The power of thinning in balanced allocation

Ohad N. Feldheim; Ori Gurel-Gurevich

doi:10.1214/21-ECP400

The power of thinning in balanced allocation

Ohad N. Feldheim, Ori Gurel-Gurevich

Einstein Institute of Mathematics

Research output: Contribution to journal › Article › peer-review

9 Scopus citations

Abstract

Balls are sequentially allocated into n bins as follows: for each ball, an independent, uniformly random bin is generated. An overseer may then choose to either allocate the ball to this bin, or else the ball is allocated to a new independent uniformly random bin. The goal of the overseer is to reduce the load of the most heavily loaded bin after Θ(n) balls have been allocated. We provide an asymptotically optimal strategy √ yielding a maximum load of (1 + o(1)) 8 log n log log n^balls..

Original language	English
Article number	34
Journal	Electronic Communications in Probability
Volume	26
DOIs	https://doi.org/10.1214/21-ECP400
State	Published - 2021

Bibliographical note

Funding Information:
*Research was supported by the Israel Science Foundation (grant No. 1707/16). †Hebrew University of Jerusalem, Israel. E-mail: [email protected] ‡Hebrew University of Jerusalem, Israel. E-mail: [email protected]

Publisher Copyright:
© 2021, Institute of Mathematical Statistics. All rights reserved.

Keywords

(1 + α)-choice
Balanced allocation
Load balancing, balls and bins
One-retry
Subsampling
Thinning
Two-choices

Access to Document

10.1214/21-ECP400

Cite this

@article{ae4228deaad74c13a9e0359165d42cbf,

title = "The power of thinning in balanced allocation",

abstract = "Balls are sequentially allocated into n bins as follows: for each ball, an independent, uniformly random bin is generated. An overseer may then choose to either allocate the ball to this bin, or else the ball is allocated to a new independent uniformly random bin. The goal of the overseer is to reduce the load of the most heavily loaded bin after Θ(n) balls have been allocated. We provide an asymptotically optimal strategy √ yielding a maximum load of (1 + o(1)) 8 log n log log nballs..",

keywords = "(1 + α)-choice, Balanced allocation, Load balancing, balls and bins, One-retry, Subsampling, Thinning, Two-choices",

author = "Feldheim, {Ohad N.} and Ori Gurel-Gurevich",

year = "2021",

doi = "10.1214/21-ECP400",

language = "אנגלית",

volume = "26",

journal = "Electronic Communications in Probability",

issn = "1083-589X",

publisher = "Institute of Mathematical Statistics",

}

TY - JOUR

T1 - The power of thinning in balanced allocation

AU - Feldheim, Ohad N.

AU - Gurel-Gurevich, Ori

PY - 2021

Y1 - 2021

N2 - Balls are sequentially allocated into n bins as follows: for each ball, an independent, uniformly random bin is generated. An overseer may then choose to either allocate the ball to this bin, or else the ball is allocated to a new independent uniformly random bin. The goal of the overseer is to reduce the load of the most heavily loaded bin after Θ(n) balls have been allocated. We provide an asymptotically optimal strategy √ yielding a maximum load of (1 + o(1)) 8 log n log log nballs..

AB - Balls are sequentially allocated into n bins as follows: for each ball, an independent, uniformly random bin is generated. An overseer may then choose to either allocate the ball to this bin, or else the ball is allocated to a new independent uniformly random bin. The goal of the overseer is to reduce the load of the most heavily loaded bin after Θ(n) balls have been allocated. We provide an asymptotically optimal strategy √ yielding a maximum load of (1 + o(1)) 8 log n log log nballs..

KW - (1 + α)-choice

KW - Balanced allocation

KW - Load balancing, balls and bins

KW - One-retry

KW - Subsampling

KW - Thinning

KW - Two-choices

UR - http://www.scopus.com/inward/record.url?scp=85109304061&partnerID=8YFLogxK

U2 - 10.1214/21-ECP400

DO - 10.1214/21-ECP400

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:85109304061

SN - 1083-589X

VL - 26

JO - Electronic Communications in Probability

JF - Electronic Communications in Probability

M1 - 34

ER -

The power of thinning in balanced allocation

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this