๐”– Bobbio Scriptorium
โœฆ   LIBER   โœฆ

Characters and Random Walks on Finite Classical Groups

โœ Scribed by David Gluck


Publisher
Elsevier Science
Year
1997
Tongue
English
Weight
447 KB
Volume
129
Category
Article
ISSN
0001-8708

No coin nor oath required. For personal study only.

โœฆ Synopsis


Let G be a finite group and E a generating set for G. Let P be a probability measure on G whose support is E. We define a random walk on G as follows. At the zeroth stage, we set w 0 =1. At the k th stage, we set w k =w k&1 x, where x # E is chosen with probability P(x). For g # G, the probability that w k = g is given by P* k (g), where P* k denotes the k-fold convolution power. We are interested in when, and more importantly, how fast P* k converges to the uniform distribution U on G.

Character theory of finite groups is directly relevant in the important special case that E is a union of conjugacy classes and P is constant on conjugacy classes. In this case we let P = g # G P( g) g # Z(C[G]). Then P* k converges to U if and only if P k converges to the primitive central idempotent e 1 =|G| &1 g # G g. Using the familiar central characters | / : Z(C[G]) ร„ C defined for each / # Irr(G) by | / (z)=/(z)ร‚/(1), one sees that P k converges to e 1 if and only if no nonprincipal degree one character of G is constant on E. Character theory is indirectly relevant even when E is not a union of conjugacy classes, since the rate of convergence for a random walk based on E can be compared to the rate for another set F which is a union of conjugacy classes; see [4]. From now on we assume that E is a union of conjugacy classes and P is constant on conjugacy classes.

It was shown in [5] that near-uniformity is achieved in n log n steps when G=S n and E is the set of all transpositions in G (actually ``parity problems'' require that the identity be included in E). This has an important analog when G is a finite classical group over GF(q) of rank n, with natural module V. We fix a positive integer d and require that E consist of elements x such that dim V(x&1) is at most d. We want to show that near-uniformity is achieved in O(n) steps, where the implied constant depends only on d and q. Repeated use of the relation V(xy&1)/ V(x&1) y+V( y&1) shows that at least O(n) steps are needed. When G=SL(n+1, q) and E is the set of all transvections in G, Hildebrand [10] article no. AI961635 46


๐Ÿ“œ SIMILAR VOLUMES


Quantum Simulations of Classical Random
โœ John Watrous ๐Ÿ“‚ Article ๐Ÿ“… 2001 ๐Ÿ› Elsevier Science ๐ŸŒ English โš– 160 KB

While it is straightforward to simulate a very general class of random processes space-efficiently by non-unitary quantum computations (e.g., quantum computations that allow intermediate measurements to occur), it is not currently known to what extent restricting quantum computations to be unitary a

Entropy and Dyadic Equivalence of Random
โœ Deborah Heicklen; Christopher Hoffman; Daniel J. Rudolph ๐Ÿ“‚ Article ๐Ÿ“… 2000 ๐Ÿ› Elsevier Science ๐ŸŒ English โš– 192 KB

For any 1 1 measure-preserving map T of a probability space, consider the [T, T &1 ] endomorphism and the corresponding decreasing sequence of \_-algebras. We demonstrate that if the decreasing sequence of \_-algebras generated by [T, T &1 ] and [S, S &1 ] are isomorphic, then T and S must have equa

Random Walk in an Alcove of an Affine We
โœ David J. Grabiner ๐Ÿ“‚ Article ๐Ÿ“… 2002 ๐Ÿ› Elsevier Science ๐ŸŒ English โš– 159 KB

We use a reflection argument, introduced by Gessel and Zeilberger, to count the number of k-step walks between two points which stay within a chamber of a Weyl group. We apply this technique to walks in the alcoves of the classical affine Weyl groups. In all cases, we get determinant formulas for th

Random walks on random simple graphs
โœ Martin Hildebrand ๐Ÿ“‚ Article ๐Ÿ“… 1996 ๐Ÿ› John Wiley and Sons ๐ŸŒ English โš– 676 KB

This paper looks at random regular simple graphs and considers nearest neighbor random walks on such graphs. This paper considers walks where the degree d of each vertex is around (logn)", where a is a constant which is at least 2 and where n is the number of vertices. By extending techniques of Dou