DARPA ITO Sponsored Research

1998 Project Summary
Bulk Quantum Computation with NMR
Stanford University

Project Website: Additional project information provided by the performing organization
Objective: The focus of this research program is to investigate and to develop an enabling technology for the near-term realization of practical computing systems based on the non-classical dynamical behavior of quantum scale systems. Such systems offer orders-of-magnitude improvement in computing power beyond foreseeable technology. Through a collaborative engineering and basic science effort, recent results demonstrating the possibility of quantum computation with bulk systems will be leveraged to accomplish the following objectives:
  • Demonstrate basic quantum gates and circuits using NMR techniques
  • Laboratory implementation of quantum algorithms, and tests of fitness for quantum computation
  • Experimental application of quantum error correction codes
  • Development of robust quantum algorithms which tolerate errors
  • Design and construct a desktop scale quantum computer and ultimately, a solid-state microminiaturized system
  • Develop and implement an optimizing "quantum compiler" and language to program quantum computers
Approach: Quantum Gates, Circuits, and Algorithms using NMR
Bulk quantum computation is a radically different approach which, in contrast to trapped ion and cavity-QED implementations, allows quantum computation to be performed using ordinary liquids at room temperature. Applying these methods to nuclear magnetic resonance, and using a simple molecule in solution, a quantum controlled-NOT gate has been realized. Quantum bits (qubits) are represented as nuclear spins, which have long coherence times due to their natural isolation from the external world, and fiducial input states are created using state labeling techniques. Cascading elementary gates will provide an exciting opportunity to experimentally explore quantum algorithms immediately.

Error Correction
Quantum computers, though inherently digital in nature, are notoriously susceptible to errors due to decoherence: leakage of quantum information to the environment. This reality can be countered by application of error correction techniques, which restore the system to a set of standardized states. Methods which generalize known classical error correction codes to work on quantum systems have been discovered, but these codes are complicated; the simplest requires five qubits to encode one. Practical quantum computers require simple, expedient codes which are short and easy to encode and decode, and target real-world errors like those resulting from T1 and T2 processes in NMR. Such codes will be developed and their application will be explored using high-field NMR techniques in the laboratory.

Robust Quantum Algorithms
Quantum computation can be made reliable at the expense of intermixing quantum error correction techniques with quantum algorithms. This is currently the subject of serious theoretical investigations; however, it can also be seen that algorithms such as Shor's quantum factoring algorithm actually use decoherence as an integral part of their procedures. In fact, simulated annealing provides an excellent classical example of the usefulness of controlled loss (heating and cooling via coupling to an external reservoir). Application of such notions toward quantum computation would provide a way to perform quantum computation with decohering systems, and thus achieve a natural robustness. These algorithms would provide practical means for accomplishing quantum computation with current technology, such as NMR.

Quantum Signature Tests
Given a black box, what distinctive signature would its dynamics have which certifiably demonstrate its capacity to perform quantum (vs. classical) computation? From the theoretical point of view, an NMR quantum computer can be treated as a black box, as can other physical implementations. For the purposes of quantum computation, what is most important is the existence of an exponential number of degrees of freedom which are manipulated by a polynomial number of "knobs." A suitability benchmark will be designed to test for this relationship, and implemented on an NMR quantum computer.

Quantum Computer Compiler
Quantum algorithms are implemented with quantum circuits, which are networks of abstract elements such as the controlled-NOT and square-root-of-NOT gates. An important application, which will complement network design and simulation tools, is the "back-end" compiler, which outputs actual control instructions optimized for a particular physical machine model, such as an ion trap or NMR spectrometer. Depending on the experimental configuration, many simplifications can be made, and due to the imperfect nature of current experiments, such optimizations may be crucial to practical operation. For example, sequential single qubit rotations can be combined using the rules of SU(2) Lie algebra. At present, simple quantum circuits are hand-coded into pulse sequences for NMR quantum computation. An optimizing compiler will be developed to automate this important function.

Desktop Quantum Computer
Quantum computers will not be practically useful until they are either large enough to out-perform the best classical computers, or physically small enough to be useful as integrated co-processors. A technological opportunity afforded by the invention of bulk quantum computers is the development of such a quantum co-processor. As a step in this direction, a desktop scale quantum computer will be designed and implemented, using permanent magnet NMR techniques, and targeted multi-qubit molecules. Ultimately, the envisioned system is a solid-state quantum computer, realized by modern semiconductor materials technology using micromachining of complex quantum structures.

Recent FY-97 Accomplishments:

This project was initiated 8/1/97 with the goal of experimentally realizing small quantum computers using nuclear magnetic resonance (NMR) techniques. The collaboration involves three groups: Stanford developing algorithms, numerical models and micromachine based NMR technology, U.C. Berkeley synthesizing molecules and implementing algorithms at their high magnetic field NMR facility, and MIT investigating scaling to 100's of quantum bits and desktop size NMRQC apparatus. From 8/1/97 to 2/1/98 the consortium achieved a major success, by demonstrating implementations of two quantum algorithms using molecules of chloroform as our computer. These results represent the first laboratory realizations of quantum algorithms, and are reported in the two publications below. From these studies, the consortium has developed a better understanding of the potential for scaling up NMR quantum computers. In the coming year, the consortium plans to synthesize and test quantum computers with 3-4 bits, and to further investigate means for scaling these systems to significant sizes. The major contributions of each group are listed below.

Stanford University
Stanford's three accomplishments in this period relate to quantum algorithms, microfabricated devices, and error correction. First, Stanford designed and performed experiments implementing the Deutsch-Jozsa and Deutsch-Jozsa Cleve quantum algorithms, which exhibit an exponential speedup on a quantum computer, compared with a classical computer. These results are reported in the two papers referenced below. Second, Stanford fabricated and tested a series of planar RF-microcoils to develop a basis for a future implementation of a silicon microfabricated NMR quantum computer. This project brought on board a new graduate student, who is now expert in using our Ginzton Laboratory Microfabrication Facility. Third, Stanford developed a new experiment which demonstrates quantum error correction being performed with NMR. This experiment is in progress.

U.C. Berkeley
U.C. Berkeley fulfilled two main goals during this period, relating to molecular synthesis and quantum algorithms. First, special high resolution samples of several two and three qubit molecules were synthesized, including chloroform and alanine. Molecules with more qubits have been identified and their custom synthesis requirements are being explored. Second, Grover's fast quantum search algorithm was implemented using chloroform as our two qubit computer. This demonstrates the loading of a quantum computer with an initial state, performing a computation requiring fewer steps than a classical computer, and reading out the final state. The focus is to obtain NMR samples of four and five qubit molecules and implementation of quantum computing algorithms including quantum error correction developed at Stanford. The molecular synthesis and development of NMR pulse sequences is underway.

MIT
MIT's primary role in this collaborative effort has been to scale bulk quantum computing from million-dollar NMR spectrometers to the table-top, both to make it more widely accessible, and to overcome the limits of an instrument designed for spectroscopy instead of computation. The first step in this effort was an analysis of the experimental scaling constraints, leading to the construction of prototypes using electromagnets and permanent magnets. These are now resolving J-couplings and hence close to reproducing the high-field experiments. In the coming year MIT will be focusing on online shimming down to parts per billion, the pre-polarization to increase the number of qubits, and writing the compiler for pulse sequences.

Publications

  1. Isaac L. Chuang, Neil Gershenfeld, and Mark Kubinec, "Experimental Implementation of Fast Quantum Searching," Physical Review Letters 80, 3408 (1998).
  2. Isaac L. Chuang, Lieven M. K. Vandersypen, Xinlan Zhou, Debbie W. Leung, and Seth Lloyd, "Experimental realization of a quantum algorithm," Nature 393, 143 (1998).
FY-98 Plans:
  • Demonstrate cascading of multiple quantum gates
  • Experimentally demonstrate a superfast quantum algorithm
  • Characterize new quantum error correction code performance
  • Synthesize and test a NMR quantum computer with 3-4 bits
Technology Transition: The consortium is working very closely with technology partners including IBM, HP, Motorola, Analog Devices, and Microsoft on transferring the instrumentation (an early version went to DARPA). In addition results have been communicated in technical articles (Science, Proceedings of the Royal Society, Physical Review Letters) and the popular press (New Scientist, The Economist, Discover, Scientific American). An invited paper on NMR Quantum Computation by Chuang, Vandersypen and Harris was presented at the IEEE International Solid State Circuits Conference. Also, joint seminars have been held at Stanford on Quantum Computation and Error Correction with Hewlett-Packard.
Principal Investigator: Isaac Chuang
IBM Almaden Research Center
650 Harry Rd., K10/D1
San Jose, CA 95120-6099
408-927-2845
ichuang@almaden.ibm.com

Neil Gershenfeld
MIT Media Lab, Rm. E 15-495
20 Ames St.
Cambridge, MA 02139-4307
617-253-7680
617-258-7168 fax
neilg@media.mit.edu

James Harris
Paul Allen Center for Integrated Systems
Rm. 328
Stanford, CA 94305-4075
650-723-9775
650-723-4659 fax
Harris@snow.stanford.edu

Gail Chun-Creech
CIS-327
Stanford, CA 94305-4075
650-723-0983
650-723-4659 fax
creech@snow.stanford.edu