Testing the use of lazy constraints in solving area-based adjacency formulations of harvest scheduling models

Sándor F. Tóth, Marc E. McDill, Nóra Könnyü, Sonney George

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

Spatially explicit harvest scheduling models to enforce maximum harvest opening size restrictions often lead to combinatorial problems that are hard to solve. This article shows that the inequalities required by one of the three existing formulations, the Path model are typically lazy. In other words, these constraints are rarely binding during optimization, especially if the maximum opening size is large relative to the average management unit size. By solving 60 hypothetical and 8 real forest problems with varying maximum clearcut sizes and to varying target optimality gaps, we confirm that applying the path constraints only when they are violated during optimization leads to shorter solution times. Although the Lazy Path constraints performed better than the other formulation/solution approaches, the relative superiority of the method was more obvious at larger optimality gaps. Nearly 95% of the problem instances solved fastest with the "lazy" method at a target gap of 1%, and almost 92% solved fastest at 0.05%. At 0.01%, the Lazy Path approach was still superior in the majority of cases, but the percentage was much lower (57%). This is a significant improvement compared with the 14, 10, and 19% shares of the other approaches. If only the real instances are considered, the Lazy Path approach performed best in 68% of the instances with 1 and 0.01% optimality gaps and in 61% of the instances with 0.05% gap. A closer analysis of the results suggests that the relative superiority of the approach increases with problem size and maximum clearcut size.

Original languageEnglish (US)
Pages (from-to)157-176
Number of pages20
JournalForest Science
Volume59
Issue number2
DOIs
StatePublished - Apr 1 2013

Fingerprint

testing
clearcutting
methodology
harvest
method
analysis

All Science Journal Classification (ASJC) codes

  • Forestry
  • Ecology
  • Ecological Modeling

Cite this

Tóth, Sándor F. ; McDill, Marc E. ; Könnyü, Nóra ; George, Sonney. / Testing the use of lazy constraints in solving area-based adjacency formulations of harvest scheduling models. In: Forest Science. 2013 ; Vol. 59, No. 2. pp. 157-176.
@article{d1cdca415177454bbbc44cfea13a1b7c,
title = "Testing the use of lazy constraints in solving area-based adjacency formulations of harvest scheduling models",
abstract = "Spatially explicit harvest scheduling models to enforce maximum harvest opening size restrictions often lead to combinatorial problems that are hard to solve. This article shows that the inequalities required by one of the three existing formulations, the Path model are typically lazy. In other words, these constraints are rarely binding during optimization, especially if the maximum opening size is large relative to the average management unit size. By solving 60 hypothetical and 8 real forest problems with varying maximum clearcut sizes and to varying target optimality gaps, we confirm that applying the path constraints only when they are violated during optimization leads to shorter solution times. Although the Lazy Path constraints performed better than the other formulation/solution approaches, the relative superiority of the method was more obvious at larger optimality gaps. Nearly 95{\%} of the problem instances solved fastest with the {"}lazy{"} method at a target gap of 1{\%}, and almost 92{\%} solved fastest at 0.05{\%}. At 0.01{\%}, the Lazy Path approach was still superior in the majority of cases, but the percentage was much lower (57{\%}). This is a significant improvement compared with the 14, 10, and 19{\%} shares of the other approaches. If only the real instances are considered, the Lazy Path approach performed best in 68{\%} of the instances with 1 and 0.01{\%} optimality gaps and in 61{\%} of the instances with 0.05{\%} gap. A closer analysis of the results suggests that the relative superiority of the approach increases with problem size and maximum clearcut size.",
author = "T{\'o}th, {S{\'a}ndor F.} and McDill, {Marc E.} and N{\'o}ra K{\"o}nny{\"u} and Sonney George",
year = "2013",
month = "4",
day = "1",
doi = "10.5849/forsci.11-040",
language = "English (US)",
volume = "59",
pages = "157--176",
journal = "Forest Science",
issn = "0015-749X",
publisher = "Society of American Foresters",
number = "2",

}

Testing the use of lazy constraints in solving area-based adjacency formulations of harvest scheduling models. / Tóth, Sándor F.; McDill, Marc E.; Könnyü, Nóra; George, Sonney.

In: Forest Science, Vol. 59, No. 2, 01.04.2013, p. 157-176.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Testing the use of lazy constraints in solving area-based adjacency formulations of harvest scheduling models

AU - Tóth, Sándor F.

AU - McDill, Marc E.

AU - Könnyü, Nóra

AU - George, Sonney

PY - 2013/4/1

Y1 - 2013/4/1

N2 - Spatially explicit harvest scheduling models to enforce maximum harvest opening size restrictions often lead to combinatorial problems that are hard to solve. This article shows that the inequalities required by one of the three existing formulations, the Path model are typically lazy. In other words, these constraints are rarely binding during optimization, especially if the maximum opening size is large relative to the average management unit size. By solving 60 hypothetical and 8 real forest problems with varying maximum clearcut sizes and to varying target optimality gaps, we confirm that applying the path constraints only when they are violated during optimization leads to shorter solution times. Although the Lazy Path constraints performed better than the other formulation/solution approaches, the relative superiority of the method was more obvious at larger optimality gaps. Nearly 95% of the problem instances solved fastest with the "lazy" method at a target gap of 1%, and almost 92% solved fastest at 0.05%. At 0.01%, the Lazy Path approach was still superior in the majority of cases, but the percentage was much lower (57%). This is a significant improvement compared with the 14, 10, and 19% shares of the other approaches. If only the real instances are considered, the Lazy Path approach performed best in 68% of the instances with 1 and 0.01% optimality gaps and in 61% of the instances with 0.05% gap. A closer analysis of the results suggests that the relative superiority of the approach increases with problem size and maximum clearcut size.

AB - Spatially explicit harvest scheduling models to enforce maximum harvest opening size restrictions often lead to combinatorial problems that are hard to solve. This article shows that the inequalities required by one of the three existing formulations, the Path model are typically lazy. In other words, these constraints are rarely binding during optimization, especially if the maximum opening size is large relative to the average management unit size. By solving 60 hypothetical and 8 real forest problems with varying maximum clearcut sizes and to varying target optimality gaps, we confirm that applying the path constraints only when they are violated during optimization leads to shorter solution times. Although the Lazy Path constraints performed better than the other formulation/solution approaches, the relative superiority of the method was more obvious at larger optimality gaps. Nearly 95% of the problem instances solved fastest with the "lazy" method at a target gap of 1%, and almost 92% solved fastest at 0.05%. At 0.01%, the Lazy Path approach was still superior in the majority of cases, but the percentage was much lower (57%). This is a significant improvement compared with the 14, 10, and 19% shares of the other approaches. If only the real instances are considered, the Lazy Path approach performed best in 68% of the instances with 1 and 0.01% optimality gaps and in 61% of the instances with 0.05% gap. A closer analysis of the results suggests that the relative superiority of the approach increases with problem size and maximum clearcut size.

UR - http://www.scopus.com/inward/record.url?scp=84875986791&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875986791&partnerID=8YFLogxK

U2 - 10.5849/forsci.11-040

DO - 10.5849/forsci.11-040

M3 - Article

AN - SCOPUS:84875986791

VL - 59

SP - 157

EP - 176

JO - Forest Science

JF - Forest Science

SN - 0015-749X

IS - 2

ER -