Algebra for Optical Computing and Quantum Computing

Thomas Beth, Markus Grassl, Joern Mueller-Quade

Date: July 17th (Wednesday)
Time: 14:30-15:00
Abstract
Optical and quantum mechanical systems enjoy a great intrinsic parallelism that can be used to speed-up computations. As both types of systems are linear, designing these systems leads to the problem of factoring a given matrix into matrices corresponding to the computational primitives of the physical system.
We present an algebraic approach to this "programming" problem for optical computing and quantum computing.
For optical systems, the computational primitives are diffractive elements corresponding to diagonal matrices and wave propagation modelled by the Fresnel transformation. We show that, in principle, any linear transformation can be realized optically up to a scalar factor, but possibly with many diffractive elements. In practice, optical setups with a small number of diffractive elements are of interest. We present a divide and conquer algorithm for the design of 8f-systems, i. e., systems with up to four lenses and three arbitrary diffractive elements.
Computations in quantum mechanical systems are specified by unitary matrices acting on a Hilbert space which is a tensor product of smaller subspaces. The computational primitives are unitary transformations acting on these subspaces. In principle, any unitary transformation can be factored into such basic transformations. For complexity reasons we are interested in a factorization with a small number of factors. We present an algebraic approach to find a "factorization" similar to that for optical systems.
Solving the equations arising in the design problem is equivalent to computing the inverse of a rational mapping. For this mapping, a decomposition can be derived from a tower of unirational function fields. We discuss how to compute the intermediate fields using Groebner bases and primary decomposition. Algorithms using such a functional decomposition solve the nonlinear equations faster than more universal methods.

______________
__________________________________________

Previous page RISC SWP Linz Austria