Autoregressive models are a class of generative model that probabilistically predict the next output of a sequence based on previous inputs. The autoregressive sequence is by definition one-dimensional (1D), which is natural for language tasks and hence an important component of modern architectures like recurrent neural networks (RNNs) and transformers. However, when language models are used to predict outputs on physical systems that are not intrinsically 1D, the question arises of which choice of autoregressive sequence -- if any -- is optimal. In this paper, we study the reconstruction of critical correlations in the two-dimensional (2D) Ising model, using RNNs and transformers trained on binary spin data obtained near the thermal phase transition. We compare the training performance for a number of different 1D autoregressive sequences imposed on finite-size 2D lattices. We find that paths with long 1D segments are more efficient at training the autoregressive models compared to space-filling curves that better preserve the 2D locality. Our results illustrate the potential importance in choosing the optimal autoregressive sequence ordering when training modern language models for tasks in physics.

Rydberg atom array experiments have demonstrated the ability to act as powerful quantum simulators, preparing strongly-correlated phases of matter which are challenging to study for conventional computer simulations. A key direction has been the implementation of interactions on frustrated geometries, in an effort to prepare exotic many-body states such as spin liquids and glasses. In this paper, we apply two-dimensional recurrent neural network (RNN) wave functions to study the ground states of Rydberg atom arrays on the kagome lattice. We implement an annealing scheme to find the RNN variational parameters in regions of the phase diagram where exotic phases may occur, corresponding to rough optimization landscapes. For Rydberg atom array Hamiltonians studied previously on the kagome lattice, our RNN ground states show no evidence of exotic spin liquid or emergent glassy behavior. In the latter case, we argue that the presence of a non-zero Edwards-Anderson order parameter is an artifact of the long autocorrelations times experienced with quantum Monte Carlo simulations. This result emphasizes the utility of autoregressive models, such as RNNs, to explore Rydberg atom array physics on frustrated lattices and beyond.

The last few years have seen an explosion of interest in quantum machine learning to accelerate scientific discovery in a range of fields, from quantum computing to the development of new materials and medicines. That effort deepened in July as researchers from industry and academia gathered for the week-long workshop "Machine Learning for Quantum Design" at Perimeter Institute. Conference co-organizer Roger Melko said the conference demonstrated the remarkable progress researchers have made in just a few years since the previous gathering of its kind at Perimeter. "We first had this conference on quantum machine learning three years ago, and it was largely blue-sky proposals and ideas back then," he said. "Now, the scientists here are actually implementing those ideas. The field is changing fast and the pace of that change is accelerating."

Last year, Google's DeepMind AI beat Lee Sedol at Go, a strategy game like chess, but orders of magnitude more complicated. The win was a remarkable step forward for the field of artificial intelligence, but it got Roger Melko, a physicist at the Perimeter Institute for Theoretical Physics, thinking about how neural networks--a type of AI modeled after the human brain--might be used to solve some of the toughest problems in quantum physics. Indeed, intelligent machines may be necessary to solve these problems. "The thing about quantum physics is it's highly complex in a very precise mathematical sense. A big problem we face when we study these quantum systems [without machine learning] is how to deal with this complexity," Melko told me.

A machine learning algorithm designed to teach computers how to recognize photos, speech patterns, and hand-written digits has now been applied to a vastly different set a data: identifying different phases of condensed matter. In a project half-jokingly called "Phasebook," two Perimeter researchers showed that a neural network system – a standard part of today's powerful artificial intelligence (AI) algorithms – can also identify phase transitions between states of matter. The research, published today in the journal Nature Physics, validates the idea that the relationship between theoretical physics and AI can be a fruitful, two-way exchange. The fields have long been linked. AI research has often tapped physicists to help develop machine learning for industry.

A machine learning algorithm designed to teach computers how to recognize photos, speech patterns, and hand-written digits has now been applied to a vastly different set of data: identifying phase transitions between states of matter. This new research, published today in Nature Physics by two Perimeter Institute researchers, was built on a simple question: could industry-standard machine learning algorithms help fuel physics research? To find out, former Perimeter Institute postdoctoral fellow Juan Cassasquilla and Roger Melko, an Associate Faculty member at Perimeter and Associate Professor at the University of Waterloo, repurposed Google's TensorFlow, an open-source software library for machine learning, and applied it to a physical system. Melko says they didn't know what to expect. "I thought it was a long shot," he admits. Using gigabytes of data representing different state configurations created using simulation software on supercomputers, Carrasquilla and Melko created a large collection of "images" to introduce into the machine learning algorithm (also known as a neural network).