Summary
A general method is proposed for assigning a probability distribution to a random matrix. Its principle is that the amount of information contained in the probability distribution should not exceed the minimum amount needed to satisfy relevant properties of the matrix. As examples, classical random matrices are recovered, and the case of a given density of eigenvalues is treated.
Riassunto
Si propone un metodo generale per attribuire una distribuzione di probabilità ad una matrice aleatoria. Il metodo si basa sul principio che la distribuzione di probabilità non debba contenere informazioni maggiori del minimo necessario per soddisfare le proprietà importanti della matrice. Come esempio, si ritrovano le matrici aleatorie classiche e si tratta il caso di una densità data di autovalori.
Реэюме
Предлагается обший метод для приписывания распределения вероятности для случайной матрицы. Суть метода состоит в том, что количество информации, содержашейся в распределении вероятности, не должно превыщать минимального количества, необходимого для удовлетворения соответствуюших свойств матрицы. Как пример, эаново выводятся классические случайные матрицы, и рассматривается случай для эаданной плотности собственных эначений.
Similar content being viewed by others
References
Statistical Theories of Spectra: Fluctuations,C. E. Porter, Ed. (New York, 1965), hereafter referred to as STSF.
M. L. Mehta:Random Matrices and the Theory of Energy Levels (New York, 1967).
E. P. Wigner:SIAM Rev.,9, 1 (1967).
E. P. Wigner:Proc. Fourth Can. Math. Congr. (Toronto, 1957), p. 174; STSF, p. 188.
M. L. Mehta andM. Gaudin:Nucl. Phys.,18, 420 (1960); STSF, p. 342;B. V. Bronk:Journ. Math. Phys.,5, 215 (1964).
F. J. Dyson:Journ. Math. Phys.,3, 140 (1962); STSF, p. 379.
E. P. Wigner: STSF, p. 446.
J. Ginibre:Journ. Math. Phys.,6, 440 (1965).
C. E. Porter andN. Rosenzweig:Suomalaisen Tiedeakatemian Toimituksia, vol. A-VI, no. 44 (1960), p. 19; STSF, p. 251.
N. Rosenzweig andC. E. Porter:Phys. Rev.,120, 1698 (1960); STSF, p. 312;F. J. Dyson:Journ. Math. Phys.,3, 1191 (1962); STSF, p. 424;N. Rosenzweig, J. E. Monahan andM. L. Mehta:Nucl. Phys., A109, 437 (1968).
D. Fox andP. B. Kahn:Phys. Rev.,134, B 1151 (1964);H. S. Leff:Journ. Math. Phys.,5, 756, 763 (1964);B. V. Bronk:Journ. Math. Phys.,6, 228 (1965).
B. V. Bronk:Thesis, Princeton University (1964).
C. E. Porter: STSF, p. 49.Bronk was the first to use information theory (ref. (11,12)); his work needed two improvements. On the one hand, instead of the unique constraint (8), which is sufficient to obtain the Gaussian ensemble (9), the variances and the correlations ofall matrix elementsM αβ are assumed to be given, without justification for the absence of correlations, nor for the factor √2 between variances of diagonal and off-diagonal matrix elements. On the other hand, the volume element d[M] entering in the definition of information for positive random matrices is chosen somewhat arbitrarily; this is troublesome, since one may derive, from a fixed set of constraints, anyP[M] by simply changing the definition of the basic measure d[M].
L. Brillouin:Science and Information Theory (New York, 1956);E. T. Jaynes:Phys. Rev.,106, 620 (1957);A. Katz:Principles of Statistical Mechanics, The Information Theory Approach (New York, 1967).
C. E. Shannon andW. Weaver:The Mathematical Theory of Communication (Urbana, Ill.,, 1949);A. Feinstein:Foundations of Information Theory (New York, 1958);D. Middleton:Introduction to Statistical Communication Theory, Chap. 6 (New York, 1960).
C. E. Porter: STSF, p. 32.
N. Rosenzweig: inStatistical Physics, Brandeis Lectures, vol.3 (New York, 1963), p. 91.
F. J. Dyson:Journ. Math. Phys.,3, 154 (1962); STSF, p. 393.
H. A. Bethe:Rev. Mod. Phys.,9, 53 (1937);C. Bloch:Phys. Rev.,93, 1094 (1954);J. M. B. Lang andK. J. Le Couteur:Proc. Phys. Soc.,67 A, 586 (1954);T. D. Newton:Can. Journ. Phys.,34, 804 (1956).
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Balian, R. Random matrices and information theory. Nuovo Cimento B (1965-1970) 57, 183–193 (1968). https://doi.org/10.1007/BF02710326
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF02710326