimization (HUBO) problems implement the CSP problem and is solved using simulated and quantum annealing. Using the example of Lennard-Jones clusters we show that it is not necessary to include the target atom number in the formulation allowing for simultaneous optimization of both the particle density and the configuration and argue that this is advantageous for use on annealing machines as it reduces the total amount of interactions. We further provide a scheme that allows for reduction of higher-order interaction terms that is inspired by the underlying physics. We show for a covalently bonded monolayer MoS2 crystal that we can simultaneously optimize for the particle density as well as the crystal structure using simulated annealing. We also show that we reproduce ground states of the interatomic potential with high probability that are not represented on the initial discretization of the unit cell.
△ Less
Submitted 11 September, 2023; v1 submitted 6 July, 2023;
originally announced July 2023.
-
The multicolour East model
Authors:
Yannick Couzinié
Abstract:
We consider the multicolour East model, a model of glass forming liquids closely related to the East model on $\mathbb{Z}^d$. The state space ${(G\cup \{\star\})}^{\mathbb{Z}^d}$ consists of $|G|\le 2^d$ different vacancy types and the neutral state $\star$. To each $h\in G$ we associate unique facilitation mechanisms ${\{c_x^{h}\}}_{x\in \mathbb{Z}^d}$ that correspond to rotated versions of the E…
▽ More
We consider the multicolour East model, a model of glass forming liquids closely related to the East model on $\mathbb{Z}^d$. The state space ${(G\cup \{\star\})}^{\mathbb{Z}^d}$ consists of $|G|\le 2^d$ different vacancy types and the neutral state $\star$. To each $h\in G$ we associate unique facilitation mechanisms ${\{c_x^{h}\}}_{x\in \mathbb{Z}^d}$ that correspond to rotated versions of the East model constraints. If $c_x^{h}$ is satisfied, the state on $x$ can transition from $h$ to $\star$ with rate $p\in (0,1)$ or vice versa with rate $q_h\in (0,1)$, where generally $q_h\neq q_{h'}$ if $h'\neq h$. Notably, vertices in the state $h$ cannot transition directly to $h'\neq h$ and neighbouring $h'$-vacancies do not contribute in satisfying $c_x^{h}$. Thus, there is a novel blocking mechanism between vacancies of differing type. We find sufficient conditions on the model geometry to have a positive spectral gap and prove that with $|G|=2^d$ the model is not ergodic. For $d=2$ we prove that the model with $|G|\le 3$ has positive spectral gap and we find sufficient conditions on the transition rates for the spectral gap to be given in the leading order by the spectral gap of the East model on $\mathbb{Z}^2$ with parameter $q_{\min}=\min_{h\in G}q_h$ in the limit $q_{\min}\rightarrow 0$. In particular, we prove this when there are $h\in G$ with $q_h\gg q_{\min}$ by explicitly constructing mechanisms on which the frequent vacancy types cooperate to facilitate the East movement of the least frequent vacancies.
△ Less
Submitted 20 July, 2022;
originally announced July 2022.
-
On a front evolution problem for the multidimensional East model
Authors:
Yannick Couzinié,
Fabio Martinelli
Abstract:
We consider a natural front evolution problem the East process on $\mathbb{Z}^d, d\ge 2,$ a well studied kinetically constrained model for which the facilitation mechanism is oriented along the coordinate directions, as the equilibrium density $q$ of the facilitating vertices vanishes. Starting with a unique unconstrained vertex at the origin, let $S(t)$ consist of those vertices which became unco…
▽ More
We consider a natural front evolution problem the East process on $\mathbb{Z}^d, d\ge 2,$ a well studied kinetically constrained model for which the facilitation mechanism is oriented along the coordinate directions, as the equilibrium density $q$ of the facilitating vertices vanishes. Starting with a unique unconstrained vertex at the origin, let $S(t)$ consist of those vertices which became unconstrained within time $t$ and, for an arbitrary positive direction $\mathbf x,$ let $v_{\max}(\mathbf x),v_{\min}(\mathbf x )$ be the maximal/minimal velocities at which $S(t)$ grows in that direction. If $\mathbf x$ is independent of $q$, we prove that $v_{\max}(\mathbf x)= v_{\min}(\mathbf x)^{(1+o(1))}=γ(d) ^{(1+o(1))}$ as $q\to 0$, where $γ(d)$ is the spectral gap of the process on $\mathbb{Z}^d$. We also analyse the case in which some of the coordinates of $\mathbf x$ vanish as $q\to 0$. In particular, for $d=2$ we prove that if $\mathbf x$ approaches one of the two coordinate directions fast enough, then $v_{\max}(\mathbf x)= v_{\min}(\mathbf x)^{(1+o(1))}=γ(1) ^{(1+o(1))}=γ(d)^{d(1+o(1))},$ i.e. the growth of $S(t)$ close to the coordinate directions is dictated by the one dimensional process. As a result the region $S(t)$ becomes extremely elongated inside $\mathbb{Z}^d_+$. We also establish mixing time cutoff for the chain in finite boxes with minimal boundary conditions. A key ingredient of our analysis is the renormalisation technique of arXiv:1404.7257 to estimate the spectral gap of the East process. Here we extend this technique to get the main asymptotics of a suitable principal Dirichlet eigenvalue of the process.
△ Less
Submitted 9 November, 2022; v1 submitted 29 December, 2021;
originally announced December 2021.
-
Weakly reinforced Pólya urns on countable networks
Authors:
Yannick Couzinié,
Christian Hirsch
Abstract:
We study the long-time asymptotics of a network of weakly reinforced Pólya urns. In this system, which extends the WARM introduced by R. van der Hofstad et. al. (2016) to countable networks, the nodes fire at times given by a Poisson point process. When a node fires, one of the incident edges is selected with a probability proportional to its weight raised to a power $α< 1$, and then this weight i…
▽ More
We study the long-time asymptotics of a network of weakly reinforced Pólya urns. In this system, which extends the WARM introduced by R. van der Hofstad et. al. (2016) to countable networks, the nodes fire at times given by a Poisson point process. When a node fires, one of the incident edges is selected with a probability proportional to its weight raised to a power $α< 1$, and then this weight is increased by $1$.
We show that for $α< 1/2$ on a network of bounded degrees, every edge is reinforced a positive proportion of time, and that the limiting proportion can be interpreted as an equilibrium in a countable network. Moreover, in the special case of regular graphs, this homogenization remains valid beyond the threshold $α= 1/2$.
△ Less
Submitted 25 June, 2021; v1 submitted 7 October, 2020;
originally announced October 2020.
