Abstract:Allocation of balls into bins is a well studied abstraction for load balancing <a class="link-external link-http" href="http://problems.The" rel="external noopener nofollow">this http URL</a> literature hosts numerous results for sequential(single dimensional) allocation case when m balls are thrown into n bins. In this paper we study the symmetric multiple choice process for both unweighted and weighted balls as well as for both multidimensional and scalar <a class="link-external link-http" href="http://models.Additionally" rel="external noopener nofollow">this http URL</a>,we present the results on bounds on gap for (1+beta) choice process with multidimensional balls and bins. We show that for the symmetric d choice process and with m=O(n), the upper bound on the gap is O(lnln(n)) <a class="link-external link-http" href="http://w.h.p.This" rel="external noopener nofollow">this http URL</a> upper bound on the gap is within D=f factor of the lower bound. This is the first such tight <a class="link-external link-http" href="http://result.For" rel="external noopener nofollow">this http URL</a> the general case of m>>n the expected gap is bounded by O(lnln(n)).For variable f and non-uniform distribution of the populated dimensions,we obtain the upper bound on the expected gap as O(log(n)). Further,for the multiple round parallel balls and bins,we show that the gap is also bounded by O(loglog(n)) for m=O(n).The same bound holds for the expected gap when m>>n. Our analysis also has strong implications in the sequential scalar <a class="link-external link-http" href="http://case.For" rel="external noopener nofollow">this http URL</a> the weighted balls and bins and general case m>>n,we show that the upper bound on the expected gap is O(log(n)) which improves upon the best prior bound of n^<a class="link-external link-http" href="http://c.Moreover" rel="external noopener nofollow">this http URL</a>,we show that for the (1 + beta) choice process and m=O(n) the upper bound(assuming uniform distribution of f populated dimensions over D total dimensions) on the gap is O(log(n)/beta),which is within D=f factor of the lower <a class="link-external link-http" href="http://bound.For" rel="external noopener nofollow">this http URL</a> fixed f with non-uniform distribution and for random f with Binomial distribution the expected gap remains O(log(n)/beta) independent of the total number of balls thrown. This is the first such tight result for (1 +beta) paradigm with multidimensional balls and bins.

The Power of Filling in Balanced Allocations

The Power of Filling in Balanced Allocations

Balanced Allocations: Caching and Packing, Twinning and Thinning

Mean-Biased Processes for Balanced Allocations

A Best Cost-Sharing Rule for Selfish Bin Packing

Best Cost-Sharing Rule Design for Selfish Bin Packing

Online Algorithms for 1-Space Bounded 2-Dimensional Bin Packing and Square Packing

A New Upper Bound on 2D Online Bin Packing

Cost-Sharing Mechanisms For Selfish Bin Packing

A New Upper Bound 2.5545 on 2D Online Bin Packing.

Dynamic Bin Packing with Unit Fraction Items Revisited

From Packing Rules to Cost-Sharing Mechanisms.

On-Line Extensible Bin Packing with Unequal Bin Sizes

Load Balancing with Dynamic Set of Balls and Bins

Bin Packing with "largest in Bottom" Constraint: Tighter Bounds and Generalizations

Multidimensional Balanced Allocation for Multiple Choice & (1 + Beta) Processes

Parallel Balanced Allocations: The Heavily Loaded Case

Adaptive Bin Packing with Overflow

The power of thinning in balanced allocation

Balanced allocation on hypergraphs

Perfectly Balanced Allocation With Estimated Average Using Expected Constant Retries