Our panel discussion on Quantum Computing at Supercomputing 2007 gives an update on current activity on quantum systems, as they are being studied for future high performance computing (HPC). The concept of a quantum computer has been with us now for over twenty-five years, so it is reasonable to ask whether its promise can be fulfilled anytime in the foreseeable future.

Probably everyone has heard what you can do with a quantum computer if you have one: factor very large integers efficiently (this is Shor's algorithm). It is also often said that this is the only efficient quantum algorithm, which would mean that this form of HPC would only be of interest to a few spooks. But there are definitely other interesting quantum algorithms. None of them obviously match the biggest current applications of HPC (although there are some interesting results on the speedup of integration and differential-equation solvers), but they are diverse. Here's an example of one that has only been discovered in the past year, describable as the following genetics problem: Suppose you know an ancient gene X, the ancestor of present day genes Y and Z. The possible mutations are all known, in the form of string-rewriting rules (e.g., ACgAT). Call N(XdY) then number of distinct mutation sequences that lead from X to Y. Then there is an efficient quantum algorithm for N(XdY) - N(XdZ), in regimes where computation of this quantity on a digital computer is almost certainly inaccurate and/or inefficient.

I have presented this particular example not to imply that by next year all bio-information departments will urgently need quantum computers, but to illustrate that quantum computers have the potential to be valuable in realms that have (apparently) nothing to do with quantum mechanics, at least no more than prime factorization does.

So, will we have quantum computers someday soon? Let me first say what defines a piece of hardware as a quantum computer. It is sometimes said that quantum computing is what happens to logic gates when they are manufactured at the atomic scale. This isn't necessarily so; "nanoscale" does not equate to "quantum". The essence of quantumness is not in size but in the degree of control. In a quantum switch, the information-carrying degree of freedom (e.g., the charge on a capacitor) must be held and switched with a precision almost reaching the ultimate limit prescribed by the quantum uncertainty principle. It is only in this regime that the switch's state can exhibit the quantum properties of coherent superposition and entanglement that enable a quantum form of information processing. One of our panelists, Geordie Rose, will discuss an intermediate point of view that his work represents. He hopes that his superconducting hardware, possessing some, but not all, attributes of quantum switches, will enable other novel HPC applications.

The quest to build fully functioning quantum hardware is active on many fronts. Atomic physicists have seen for some years that the quantum states of a single atom held at rest in a trap, manipulated by laser pulses, functions as a highly coherent quantum information carrier. The ability to perform elementary logic operations on such a qubit has been well demonstrated.

Unfortunately, atomic physicists are not skilled HPC designers. So, much work also goes on in the area of novel integrated-circuit devices, in which the necessary quantum control is harder to demonstrate, but from which a large-scale device could be more readily created than it could be with trapped-atom technology. Two of these efforts are represented by leading practitioners on our panel: Will Oliver is a specialist in superconducting electronics, in which quantum behavior results not because the circuits are atomic-scale, but because of the special physical properties of the superconducting state. He has interesting results on a potentially scalable Josephson-junction circuit. Another panelist, Eli Yablonovich, is an expert on the creation of qubits using individual atomic impurities in semiconductors (yes, he is also the inventor of the photonic bandgap effect). The control of individual atomic impurities and individual electrons in electronic devices has been a beautiful technological feat of recent years, which has opened up many novel possibilities, quantum and otherwise, for new, ultradense integrated devices.

So, quantum computers are still (mostly) in the basic research lab, and there is no sure answer to the question of when they will become a relevant technology in HPC. But if you are designing encryption for future systems or networks, don't bet against the spooks having a quantum codecracker before too long.

-----

Since 1985, David P. DiVincenzo has been a Research Staff Member in the Physical Sciences Department at the IBM T. J. Watson Research Center in Yorktown Heights, NY. He has worked throughout his career in various problems in condensed matter physics.

Since 1993, he has explored quantum information theory and the physical realizations of quantum computers. He is known for proposing a set of five criteria (commonly called DiVincenzo's checklist) for the physical implementation of quantum computers. He is a Fellow of the American Physical Society and the Editor-in-Chief of the Virtual Journal of Quantum Information.

DiVincenzo received his Ph.D. (1983), M.S.E. (1980) and B.S.E. (1979) from the University of Pennsylvania.