Abstract
Simple exact expressions are derived for all the Lyapunov exponents of certainN-dimensional stochastic linear dynamical systems. In the case of the product of independent random matrices, each of which has independent Gaussian entries with mean zero and variance 1/N, the exponents have an exponential distribution asN→∞. In the case of the time-ordered product integral of exp[N −1/2 dW], where the entries of theN×N matrixW(t) are independent standard Wiener processes, the exponents are equally spaced for fixedN and thus have a uniform distribution as N→∞.
Similar content being viewed by others
References
Brandstäter, A., Swift, J., Swinney, H.L., Wolf, A.: Low-dimensional chaos in a hydrodynamic system. Phys. Rev. Lett.51, 1442 (1983)
See e.g., Pastur, L.A.: Spectral properties of disordered systems in one-body approximations. Commun. Math. Phys.75, 179 (1980)
Kunz, H., Souillard, B.: Sur le spectre des opérateurs aux différences finies aléatoires. Commun. Math. Phys.78, 201 (1980)
Ruelle, D.: Large volume limit of the distribution of characteristic exponents in turbulence. Commun. Math. Phys.87, 287 (1982)
Lacroix, J.: Localisation pour l'operateur de Schrödinger aléatoire dans un ruban. Ann. Inst. H. Poincaré40, 97 (1984)
Cohen, J.E., Newman, C.M.: Ann. Probab.12, 283 (1984)
Mehta, M.L.: Random matrices and the statistical theory of energy levels. New York: Academic Press 1967
It was pointed out to us by D. Ruelle that in Livi, R., Politi, A., Ruffo, S.: Distribution of characteristic exponents in the thermodynamic limit, submitted to J. Phys. A, there are numerical results concerning the high-dimensional limit for a certain nonlinear deterministic dynamical system (the “Fermi-Pasta-Ulam β-model” with the energy per particle fixed). They suggest, at least for large energy per particle, an asymptotically uniform distribution of Lyapunov exponents. On the other hand, a uniform distribution of exponents is valid exactly for the continuous time linear stochastic dynamical systems presented as “Modeld” in this paper [see Eq. (15)]. An asymptotically uniform distribution has also been obtained (numerically) for products of certain random symplectic matrices (G. Paladin and A. Vulpiani, 1985 preprint). The exact nature of the universality implicit in these “coincidences” deserves further investigation
Oseledec, V.I.: Tr. Mosk. O.-Va19, 179 (1968) [Trans. Mosc. Math. Soc.19, 197 (1968)]
Raghunathan, M.S.: Isr. J. Math.32, 356 (1979)
See p. 7 of Wigner, E.P.: SIAM Rev.9, 1 (1967)
Wachter, K.: Ann. Probab.6, 1 (1978)
Newman, C.M.: Lyapunov exponents for some products of random matrices: exact expressions and asymptotic distributions. In: Random matrices and their applications. Cohen, J.E., Kesten, H., Newman, C.M. (eds.). Providence, RI: AMS (to appear)
Harris, T.E.: Ann. Probab.9, 232 (1981)
These can also be obtained by first showing that (B(t) T B(t))1/2 is a Brownian motion on the space of positive definite matrices and then applying results of Dynkin, E.B.: Dokl. Akad. Nauk SSSR141, 288 (1961), Izv. Akad. Nauk SSSR Ser. Mat.30, 455 (1966), which are based on the theory of Martin boundaries. Our direct derivation is more elementary
LeJan, Y.: C.R. Acad. Sci. Paris, Sér. I298, 361 (1984)
LeJan, Y.: On isotropic Brownian motions, Z. Wahr. (to appear)
Baxendale, P., Harris, T.E.: Isotropic stochastic flows. Ann. Probab. (to appear)
Norris, J.R., Rogers, L.C.G., Williams, D.: Brownian motions of ellipsoids. Trans. A.M.S. (to appear)
Author information
Authors and Affiliations
Additional information
Communicated by O. E. Lanford
John S. Guggenheim Memorial Fellow. Research supported in part by NSF Grant MCS 80-19384
Rights and permissions
About this article
Cite this article
Newman, C.M. The distribution of Lyapunov exponents: Exact results for random matrices. Commun.Math. Phys. 103, 121–126 (1986). https://doi.org/10.1007/BF01464284
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01464284