Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2205.12250 · Journal of Machine Learning

Efficient Anti-Symmetrization of a Neural Network Layer by Taming the Sign Problem

Nilin Abrahamsen, Lin Lin

Published Sep 23, 2023• 4 citations
arXivHTMLPDF

Abstract

Explicit antisymmetrization of a neural network is a potential candidate for a universal function approximator for generic antisymmetric functions, which are ubiquitous in quantum physics. However, this procedure is a priori factorially costly to implement, making it impractical for large numbers of particles. The strategy also suffers from a sign problem. Namely, due to near-exact cancellation of positive and negative contributions, the magnitude of the antisymmetrized function may be significantly smaller than before antisymmetrization. We show that the anti-symmetric projection of a two-layer neural network can be evaluated efficiently, opening the door to using a generic anti-symmetric layer as a building block in anti-symmetric neural network Ansatzes. This approximation is effective when the sign problem is controlled, and we show that this property depends crucially the choice of activation function under standard Xavier/He initialization methods. As a consequence, using a smooth activation function requires rescaling of the neural network weights compared to standard initializations.

DOI
10.4208/jml.230703

Related papers

Bi-Lipschitz Ansatz for Anti-Symmetric Functions

Nadav Dym, Jianfeng Lu, Matan Mizrachi

Motivated by applications to the simulation of quantum many-body systems by neural networks, researchers have suggested several models which are antisymmetric by construction, and can approximate all antisymmetric functions. However, these works either require very high computational complexity to attain universal approximation, or suffer from discontinuities. In this paper, we introduce two antisymmetric ansatzes which do not suffer from these disadvantages. The first is based on a bi-Lipschitz embedding with respect to a naturally defined metric. The second is a modular anti-symmetrizing projection framework based on the frame-averaging methodology. Both approaches yield continuous antisymmetric models which attain universal approximation guarantees with a polynomial complexity in problem size. Moreover, for both approaches, we obtain quantitative approximation results that bound the number of parameters required to approximate Lipschitz antisymmetric functions to a given accuracy $ε$. We also provide preliminary experimental evidence suggesting improved performance in learning antisymmetric functions.

arXiv:2503.04263•Mar 6, 2025•4 citations

On Representing (Anti)Symmetric Functions

Marcus Hütter

Permutation-invariant, -equivariant, and -covariant functions and anti-symmetric functions are important in quantum physics, computer vision, and other disciplines. Applications often require most or all of the following properties: (a) a large class of such functions can be approximated, e.g. all continuous function, (b) only the (anti)symmetric functions can be represented, (c) a fast algorithm for computing the approximation, (d) the representation itself is continuous or differentiable, (e) the architecture is suitable for learning the function from data. (Anti)symmetric neural networks have recently been developed and applied with great success. A few theoretical approximation results have been proven, but many questions are still open, especially for particles in more than one dimension and the anti-symmetric case, which this work focusses on. More concretely, we derive natural polynomial approximations in the symmetric case, and approximations based on a single generalized Slater determinant in the anti-symmetric case. Unlike some previous super-exponential and discontinuous approximations, these seem a more promising basis for future tighter bounds. We provide a complete and explicit universality proof of the Equivariant MultiLayer Perceptron, which implies universality of symmetric MLPs and the FermiNet.

arXiv:2007.15298•Jul 30, 2020•45 citations

$\widetilde{O}(N^2)$ Representation of General Continuous Anti-symmetric Function

Haotian Ye, Ruichen Li, Yuntian Gu +3 more

In quantum mechanics, the wave function of fermion systems such as many-body electron systems are anti-symmetric (AS) and continuous, and it is crucial yet challenging to find an ansatz to represent them. This paper addresses this challenge by presenting an ${\widetilde O}(N^2)$ ansatz based on permutation-equivariant functions. We prove that our ansatz can represent any AS continuous functions, and can accommodate the determinant-based structure proposed by Hutter [14], solving the proposed open problems that ${O}(N)$ Slater determinants are sufficient to provide universal representation of AS continuous functions. Together, we offer a generalizable and efficient approach to representing AS continuous functions, shedding light on designing neural networks to learn wave functions.

arXiv:2402.15167•Feb 23, 2024•6 citations

Towards Antisymmetric Neural Ansatz Separation

Aaron Zweig, Joan Bruna

We study separations between two fundamental models (or \emph{Ansätze}) of antisymmetric functions, that is, functions $f$ of the form $f(x_{σ(1)}, \ldots, x_{σ(N)}) = \text{sign}(σ)f(x_1, \ldots, x_N)$, where $σ$ is any permutation. These arise in the context of quantum chemistry, and are the basic modeling tool for wavefunctions of Fermionic systems. Specifically, we consider two popular antisymmetric Ansätze: the Slater representation, which leverages the alternating structure of determinants, and the Jastrow ansatz, which augments Slater determinants with a product by an arbitrary symmetric function. We construct an antisymmetric function in $N$ dimensions that can be efficiently expressed in Jastrow form, yet provably cannot be approximated by Slater determinants unless there are exponentially (in $N^2$) many terms. This represents the first explicit quantitative separation between these two Ansätze.

arXiv:2208.03264•Aug 5, 2022•6 citations

Fermionic Neural Networks through the lens of Group Theory

J. Rozalén Sarmiento, Arnau Rios

We present an overview of the method of Neural Quantum States applied to the many-body problem of atomic nuclei. Through the lens of group representation theory, we focus on the problem of constructing neural-network ansätze that respect physical symmetries. We explicitly prove that determinants, which are among the most common methods to build antisymmetric neural-network wave functions, can be understood as the result of a group convolution. We also identify the reason why this construction is so efficient in practice compared to other group convolutional operations. We conclude that group representation theory is a promising avenue to incorporate explicitly symmetries in Neural Quantum States.

arXiv:2411.11605•International Conference on Quarks and Nuclear Physics (QNP)•Feb 7, 2025•0 citations

Combining neural networks and signed particles to simulate quantum systems more efficiently, Part III

Jean Michel Sellier, Gaétan Marceau, Jacob Leygonie

This work belongs to a series of articles which have been dedicated to the combination of signed particles and neural networks to speed up the time-dependent simulation of quantum systems. More specifically, the suggested networks are utilized to compute the function known as the Wigner kernel. In the first paper, we suggested a network which is completely defined analytically and which does not necessitate any training process. Although very useful, this approach keeps the same complexity as the more standard finite difference methods. In the second work, we presented a different architecture which has generalization capabilities. Although more convenient in terms of computational time, it still uses a similar structure compared to the previous approach, with less neurons in the hidden layer but with the same expensive activation functions (sine functions). In this work, we focus on neural networks without any prior physics based knowledge. In more details, the network now consists of different hidden layers which are based on more common, and less computationally expensive, activation functions such as rectified linear units, and which focus on predicting only one column of the discretized Wigner kernel at a time. This new suggested architecture proves to accurately learn the transform from the space of potential functions to the space of Wigner kernels and performs very well during the simulation of quantum systems (20 times faster). Finally, we simulate a one-dimensional quantum system consisting of a wave packet impinging on a potential barrier for validation purposes. This represents one further step ahead to achieve fast and reliable simulations of time-dependent quantum systems.

arXiv:1807.06939•Jun 28, 2018•0 citations
Explicit antisymmetrization of a neural network is a potential candidate for a universal function approximator for generic antisymmetric functions, which are ubiquitous in quantum physics. However, this procedure is a priori factorially costly to implement, making it impractical for large numbers of particles. The strategy also suffers from a sign problem. Namely, due to near-exact cancellation of positive and negative contributions, the magnitude of the antisymmetrized function may be significantly smaller than before antisymmetrization. We show that the anti-symmetric projection of a two-layer neural network can be evaluated efficiently, opening the door to using a generic anti-symmetric layer as a building block in anti-symmetric neural network Ansatzes. This approximation is effective when the sign problem is controlled, and we show that this property depends crucially the choice of activation function under standard Xavier/He initialization methods. As a consequence, using a smooth activation function requires rescaling of the neural network weights compared to standard initializations.
Motivated by applications to the simulation of quantum many-body systems by neural networks, researchers have suggested several models which are antisymmetric by construction, and can approximate all antisymmetric functions. However, these works either require very high computational complexity to attain universal approximation, or suffer from discontinuities. In this paper, we introduce two antisymmetric ansatzes which do not suffer from these disadvantages. The first is based on a bi-Lipschitz embedding with respect to a naturally defined metric. The second is a modular anti-symmetrizing projection framework based on the frame-averaging methodology. Both approaches yield continuous antisymmetric models which attain universal approximation guarantees with a polynomial complexity in problem size. Moreover, for both approaches, we obtain quantitative approximation results that bound the number of parameters required to approximate Lipschitz antisymmetric functions to a given accuracy εεε. We also provide preliminary experimental evidence suggesting improved performance in learning antisymmetric functions.
Permutation-invariant, -equivariant, and -covariant functions and anti-symmetric functions are important in quantum physics, computer vision, and other disciplines. Applications often require most or all of the following properties: (a) a large class of such functions can be approximated, e.g. all continuous function, (b) only the (anti)symmetric functions can be represented, (c) a fast algorithm for computing the approximation, (d) the representation itself is continuous or differentiable, (e) the architecture is suitable for learning the function from data. (Anti)symmetric neural networks have recently been developed and applied with great success. A few theoretical approximation results have been proven, but many questions are still open, especially for particles in more than one dimension and the anti-symmetric case, which this work focusses on. More concretely, we derive natural polynomial approximations in the symmetric case, and approximations based on a single generalized Slater determinant in the anti-symmetric case. Unlike some previous super-exponential and discontinuous approximations, these seem a more promising basis for future tighter bounds. We provide a complete and explicit universality proof of the Equivariant MultiLayer Perceptron, which implies universality of symmetric MLPs and the FermiNet.
O~(N2)\widetilde{O}(N^2)O(N2) Representation of General Continuous Anti-symmetric Function
In quantum mechanics, the wave function of fermion systems such as many-body electron systems are anti-symmetric (AS) and continuous, and it is crucial yet challenging to find an ansatz to represent them. This paper addresses this challenge by presenting an O~(N2){\widetilde O}(N^2)O(N2) ansatz based on permutation-equivariant functions. We prove that our ansatz can represent any AS continuous functions, and can accommodate the determinant-based structure proposed by Hutter [14], solving the proposed open problems that O(N){O}(N)O(N) Slater determinants are sufficient to provide universal representation of AS continuous functions. Together, we offer a generalizable and efficient approach to representing AS continuous functions, shedding light on designing neural networks to learn wave functions.
We study separations between two fundamental models (or \emph{Ansätze}) of antisymmetric functions, that is, functions fff of the form f(xσ(1),…,xσ(N))=sign(σ)f(x1,…,xN)f(x_{σ(1)}, \ldots, x_{σ(N)}) = \text{sign}(σ)f(x_1, \ldots, x_N)f(xσ(1)​,…,xσ(N)​)=sign(σ)f(x1​,…,xN​), where σσσ is any permutation. These arise in the context of quantum chemistry, and are the basic modeling tool for wavefunctions of Fermionic systems. Specifically, we consider two popular antisymmetric Ansätze: the Slater representation, which leverages the alternating structure of determinants, and the Jastrow ansatz, which augments Slater determinants with a product by an arbitrary symmetric function. We construct an antisymmetric function in NNN dimensions that can be efficiently expressed in Jastrow form, yet provably cannot be approximated by Slater determinants unless there are exponentially (in N2N^2N2) many terms. This represents the first explicit quantitative separation between these two Ansätze.
We present an overview of the method of Neural Quantum States applied to the many-body problem of atomic nuclei. Through the lens of group representation theory, we focus on the problem of constructing neural-network ansätze that respect physical symmetries. We explicitly prove that determinants, which are among the most common methods to build antisymmetric neural-network wave functions, can be understood as the result of a group convolution. We also identify the reason why this construction is so efficient in practice compared to other group convolutional operations. We conclude that group representation theory is a promising avenue to incorporate explicitly symmetries in Neural Quantum States.
This work belongs to a series of articles which have been dedicated to the combination of signed particles and neural networks to speed up the time-dependent simulation of quantum systems. More specifically, the suggested networks are utilized to compute the function known as the Wigner kernel. In the first paper, we suggested a network which is completely defined analytically and which does not necessitate any training process. Although very useful, this approach keeps the same complexity as the more standard finite difference methods. In the second work, we presented a different architecture which has generalization capabilities. Although more convenient in terms of computational time, it still uses a similar structure compared to the previous approach, with less neurons in the hidden layer but with the same expensive activation functions (sine functions). In this work, we focus on neural networks without any prior physics based knowledge. In more details, the network now consists of different hidden layers which are based on more common, and less computationally expensive, activation functions such as rectified linear units, and which focus on predicting only one column of the discretized Wigner kernel at a time. This new suggested architecture proves to accurately learn the transform from the space of potential functions to the space of Wigner kernels and performs very well during the simulation of quantum systems (20 times faster). Finally, we simulate a one-dimensional quantum system consisting of a wave packet impinging on a potential barrier for validation purposes. This represents one further step ahead to achieve fast and reliable simulations of time-dependent quantum systems.