Entropy Waves, The ZigZag Graph Product, and New Constant熵波锯齿形图的产品和新的常数.pptVIP

Entropy Waves, The Zigzag Graph Product, and NewConstant-Degree Expanders Omer Reingold Salil Vadhan Avi Wigderson Lecturer: Oded Levy Introduction Most random constant degree graphs are good expanders. Most applications need explicit constructions. This lecture deals with the Zigzag product which is used in order to explicitly build expanders. The zigzag product GA = (NA, DA, λA), GB = (DA, DB, λB). We define their zigzag product as a graph which Its vertices are NA × NB (each vertex is represented as a pair (v, u) such that v?NA and u?NB . For each vertex (v, k) we put an edge between a pair (v, k) to (v[k[i]], k[i][j]). Example We can separate it to three steps: Move from one vertex to another vertex at the same cloud. Jump between clouds. Move from one vertex to another vertex at the same cloud Theorem Theorem G = GA Z GB = (NA× DA, DB2, f(λA , λB)). f(λA’λB) ≤ λA + λB + λB 2. λA, λB 1→ f(λA , λB) 1. We know that . We aim to show that for each  Why does it work? Given a distribution vector π over G’s vertices, we distinguish between two cases of π: ? uniform within clouds. ? non-uniform within clouds. We aim to show that after one step in G, π becomes more uniform. ? uniform within clouds. ? uniform within clouds. ? uniform within clouds. ? non-uniform within clouds. ? non-uniform within clouds. ? non-uniform within clouds. Proof Let M be G’s normalized adjacency matrix. We’ll decompose M into three matrices, corresponding to three steps as defined before. Then is a normalized matrix where we connect vertices within each cloud. is a normalized matrix where we connect clouds. Decomposing α The relation between and B is Recall that we want to bound Thus Using triangular inequality we get We’ll first bound . Decomposing α? to its components we have For our analysis we’ll define linear map C such that: Now we’ll bound . We’ll first relate A to ?. Claim: ev is a basis, thus this claim holds for all vectors . Substituting