Bulletin of the American Physical Society
APS March Meeting 2020
Volume 65, Number 1
Monday–Friday, March 2–6, 2020; Denver, Colorado
Session L39: Machine Learning for Quantum Matter IFocus
|
Hide Abstracts |
Sponsoring Units: DCOMP GDS DMP Chair: Miles Stoudenmire, Simons Foundation Room: 703 |
Wednesday, March 4, 2020 8:00AM - 8:36AM |
L39.00001: Classifying Snapshots of the Doped Hubbard Model with Machine Learning Invited Speaker: Annabelle Bohrdt Quantum gas microscopes for ultracold atoms provide real-space snapshots of complex many-body systems with single site resolution. We use machine learning techniques to analyse and classify such snapshots of ultracold atoms. Specifically, we study the two-dimensional Fermi-Hubbard model, which is believed to capture the rich physics of high-temperature superconductivity and other phases such as the strange metal, stripe, antiferromagnet, or pseudo-gap phase. While a large number of theories exist to describe this system, each with its own merits, a unifying analytic understanding is still lacking. We compare the data from an experimental realization of the two-dimensional Fermi–Hubbard model to two theoretical approaches: a doped quantum spin liquid state of resonating valence bond type, and the geometric string theory, describing a state with hidden spin order. This method considers all available information without a potential bias towards one particular theory by the choice of an observable and can therefore select the theory that is more predictive in general. Up to intermediate doping values, our algorithm tends to classify experimental snapshots as geometric-string-like, as compared to the doped spin liquid. Our results demonstrate the potential for machine learning in processing the wealth of data obtained through quantum gas microscopy for new physical insights. |
Wednesday, March 4, 2020 8:36AM - 8:48AM |
L39.00002: AI Assisted Discovery in Quantum Gas Microscope Images Elmer Guardado-Sanchez, Benjamin M Spar, Juan Carrasquilla, Richard Theodore Scalettar, Waseem S Bakr, Ehsan Khatami Quantum gas microscopes for ultracold atoms in optical lattices have transformed quantum simulations of many-body Hamiltonians. Statistical analysis of atomic snapshots can produce expectation values for various charge and spin correlation functions and has led to new discoveries for the Fermi-Hubbard model in two dimensions. Here, we enlist the help of artificial intelligence to look for possible patterns in the snapshots not captured by conventional indicators. We try this unbiased approach on images taken in the non-Fermi liquid phase of the Hubbard model around optimal doping. |
Wednesday, March 4, 2020 8:48AM - 9:00AM |
L39.00003: Unsupervised machine learning of topological phase transitions Joaquin Rodriguez Nieva, Mathias Scheurer In the traditional theory of phase transitions, pioneered by Landau, different phases are characterized by the symmetries they break. However, physical systems can also exhibit phase transitions between two states that share the same symmetries, but can be sharply distinguished by their “topological” properties. While symmetry-breaking phase transitions are readily captured with machine learning, topological phase transitions are significantly more difficult, which is related to their non-local nature. In this talk, I will discuss an unsupervised machine-learning approach that we propose [see Nature Physics 15, 790-795 (2019)], which is capable of “learning” topological invariants from raw, unlabeled data. The success of the approach is demonstrated on several different models and we also discuss a mapping of the output of the machine learning to the eigenvalues and wavefunctions of an auxiliary quantum problem. This will allow us to use physical intuition of quantum mechanics to understand how the machine-learning algorithm performs the topological classification. |
Wednesday, March 4, 2020 9:00AM - 9:12AM |
L39.00004: Classification of optical quantum states using machine learning Shahnawaz Ahmed, Carlos Sánchez Muñoz, Franco Nori, Anton Frisk Kockum Machine-learning techniques for quantum state tomography can give significantly faster results, e.g., by using adaptive measurements to avoid redundant data acquisition, or by finding efficient parameterizations of a quantum state to escape the "curse of dimensionality". Here, we explore a similar idea for optical quantum states by using deep neural networks (DNNs). We train a DNN to classify different optical quantum states, e.g., cat or thermal states, with a high accuracy. Our DNN can also predict interesting physical properties, such as Wigner negativity, directly from measurement data without requiring a full reconstruction. We study the influence of various factors such as noise, Hilbert-space cutoff, and measurement settings on the predictions and show that the DNN approach is robust. We also apply standard methods for analyzing neural network predictions, such as Grad-CAM, to determine the features used by the network to make its predictions. To benchmark our method, we compare with a naive classifier using maximum likelihood estimation. Our results indicate that the DNN can be a fast and efficient real-time classifier to distinguish various optical quantum states in the lab. |
Wednesday, March 4, 2020 9:12AM - 9:24AM |
L39.00005: Unsupervised learning of quantum phase transitions using nonlinear dimension reduction methods Alexander Lidiak, Zhexuan Gong Quantum simulators have reached a complexity that understanding measurement data they generated has become a daunting task for traditional data analysis methods. To make scientific discoveries based on experimental data where theoretical understanding is lacking, unsupervised machine learning can be a powerful tool. However, existing approaches to unsupervised learning of quantum many-body states are largely focused linear dimension reduction methods such as principle component analysis (PCA), or generalizations such as kernel PCA. These methods often fail when order parameters of the states are nonlinear functions, i.e. states with valence-bond order, topological order, many-body localization, etc. This motivates us to investigate nonlinear dimension reduction methods such as diffusion maps and autoencoders. By studying a 1D chiral Z3 clock model (experimentally a chain of Rydberg atoms), we find PCA detects only the Z3 phase while diffusion maps detects the full phase map including a incommensurate phase. In addition, diffusion maps directly detect the Z3 symmetry of model and predict the number of clusters. We find these nonlinear dimension reduction methods also useful in learning valence-bond order and Gaussian-type topological phase transitions in quantum spin systems. |
Wednesday, March 4, 2020 9:24AM - 9:36AM |
L39.00006: Machine learning the Mattis glass transformation Daniel Lozano-Gomez, Darren Pereira, Michel J P Gingras Machine learning techniques are being actively explored to ascertain their usefulness in many areas of physics. Within condensed matter physics, one of the goals these techniques are being investigated for is per their ability to identify the different phases of a system. As such, the identification of a system's symmetries and underlying gauges is taken as a crucial step to accomplish this endeavor. In this context, we consider classical spin models in which we introduce a so-called Mattis gauge transformation. This transformation turns the standard ferromagnetic Ising model into a non-frustrated random bond Ising model, and the ferromagnetic XY model into an XY model with random Dzyaloshinskii-Moriya interactions. We show through a simple unsupervised method, the Principal Component Analysis (PCA), that PCA is able to expose the introduced, albeit hidden, gauge transformation for both models. For the Ising Mattis-model, the original Ising model structure is recovered by PCA, and a rough estimate of the gauge distribution is extracted. For the XY gauge model, we show that while the gauge transformation is ``hidden’’ in the clustering structure for the full data set, it is nevertheless uncovered when studying a specific component of the spins. |
Wednesday, March 4, 2020 9:36AM - 9:48AM |
L39.00007: Augmenting machine learning algorithms with the addition of a physics based intelligence prior Christopher Singh, Matthew Redell, Mohannad Elhamod, Jie Bu, Anuj Karpatne, Wei-Cheng Lee Improving the predictive power of machine learning models is one of the |
Wednesday, March 4, 2020 9:48AM - 10:00AM |
L39.00008: Adversarial machine learning for modeling the distribution of large-scale ultracold atom experiments Corneel Casert, Kyle Mills, Tom Vieijra, Jan Ryckebusch, Isaac Tamblyn Directly generating microstates with desired properties from the configuration space of many-body systems is infeasible due to its high-dimensional nature. Instead, traditional generation methods rely on computationally costly algorithms or carefully controlled experimental setups, which limits the number of particles that can be investigated. |
Wednesday, March 4, 2020 10:00AM - 10:12AM |
L39.00009: Using Convolutional Neural Networks to analyze phase transitions and calculate critical exponents Nishad Maskara, Evert Van Nieuwenburg, Manuel Endres Identifying phase transitions and their corresponding order parameters is a central problem in physics, but estimation of order parameters from experimental measurements is difficult. In this work, we present an alternative framework for analyzing phase transitions by using neural networks to learn order parameters directly from data. By introducing a type of convolutional architecture, we show how these methods can be made more robust by systematically increasing the convolutional window size. We investigate the extraction of correlation length critical exponents by performing finite-size scaling on the network, and use this to analyze the 1D TFIM phase transition from measurements in different bases. This work is a step towards a machine learning toolkit for characterizing phase transitions without prior knowledge. |
Wednesday, March 4, 2020 10:12AM - 10:24AM |
L39.00010: Unsupervised learning of topological indices Oleksandr Balabanov, Mats Granath I will present an unsupervised protocol for learning topological indices of quantum systems [1]. The idea is to produce ensembles of topologically equivalent data and then train a specially designed neural-network-based regressor for classifying them. The datasets of topologically equivalent samples are derived by continuously deforming some selected parent systems and this procedure does not require any knowledge of the topological numbers or how they are constructed. I will explicitly illustrate the protocol with two examples: It will be employed for classifying 1d band insulators in symmetry class AIII, characterized by a winding number, and 2d band insulators in symmetry class A, characterized by a Chern number. |
Wednesday, March 4, 2020 10:24AM - 10:36AM |
L39.00011: Machine Learning based BCS superconductivity Predictor from Normal State Properties Fei Han, Nina Andrejevic, Thanh Nguyen, Quynh Nguyen, Shreya Parjan, Mingda Li BCS theory is the widely accepted microscopic mechanism for conventional superconductivity. However, despite decades’ research effort, it is still challenging to judge that whether a material is superconducting or not, not to mention a faithful estimation on the superconducting critical temperature Tc. In this study, we employed a few deep learning architectures to correlate the normal state properties to superconductivity. A few normal state properties are found to be closely related to the formation of superconductivity with further link to Tc. Our work might offer an alternative avenue to rapidly identify superconducting materials. |
Wednesday, March 4, 2020 10:36AM - 10:48AM |
L39.00012: Unlocking quantum critical phenomena with physics guided artificial intelligence Christopher Singh, Matthew Redell, Mohannad Elhamod, Jie Bu, Wei-Cheng Lee, Anuj Karpatne Breakthroughs in cold atom experiments, advances in quantum computing, developments in spin liquids, and the proliferating importance of quantum critical phenomena compel the application of machine learning techniques to difficult quantum problems. In an age where data can drive unparalleled discoveries, expensive-to-acquire data such as measurements of quantum computer states or cold atom chains can be used by the community to distill new information. Thus, more effective ways of prediction and distillation are required to efficiently identify the criticality. While many have done this using a classification algorithm, we have pioneered a method to predict quantum critical phenomena using machine learning in the absence of direct exposure to states on either side of the transition by directly predicting the ground state wavefunction. By analyzing the predictions for the total phase space, we can confidently identify the location of criticality from the evolution of the predicted wavefunctions. Through further development, this type of machine could help researchers quickly, and cheaply, identify regions of the phase space that are of the utmost interest. |
Wednesday, March 4, 2020 10:48AM - 11:00AM |
L39.00013: Neural-Network Approach to Dissipative Quantum Many-Body Dynamics Michael Hartmann, Giuseppe Carleo In experimentally realistic situations, quantum systems are never perfectly isolated and the coupling to their environment needs to be taken into account. Often, the effect of the environment can be well approximated by a Markovian master equation. However, solving this master equation for quantum many-body systems becomes exceedingly hard due to the high dimension of the Hilbert space. Here we present an approach to the effective simulation of the dynamics of open quantum many-body systems based on machine-learning techniques. We represent the mixed many-body quantum states with neural networks in the form of restricted Boltzmann machines and derive a variational Monte Carlo algorithm for their time evolution and stationary states. We document the accuracy of the approach with numerical examples for a dissipative spin lattice system. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700