29 March 2017 in Physical Review Letters
The defining problem in frustrated quantum magnetism, the ground state of the nearest-neighbor antiferromagnetic Heisenberg model on the kagome lattice, has defied all theoretical and numerical methods employed to date. We apply the formalism of tensor-network states, specifically the method of projected entangled simplex states, which combines infinite system size with a correct accounting for multipartite entanglement. By studying the ground-state energy, the finite magnetic order appearing at finite tensor bond dimensions, and the effects of a next-nearest-neighbor coupling, we demonstrate that the ground state is a gapless spin liquid. We discuss the comparison with other numerical studies and the physical interpretation of this result.
5 September 2019 in Physical Review X
Differentiable programming is a fresh programming paradigm which composes parameterized algorithmic components and optimizes them using gradient search. The concept emerges from deep learning but is not limited to training neural networks. We present the theory and practice of programming tensor network algorithms in a fully differentiable way. By formulating the tensor network algorithm as a computation graph, one can compute higher-order derivatives of the program accurately and efficiently using automatic differentiation. We present essential techniques to differentiate through the tensor networks contraction algorithms, including numerical stable differentiation for tensor decompositions and efficient backpropagation through fixed-point iterations. As a demonstration, we compute the specific heat of the Ising model directly by taking the second-order derivative of the free energy obtained in the tensor renormalization group calculation. Next, we perform gradient-based variational optimization of infinite projected entangled pair states for the quantum antiferromagnetic Heisenberg model and obtain state-of-the-art variational energy and magnetization with moderate efforts. Differentiable programming removes laborious human efforts in deriving and implementing analytical gradients for tensor network programs, which opens the door to more innovations in tensor network algorithms and applications.
2 February 2018 in Physical Review B
The restricted Boltzmann machine (RBM) is one of the fundamental building blocks of deep learning. RBM finds wide applications in dimensional reduction, feature extraction, and recommender systems via modeling the probability distributions of a variety of input data including natural images, speech signals, and customer ratings, etc. We build a bridge between RBM and tensor network states (TNS) widely used in quantum many-body physics research. We devise efficient algorithms to translate an RBM into the commonly used TNS. Conversely, we give sufficient and necessary conditions to determine whether a TNS can be transformed into an RBM of given architectures. Revealing these general and constructive connections can cross fertilize both deep learning and quantum many-body physics. Notably, by exploiting the entanglement entropy bound of TNS, we can rigorously quantify the expressive power of RBM on complex data sets. Insights into TNS and its entanglement capacity can guide the design of more powerful deep learning architectures. On the other hand, RBM can represent quantum many-body states with fewer parameters compared to TNS, which may allow more efficient classical simulations.
18 April 2019 in Physical Review B
Matrix product states (MPSs), a tensor network designed for one-dimensional quantum systems, were recently proposed for generative modeling of natural data (such as images) in terms of the “Born machine.” However, the exponential decay of correlation in MPSs restricts its representation power heavily for modeling complex data such as natural images. In this work, we push forward the effort of applying tensor networks to machine learning by employing the tree tensor network (TTN), which exhibits balanced performance in expressibility and efficient training and sampling. We design the tree tensor network to utilize the two-dimensional prior of the natural images and develop sweeping learning and sampling algorithms which can be efficiently implemented utilizing graphical processing units. We apply our model to random binary patterns and the binary MNIST data sets of handwritten digits. We show that the TTN is superior to MPSs for generative modeling in keeping the correlation of pixels in natural images, as well as giving better log-likelihood scores in standard data sets of handwritten digits. We also compare its performance with state-of-the-art generative models such as variational autoencoders, restricted Boltzmann machines, and PixelCNN. Finally, we discuss the future development of tensor network states in machine learning problems.
23 June 2016 in Physical Review Letters
The sign problem is a major obstacle in quantum Monte Carlo simulations for many-body fermion systems. We examine this problem with a new perspective based on the Majorana reflection positivity and Majorana Kramers positivity. Two sufficient conditions are proven for the absence of the fermion sign problem. Our proof provides a unified description for all the interacting lattice fermion models previously known to be free of the sign problem based on the auxiliary field quantum Monte Carlo method. It also allows us to identify a number of new sign-problem-free interacting fermion models including, but not limited to, lattice fermion models with repulsive interactions but without particle-hole symmetry, and interacting topological insulators with spin-flip terms.
24 July 2024 in Physical Review B
The discovery of superconductivity at 80 K under high pressure in presents the groundbreaking confirmation that high- superconductivity is a property of strongly correlated materials beyond cuprates. We use density functional theory calculations of the band structure of under pressure to verify that the low-energy bands are composed almost exclusively of Ni and O orbitals. We deduce that the Ni orbitals are essentially decoupled by the geometry of the high-pressure structure and by the effect of the Ni Hund coupling being strongly suppressed, which results from the enhanced interlayer antiferromagnetic interaction between orbitals and the strong intralayer hybridization of the orbitals with O . By introducing a tight-binding model for the Fermi surfaces and low-energy dispersions, we arrive at a bilayer model with strong interlayer hopping, which we show is a framework unifying with cuprate materials possessing similar band structures, particularly the compounds , , and . We use a renormalized mean-field theory to show that these systems should have -wave superconductivity, with a dominant -wave component and the high driven by the near-optimally doped band, while the band adds an -wave component that should lead to clear experimental signatures.