Seminar: Programming a Quantum Annealer

Scott Pakin

Scientist, Los Alamos National Labratory

Tuesday, May 7
4:00pm - 5:00pm
310 Kelly Hall

Scott Pakin

Abstract:

A quantum annealer is a special-purpose quantum computer that exploits quantum effects to approximate the solution to a particular NP-hard problem: minimizing a 2-local Ising-model Hamiltonian function. Essentially, one can think of this as simulated annealing performed in hardware. Quantum annealers containing thousands of qubits (quantum bits) are already commercially available, representing orders of magnitude more qubits than what is provided by any current general-purpose quantum computer.

A timely research challenge -- and the subject of this talk is how to effectively program a quantum annealer. At the lowest level, a quantum-annealing program is merely a list of coefficients of a quadratic pseudo-Boolean function. Virtually all quantum-annealing programs written to date required painstaking manual effort to map problems to Hamiltonian coefficients. In this talk we demonstrate how to map arbitrary programs written in a classical programming language into a minimization problem. We argue that with this approach, efficient approximate solutions to difficult (i.e., NP) problems can in fact be *easier* to express than with conventional, classical techniques.  

Biography:

Scott Pakin has worked since 2002 as a scientist at Los Alamos National Laboratory. He has researched over time a variety of Computer Science topics related to high-performance computing, including programming models, application performance analysis, energy efficiency, and high-speed communication. Scott has recently begun investigating quantum computing and currently serves as the technical/scientific point of contact for LANL's D-Wave quantum annealer. Scott holds an M.S. and a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign and a B.S. in Mathematics/Computer Science from Carnegie Mellon University.

*
Pakin Seminar Slides.pdf