Testing lipschitz functions on hypergrid domains

Pranjal Awasthi, Madhav Jha, Marco Molinaro, Sofya Raskhodnikova

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

A function f(x 1, ... , x d ), where each input is an integer from 1 to n and output is a real number, is Lipschitz if changing one of the inputs by 1 changes the output by at most 1. In other words, Lipschitz functions are not very sensitive to small changes in the input. Our main result is an efficient tester for the Lipschitz property of functions f : [n] d → δℤ, where δ ∈ (0,1] and δℤ is the set of integer multiples of δ. The main tool in the analysis of our tester is a smoothing procedure that makes a function Lipschitz by modifying it at a few points. Its analysis is already nontrivial for the 1-dimensional version, which we call Bubble Smooth, in analogy to Bubble Sort. In one step, Bubble Smooth modifies two values that violate the Lipschitz property, i.e., differ by more than 1, by transferring δ units from the larger to the smaller. We define a transfer graph to keep track of the transfers, and use it to show that the ℓ 1 distance between f and BubbleSmooth(f) is at most twice the ℓ 1 distance from f to the nearest Lipschitz function. Bubble Smooth has other important properties, which allow us to obtain a dimension reduction, i.e., a reduction from testing functions on multidimensional domains to testing functions on the 1-dimensional domain, that incurs only a small multiplicative overhead in the running time and thus avoids the exponential dependence on the dimension.

Original languageEnglish (US)
Title of host publicationApproximation, Randomization, and Combinatorial Optimization
Subtitle of host publicationAlgorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings
Pages387-398
Number of pages12
DOIs
StatePublished - Aug 28 2012
Event15th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2012 and the 16th International Workshop on Randomization and Computation, RANDOM 2012 - Cambridge, MA, United States
Duration: Aug 15 2012Aug 17 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7408 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2012 and the 16th International Workshop on Randomization and Computation, RANDOM 2012
CountryUnited States
CityCambridge, MA
Period8/15/128/17/12

Fingerprint

Lipschitz Function
Lipschitz Property
Bubble
Testing
Bubble sort
Integer
Output
Dimension Reduction
Violate
Lipschitz
Analogy
Smoothing
Multiplicative
Unit
Graph in graph theory

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Awasthi, P., Jha, M., Molinaro, M., & Raskhodnikova, S. (2012). Testing lipschitz functions on hypergrid domains. In Approximation, Randomization, and Combinatorial Optimization: Algorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings (pp. 387-398). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7408 LNCS). https://doi.org/10.1007/978-3-642-32512-0_33
Awasthi, Pranjal ; Jha, Madhav ; Molinaro, Marco ; Raskhodnikova, Sofya. / Testing lipschitz functions on hypergrid domains. Approximation, Randomization, and Combinatorial Optimization: Algorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings. 2012. pp. 387-398 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{b870919f913c442a83dff1e2a9f69108,
title = "Testing lipschitz functions on hypergrid domains",
abstract = "A function f(x 1, ... , x d ), where each input is an integer from 1 to n and output is a real number, is Lipschitz if changing one of the inputs by 1 changes the output by at most 1. In other words, Lipschitz functions are not very sensitive to small changes in the input. Our main result is an efficient tester for the Lipschitz property of functions f : [n] d → δℤ, where δ ∈ (0,1] and δℤ is the set of integer multiples of δ. The main tool in the analysis of our tester is a smoothing procedure that makes a function Lipschitz by modifying it at a few points. Its analysis is already nontrivial for the 1-dimensional version, which we call Bubble Smooth, in analogy to Bubble Sort. In one step, Bubble Smooth modifies two values that violate the Lipschitz property, i.e., differ by more than 1, by transferring δ units from the larger to the smaller. We define a transfer graph to keep track of the transfers, and use it to show that the ℓ 1 distance between f and BubbleSmooth(f) is at most twice the ℓ 1 distance from f to the nearest Lipschitz function. Bubble Smooth has other important properties, which allow us to obtain a dimension reduction, i.e., a reduction from testing functions on multidimensional domains to testing functions on the 1-dimensional domain, that incurs only a small multiplicative overhead in the running time and thus avoids the exponential dependence on the dimension.",
author = "Pranjal Awasthi and Madhav Jha and Marco Molinaro and Sofya Raskhodnikova",
year = "2012",
month = "8",
day = "28",
doi = "10.1007/978-3-642-32512-0_33",
language = "English (US)",
isbn = "9783642325113",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "387--398",
booktitle = "Approximation, Randomization, and Combinatorial Optimization",

}

Awasthi, P, Jha, M, Molinaro, M & Raskhodnikova, S 2012, Testing lipschitz functions on hypergrid domains. in Approximation, Randomization, and Combinatorial Optimization: Algorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7408 LNCS, pp. 387-398, 15th International Workshop on Approximation Algorithms for Combinatorial Optimization Problems, APPROX 2012 and the 16th International Workshop on Randomization and Computation, RANDOM 2012, Cambridge, MA, United States, 8/15/12. https://doi.org/10.1007/978-3-642-32512-0_33

Testing lipschitz functions on hypergrid domains. / Awasthi, Pranjal; Jha, Madhav; Molinaro, Marco; Raskhodnikova, Sofya.

Approximation, Randomization, and Combinatorial Optimization: Algorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings. 2012. p. 387-398 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7408 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Testing lipschitz functions on hypergrid domains

AU - Awasthi, Pranjal

AU - Jha, Madhav

AU - Molinaro, Marco

AU - Raskhodnikova, Sofya

PY - 2012/8/28

Y1 - 2012/8/28

N2 - A function f(x 1, ... , x d ), where each input is an integer from 1 to n and output is a real number, is Lipschitz if changing one of the inputs by 1 changes the output by at most 1. In other words, Lipschitz functions are not very sensitive to small changes in the input. Our main result is an efficient tester for the Lipschitz property of functions f : [n] d → δℤ, where δ ∈ (0,1] and δℤ is the set of integer multiples of δ. The main tool in the analysis of our tester is a smoothing procedure that makes a function Lipschitz by modifying it at a few points. Its analysis is already nontrivial for the 1-dimensional version, which we call Bubble Smooth, in analogy to Bubble Sort. In one step, Bubble Smooth modifies two values that violate the Lipschitz property, i.e., differ by more than 1, by transferring δ units from the larger to the smaller. We define a transfer graph to keep track of the transfers, and use it to show that the ℓ 1 distance between f and BubbleSmooth(f) is at most twice the ℓ 1 distance from f to the nearest Lipschitz function. Bubble Smooth has other important properties, which allow us to obtain a dimension reduction, i.e., a reduction from testing functions on multidimensional domains to testing functions on the 1-dimensional domain, that incurs only a small multiplicative overhead in the running time and thus avoids the exponential dependence on the dimension.

AB - A function f(x 1, ... , x d ), where each input is an integer from 1 to n and output is a real number, is Lipschitz if changing one of the inputs by 1 changes the output by at most 1. In other words, Lipschitz functions are not very sensitive to small changes in the input. Our main result is an efficient tester for the Lipschitz property of functions f : [n] d → δℤ, where δ ∈ (0,1] and δℤ is the set of integer multiples of δ. The main tool in the analysis of our tester is a smoothing procedure that makes a function Lipschitz by modifying it at a few points. Its analysis is already nontrivial for the 1-dimensional version, which we call Bubble Smooth, in analogy to Bubble Sort. In one step, Bubble Smooth modifies two values that violate the Lipschitz property, i.e., differ by more than 1, by transferring δ units from the larger to the smaller. We define a transfer graph to keep track of the transfers, and use it to show that the ℓ 1 distance between f and BubbleSmooth(f) is at most twice the ℓ 1 distance from f to the nearest Lipschitz function. Bubble Smooth has other important properties, which allow us to obtain a dimension reduction, i.e., a reduction from testing functions on multidimensional domains to testing functions on the 1-dimensional domain, that incurs only a small multiplicative overhead in the running time and thus avoids the exponential dependence on the dimension.

UR - http://www.scopus.com/inward/record.url?scp=84865288626&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84865288626&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-32512-0_33

DO - 10.1007/978-3-642-32512-0_33

M3 - Conference contribution

AN - SCOPUS:84865288626

SN - 9783642325113

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 387

EP - 398

BT - Approximation, Randomization, and Combinatorial Optimization

ER -

Awasthi P, Jha M, Molinaro M, Raskhodnikova S. Testing lipschitz functions on hypergrid domains. In Approximation, Randomization, and Combinatorial Optimization: Algorithms and Techniques - 15th International Workshop, APPROX 2012, and 16th International Workshop, RANDOM 2012, Proceedings. 2012. p. 387-398. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-32512-0_33