Abstract:In this work we apply deep neural networks to find the non-equilibrium steady state solution to correlated open quantum many-body systems. Motivated by the ongoing search to find more powerful representations of (mixed) quantum states, we design a simple prototypical convolutional neural network and show that parametrizing the density matrix directly with more powerful models can yield better variational ansatz functions and improve upon results reached by neural density operator based on the restricted Boltzmann machine. Hereby we give up the explicit restriction to positive semi-definite density matrices. However, this is fulfilled again to good approximation by optimizing the parameters. The great advantage of this approach is that it opens up the possibility of exploring more complex network architectures that can be tailored to specific physical properties. We show how translation invariance can be enforced effortlessly and reach better results with fewer parameters. We present results for the dissipative one-dimensional transverse-field Ising model and a two-dimensional dissipative Heisenberg model compared to exact values.

What problem does this paper attempt to address?

This paper attempts to address the problem of finding non-equilibrium steady state (NESS) solutions in open quantum many-body systems. Specifically, the authors apply deep neural networks (particularly convolutional neural networks, CNNs) to directly parameterize the mixed density matrices describing these systems, overcoming limitations in traditional methods such as positive definiteness and Hermiticity constraints, and enhancing the ability to express complex physical properties. ### Main Problem Background: 1. **Open Quantum Systems**: These systems interact with the environment, leading to dissipative processes, typically described by quantum master equations such as the Lindblad equation. 2. **Limitations of Traditional Methods**: Traditional neural network quantum state (NQS) methods, such as those based on restricted Boltzmann machines (RBMs), can handle some problems but have limitations like a large number of parameters and difficulty scaling to higher-dimensional systems. 3. **Need for New Methods**: To better represent complex quantum states, more powerful models and architectures are needed, especially for mixed density matrices in open quantum systems. ### Main Contributions of the Paper: 1. **Direct Parameterization of Density Matrices**: By directly parameterizing the density matrices instead of relying on purification methods (like RBMs), the authors can use more powerful deep neural network models. 2. **Application of Convolutional Neural Networks**: The introduction of convolutional neural networks (CNNs) leverages their advantages in handling local and long-range correlations, particularly in systems with translational symmetry. 3. **Reduction in Number of Parameters**: The design of convolutional layers and pooling layers allows for a significant reduction in the number of parameters while maintaining high accuracy. 4. **Optimization Method**: The use of the backtracking Nesterov accelerated gradient descent (NAGD+) optimization method improves convergence speed and result quality. ### Experimental Validation: - **1D Dissipative Transverse Field Ising Model**: Demonstrated that the CNN model can accurately predict the expectation values of spins under different magnetic field strengths, especially in parameter ranges where RBM methods struggle. - **2D Dissipative Heisenberg Model**: Further validated the effectiveness of the CNN model in higher-dimensional systems, achieving results comparable to exact values in the calculation of σz expectation values on a 3×3 lattice. ### Conclusion: By directly parameterizing mixed density matrices and using convolutional neural networks, the proposed method in this paper shows significant advantages in handling open quantum many-body systems, particularly in reducing the number of parameters and improving computational efficiency. This opens new directions for exploring more complex neural network architectures and applications in the future.

Deep Neural Networks as Variational Solutions for Correlated Open Quantum Systems

Neural-network Variational Quantum Algorithm for Simulating Many-Body Dynamics

Neural-network quantum states for many-body physics

Tensor-network-based variational Monte Carlo approach to the non-equilibrium steady state of open quantum systems

Variational Neural-Network Ansatz for Continuum Quantum Field Theory

Solving the Quantum Many-Body Problem with Artificial Neural Networks

Symmetries and many-body excited states with neural-network quantum states

Deep learning of spatial densities in inhomogeneous correlated quantum systems

Quantum neural network approach to Markovian dissipative dynamics of many-body open quantum systems

Solving Quantum Master Equations with Deep Quantum Neural Networks

Variational Monte Carlo with Neural Network Quantum States for Yang-Mills Matrix Model

Neural Quantum States in Variational Monte Carlo Method: A Brief Summary

Variational Neural and Tensor Network Approximations of Thermal States

Entanglement transition in deep neural quantum states

Autoregressive neural quantum states of Fermi Hubbard models

Liouville Space Neural Network Representation of Density Matrices

High-accuracy variational Monte Carlo for frustrated magnets with deep neural networks

Efficient solutions of fermionic systems using artificial neural networks

Constructing exact representations of quantum many-body systems with deep neural networks