Bulletin of the American Physical Society
APS March Meeting 2020
Volume 65, Number 1
Monday–Friday, March 2–6, 2020; Denver, Colorado
Session L16: Quantum Annealing and Optimization IIFocus
|
Hide Abstracts |
Sponsoring Units: DQI Chair: Yudong Cao, Zapata Room: 201 |
Wednesday, March 4, 2020 8:00AM - 8:12AM |
L16.00001: Adaptive measurement approach towards controlling non-adiabatic transitions in quantum annealing Salil Bedkihal, Mehmet Canturk, Yongchao Tang, Antonio Martinez, Adrian Lupascu, Song Zhang, Juan Atalaya, Birgitta K Whaley Non-adiabatic transitions arising from extremely small energy gaps present a challenge to quantum annealing and adiabatic quantum computing. In realistic quantum annealing scenarios, one does not have a priori knowledge of the energy-spectrum and hence the location of minimum energy gaps; thereby making it difficult to design strategies that will slow down close to the points of enhanced transition probability out of the ground state. In this work, we present an adaptive annealing protocol based on measurement of the energy-level curvature. Numerical results are presented for a random transverse field Ising model. We also discuss the relationship between the measurement operator and the fidelity-susceptibility, a measure allowing analysis of quantum phase transitions in many-body systems. |
Wednesday, March 4, 2020 8:12AM - 8:24AM |
L16.00002: Elucidating the interplay between non-stoquasticity and the sign problem Lalit Gupta, Itay Hen The sign problem is a key challenge in computational physics, encapsulating our inability to properly understand many important quantum many-body phenomena in physics, chemistry and the material sciences. Despite its centrality, the circumstances under which the problem arises or can be resolved as well as its interplay with the related notion of ‘non-stoquasticity’ are often not very well understood. In this study, we make an attempt to elucidate the circumstances under which the sign problem emerges and to clear up some of the confusion surrounding this intricate computational phenomenon. To that aim, we make use of the recently introduced off-diagonal series expansion quantum Monte Carlo scheme with which we analyze in detail a number of examples that capture the essence of our results. |
Wednesday, March 4, 2020 8:24AM - 8:36AM |
L16.00003: All-optical Ising machine by spatial light modulation Davide Pierangeli, Giulia Marcucci, Claudio Conti A broad class of computationally intractable problems maps to the search of the ground state of a spin system. Quantum and classical setups that evolve according to an Ising Hamiltonian are thus emerging as novel computing architectures for solving combinatorial optimizations that cannot be tackled on large scales by conventional hardware. Among these, photonic platforms can process data at light speed and in parallel, through multiple spatial or frequency channels. However, the realized photonic Ising machines either involve a limited number of spins or electronic spin couplings or lack of scalability. |
Wednesday, March 4, 2020 8:36AM - 8:48AM |
L16.00004: Is Fault Welcoming Quantum Computing Realistic? Eliot Kapit, Vadim Oganesyan An error-corrected, fault tolerant quantum computer is one of the most important long term goals of quantum computing research. In these systems random noise is an obstacle that must be overcome through error correction. Here we explore a new possibility, fault-welcoming quantum computing, where a system can not only maintain a quantum speedup for a given (likely non-universal) class of quantum algorithms against realistic noise, but actually performs better than an idealized copy with no noise. We modify flux qubit quantum annealing by including random, coherent low-frequency oscillations in the directions of the transverse field terms during evolution. Through analytical and numerical calculations, we show that this produces a quantum speedup for finding ground states in the Grover problem and quantum random energy model, and thus should be widely applicable to other hard spin glass problems. Further, we show that this speedup should be resilient to two realistic noise channels, and that another channel, bath-assisted phase transitions, accelerates optimization and may outweigh the others, thus potentially making the system fault welcoming. The modifications we consider could be explored with current technology. |
Wednesday, March 4, 2020 8:48AM - 9:00AM |
L16.00005: Schrieffer-Wolff Methods for Annealing Qubits Rudolph Magyar, David george Ferguson A robust method for characterizing the low-energy Pauli decomposition of coupled superconducting qubits is based on a generalization of Bravyi et al’s Schrieffer-Wolff transformation but applied in infinitesimal steps [1]. This method has several desirable properties. The effective Hamiltonian remains block diagonal so that the computational subspace is completely decoupled from the non-computational one. The adiabatic connection terms generated by the time-dependence of the computational subspace are block off-diagonal and thus do not generate any terms within the computational subspace. This work is a continuation of work described last year focused on the readout of single flux qubits. Subsequent numerical advances now permit the extension of the method to interacting tunable flux qubits including novel flux qubits capable of implementing strong non-stoquastic interactions. |
Wednesday, March 4, 2020 9:00AM - 9:12AM |
L16.00006: Beyond Standard Quantum Annealing Invited Speaker: Tameem Albash Quantum annealing is typically studied (and experimentally realized) in terms of an interpolation between a driver and a problem Hamiltonian, often taken to be a uniform transverse field and an Ising Hamiltonian respectively. Nothing restricts the time-dependent Hamiltonian to take such a simple form, and greater control of experimental systems has revived the study of more exotic interpolations. We review recent work exploring new quantum annealing protocols, including adiabatic reverse annealing and the introduction of catalyst Hamiltonians. While providing an exponential improvement in performance for solving certain highly symmetric toy models, no recipe is known for how to use these new proposals to give performance enhancements more generally. This highlights the need for dramatically new insights and methods but also more experimental capabilities to further explore the performance of non-standard quantum annealing. |
Wednesday, March 4, 2020 9:12AM - 9:24AM |
L16.00007: A real-time path integral representation of driven quantum algorithms Frank Wilhelm, David K Headley, Peter Schuhmacher Both adiabatic quantum computing / quantum annealing and the quantum approximate optimization algorithm combine a problem Hamiltonian with a non-commuting driver Hamiltonian in order to efficiently explore the complete state space of an optimization problem. We develop a representation of such algorithms as a real-time path integral that directly and rigorously implements the otherwise colloquial idea that quantum algorithms follow all possible computations at the same time. We apply path integral techniques such as eikonals and semiclassics in order to provide a way to better understanding under which conditions we can expect these algorithms to reliably converge. |
Wednesday, March 4, 2020 9:24AM - 9:36AM |
L16.00008: Updates to Hybrid Quantum-Classical Annealing Peter Schuhmacher, Aditi Misra, Salil Bedkihal, Xi Dai, Adrian Lupascu, Frank Wilhelm Last year. we proposed an efficient gap-independent cooling scheme for a quantum annealer that benefits from finite temperatures. We chose a system based on superconducting flux qubits as a prominent example of current quantum annealing platforms and proposed coupling the qubit systemtransversely to a coplanar waveguide to counter noise and heating that arise from always-present longitudinal thermal noise. We provide a schematic circuit layout for the system and showed we achieve global performance enhancements. However, the work covered only single-qubit annealing. In this work, we discuss different strategies to generalize HQCA to larger qubit numbers. |
Wednesday, March 4, 2020 9:36AM - 9:48AM |
L16.00009: Non-stoquastic interactions of superconducting circuits in the low frequency regime Marius Schöndorf, Frank Wilhelm Non-stoquastic interactions are hard to realize in experimental setups using superconducting qubits. On the other hand they are important or even necessary for the construction of adiabatic quantum computers wich show a real quantum speedup. In ArXiv:1903.06139, Ozfidan et al. show that they can realize non-stoquastic qubit-qubit interactions in a superconducting circuit architecture. The non-stoquastic nature only appears when the system is restricted to the low energy qubit subspace, since the full circuit Hamiltonian itself is stoquastic. Here we study the origin of these non-stoquastic interactions arising when projecting stoquastic Hamiltonians to the low energy spectrum. For this we use different theoretical tools, e.g. renormalization group techniques. |
Wednesday, March 4, 2020 9:48AM - 10:00AM |
L16.00010: Oscillatory quantum optimization methods applied to problems with large ground state bands Zhijie Tang, Eliot Kapit RFQA is a promising new quantum method for solving optimization problems, where by adding local oscillations to transverse fields, it can provide a polynomial quantum speedup over traditional quantum annealing methods. Inspired both by the performance of RFQA in trial problems with few ground states, and by the phenomenology of NP-complete problems, we consider RFQA applied to problems with exponentially many ground states, but where these states are an exponentially small fraction of the total configuration space. We explore how accelerated thermalization in low energy bands can provide a potentially noise tolerant quantum speedup for optimization and machine learning. |
Wednesday, March 4, 2020 10:00AM - 10:12AM |
L16.00011: How Quantum is the Speedup in Adiabatic Unstructured Search? Itay Hen In classical computing, analog approaches have sometimes appeared to be more powerful than they really are. This occurs when resources, particularly precision, are not appropriately taken into account. While the same should also hold for analog quantum computing, precision issues are often neglected from the analysis. I will discuss in the above context the sensitivity of the quantum adiabatic unstructured search algorithm [Roland and Cerf, Phys. Rev. A 65, 042308 (2002)] against various types of imperfections and show that the speedup associated with the algorithm is generally not robust against the presence of finite precision. In addition, I will present a classical analog algorithm for unstructured search that can be viewed as analogous to the quantum adiabatic unstructured search algorithm and which provides a quadratic speedup over standard digital unstructured search. |
Wednesday, March 4, 2020 10:12AM - 10:24AM |
L16.00012: On Constructing Driver Hamiltonians for Several Linear Constraints Hannes Leipold, Federico Maximiliano Spedalieri Recent advances in adiabatic quantum computing and quantum annealers has centered around using more advance and novel Hamiltonians to solve optimization problems. One of these advances has centered around the development of driver Hamiltonians that commute with the constraints of an optimization problem. This approach has been shown to be able to use sparser connectivity to embed several practical problems on quantum devices in comparison to other methods. Designing the driver Hamiltonians that successfully commute with several constraints has largely been based on strong intuition for specific problems and with no general algorithm for arbitrary constraints. In this work, we develop an algebraic framework for reasoning about the commutation of Hamiltonians with linear constraints - one that allows us to classify the complexity of finding a driver Hamiltonian for a set of constraints as NP-Hard through a reduction to the Subset Equal Sums problem as well as design a simple algorithm to solve the problem for Hamiltonians with bounded number of higher body interaction terms. |
Wednesday, March 4, 2020 10:24AM - 10:36AM |
L16.00013: Generating Weighted MAX-2-SAT Instances with Tunable Frustration on an RBM Yan Ru Pei, Haik Manukian, Massimiliano Di Ventra Many optimization problems can be cast into the maximum satisfiability (MAX-SAT) form, and many solvers have been developed for tackling such problems. To evaluate a MAX-SAT solver, it is convenient to generate hard MAX-SAT instances with known solutions. Here, we propose a method of generating weighted MAX-2-SAT instances inspired by the frustrated-loop algorithm used by the quantum annealing community to generate Ising instances on a cubic lattice with nearest-neighbor couplings. We extend the algorithm for instances of general bipartite couplings, with the associated optimization problem being the minimization of the restricted Boltzmann machine (RBM) energy over the nodal values, which is useful for an effective pre-training of the RBM. The difficulty of the generated instances can be tuned through a central parameter known as the frustration index. It is observed through simulation that the frustration index drives a double phase transition in the hardness scaling behavior of the generated instances with respect to the size of the system [1]. Work supported in part by CMRR and DARPA. |
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