- Access by Xinjiang University
Computing with distributed chaos
Phys. Rev. E 60, 363 – Published 1 July, 1999
DOI: https://doi.org/10.1103/PhysRevE.60.363
Abstract
We describe and discuss in detail some recent results by Sinha and Ditto [Phys. Rev. Lett. 81, 2156 (1998)] demonstrating the capacity of a lattice of threshold coupled chaotic maps to perform computations. Such systems are shown to emulate logic gates, encode numbers, and perform specific arithmetic operations, such as addition and multiplication, as well as yield more specialized operations such as the calculation of the least common multiplier of a sequence of numbers. Furthermore, we extend the scheme to multidimensional continuous time dynamics, in particular to a system relevant to chaotic lasers.
References (26)
- T. Shinbrot, C. Grebogi, E. Ott, and J. Yorke, Nature (London) 363, 411 (1993).
- L. M. Pecora and T. L. Carroll, Phys. Rev. A 44, 2374 (1991); W. L. Ditto and L. M. Pecora, Sci. Am. (Int. Ed.) 269, 62 (1993).
- S. Hayes, C. Grebogi, E. Ott, and A. Mark, Phys. Rev. Lett. 73, 1781 (1994).
- G. D. Van Wiggeren and R. Roy, Science 279, 1198 (1997).
- S. Sinha and W. L. Ditto, Phys. Rev. Lett. 81, 2156 (1998).
- C. Moore, Phys. Rev. Lett. 64, 2354 (1990).
- A. V. Holden et al., Chaos 2, 367 (1992).
- A. Toth and K. Showalter, J. Chem. Phys. 103, 2058 (1995).
- S. Sinha and D. Biswas, Phys. Rev. Lett. 71, 2010 (1993).
- S. Sinha, Phys. Rev. E 49, 4832 (1994); Int. J. Mod. Phys. B 9, 875 (1995).
- S. Sinha, Phys. Lett. A 199, 365 (1995).
- P. Bak, C. Tang, and K. Wiesenfeld, Phys. Rev. Lett. 59, 381 (1987); Phys. Rev. A 38, 364 (1988).
- There is an additional feature that can potentially lend flexibility to the response of the chaotic network. The variation in the number of updates after which the emitted excess is “measured” also lends rich variety to the emergent responses. This feature is exploited later in an arithmetic application.
- To do the multiplication one can also feed in the excess emerging from the open edge of the lattice representing m back to the first element for n time steps. After that one can simply measure the excess ejected from the open edge, which gives the “answer.”
- U. Hubner, N. B. Abraham, and C. O. Weiss, Phys. Rev. A 40, 6354 (1989); C. O. Weiss et al., Appl. Phys. B: Lasers Opt. 61, 223 (1995).
- N. Margolus, Physica D 10, 81 (1984); T. Toffoli and N. Margolus, Cellular Automata Machines: A New Environment for Modeling (MIT Press, Cambridge, 1987); Physica D 47, 263 (1990).
- J. P. Crutchfield and K. Young, Phys. Rev. Lett. 63, 105 (1989); J. P. Crutchfield, Physica D 75, 11 (1994). The above efforts by Crutchfield and Young were primarily focused on the alternate question: Can the theory of computation help describe/quantify the complexity of physical systems. Here we are approaching quite the opposite question. We want to harness (in an explicit realization) a very complex physical system to do computations for us (in a controlled, direct, and hopefully clean and simple fashion). So while Crutchfield’s work has bearing on the computational effort required in modeling complex behavior, which is used to quantify the emergence of complexity, it does not exploit the computational capability of nonlinear processes to solve “tasks” extrinsic to it.
- G. Taubes, Science 277, 1935 (1997).
- L. M. Aldeman, Science 266, 1021 (1994).
- R. Pool, Science 268, 498 (1995).
- P. W. Shor, in Proceedings of the 35th Annual Symposium on the Foundations of Computer Science, edited by S. Goldwasser (IEEE Computer Society Press, Los Alamitos, 1994), p. 124.
- A. Steane, Rep. Prog. Phys. 61, 117 (1998).
- In fact, after years of intense research, some fundamental problems still remain in DNA and quantum computing paradigms. For instance, quantum computers are exponentially more sensitive to external noise and to slight deviations of their components from their design specifications. Quantum computing relies heavily on the quantum correlations of entangled states and these are extremely sensitive to thermal noise, which would tend to randomize each bit separately, making the whole system “decohere.” All pure states of a quantum system are connected by a continuous symmetry, so there are no discrete positions in which to stabilize such states. Moreover, any mechanism working “locally” on each qubit separately would inevitably destroy entanglement. Controlling large-scale entanglement is thus a fundamental issue in quantum computing that is yet to be solved. In DNA computing, there is the problem of really slow molecular biological operations and the hazard of fracture in the DNA molecule. (Unlike biological computers, we are fortunately free to design and use very fast dynamical systems ranging from electronic circuits to chaotic lasers.) There is also the inability to transmit information from one molecule to another in a DNA computer, which limits their flexibility. Further DNA computing involves somewhat random operations, that is to say, inherently noisy components (unlike the determinism of our prototype).
- Of course the speed of these devices will depend strongly on how they are implemented. To make the proposal work well we need to design elements where the threshold can be set/reset easily and this implementation issue also sets the possible limitations.
- SCA is a model used in computer science to give a unified theory for a number models of parallel computing (including neural nets) [7].
- I. Aleksander and H. Morton, An Introduction to Neural Computing (Champman and Hall, London, 1990).