## Toward More Powerful Recombinations (1995)

Venue: | Proceedings of the 6 th International Conference on Genetic Algorithms |

Citations: | 21 - 9 self |

### BibTeX

@INPROCEEDINGS{Kahng95towardmore,

author = {Andrew B. Kahng and Byung Ro Moon},

title = {Toward More Powerful Recombinations},

booktitle = {Proceedings of the 6 th International Conference on Genetic Algorithms},

year = {1995},

pages = {96--103},

publisher = {Morgan Kaufmann}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper suggests a flexible framework for n-dimensional crossover, consisting of cutting, classification, and copying of genes. We prove that under this framework, any cutting strategy generates two equivalence classes of genes, making the framework appropriate as a crossover scheme. Three notable features of this framework are: (i) it enables more effective use of genes' geographical linkage, (ii) it enables more diverse ways of cutting than traditional multi-point crossover on linear strings or existing 2-dimensional crossover schemes, and (iii) it can be readily used within most existing genetic algorithm implementations, i.e., the underlying problem need not be inherently multidimensional. We provide guidelines for designing a crossover strategy under this framework, along with two example crossovers. Experimental results show that one can design new crossovers under this framework which outperform traditional crossover on linear strings, uniform crossover, and existing two-di...

### Citations

11502 |
Computers and Intractability, A Guide to the Theory of NPCompleteness
- Garey, Johnson
- 1979
(Show Context)
Citation Context ...ze of a bisection is the number of edges whose endpoints are in different subsets in the bisection. The graph bisection problem seeks a bisection with minimum cut size and is well-known to be NP-hard =-=[12]-=-. We tested our GA implementations on five types of benchmark graphs: random graphs, random geometric graphs, random regular graphs with known small bisection widths, caterpillar graphs, and grid grap... |

533 |
Optimization by simulated annealing: An experimental evaluation (part i), preliminary draft
- Johnson, Aragon, et al.
- 1987
(Show Context)
Citation Context ...raphs, random regular graphs with known small bisection widths, caterpillar graphs, and grid graphs. These have been extensively studied in the discrete algorithms and optimization literatures, e.g., =-=[14, 15, 6, 19]-=-. 2 All implementations in this paper used steady-state GAs with five cuts for crossover unless otherwise noted; we use the code of [6] except for the crossover part. 3 The GAs stop when 80% of the po... |

455 | The Geometry of Graphs and some of Its Algorithmic Applications
- Linial, London, et al.
- 1995
(Show Context)
Citation Context ...tive to traditional linear crossover performance when a linear encoding is used [4, 5]. We believe that this phenomenon will persist with high-dimensional encodings. Recently, Linial-LondonRabinovich =-=[17]-=- proved that we need a fairly highdimensional encoding to embed a graph with low distortion. Since a too high-dimensional encoding is hard to implement, a practical alternative is to embed the graph i... |

321 | Performance Analysis of k-ary n-cube Interconnection Networks
- Dally
- 1990
(Show Context)
Citation Context ...onal linear encoding [7]. In the following discussion, we assume an l-ary n-cube structured chromosome. This structure is analogous to, e.g., the node structure of interconnection networks studied in =-=[10, 1]-=-. We use locus-based encodings with this chromosome structure. For example, linear locusbased encodings for graph problems mostly map each vertex to a fixed corresponding position (locus) on the linea... |

182 | Limits on interconnection network performance
- Agarwal
- 1991
(Show Context)
Citation Context ...onal linear encoding [7]. In the following discussion, we assume an l-ary n-cube structured chromosome. This structure is analogous to, e.g., the node structure of interconnection networks studied in =-=[10, 1]-=-. We use locus-based encodings with this chromosome structure. For example, linear locusbased encodings for graph problems mostly map each vertex to a fixed corresponding position (locus) on the linea... |

135 | Using genetic algorithms to solve NP-complete problems
- Jong, Spears
(Show Context)
Citation Context ...locus-based encodings with this chromosome structure. For example, linear locusbased encodings for graph problems mostly map each vertex to a fixed corresponding position (locus) on the linear string =-=[11, 20, 18, 16, 21, 19]. The term-=- "geographical linkage" is somewhat vague; here we describe the term as it is used below. If two genes are more likely to move together in a crossover (from a parent to the offspring) than a... |

119 | Graph Bisection Algorithms with Good Average Case Behavior - Bui, Chaudhuri, et al. - 1987 |

73 |
Genetic placement
- Cohoon, Paris
- 1986
(Show Context)
Citation Context ...ature of the genetic algorithm (GA) approach, and most GA implementations have been based on linear encodings [13]. At the same time, linear encodings have also been viewed as limiting GA performance =-=[9, 2, 19]-=-. For example, with graph problems it is inevitable that considerable adjacency is lost when mapping a graph (e.g., a multi-dimensional mesh) onto a linear string [7]. Cohoon and Paris [9] proposed a ... |

69 | A Parallel Genetic Algorithm for the Set Partitioning Problem
- Levine
- 1994
(Show Context)
Citation Context ...locus-based encodings with this chromosome structure. For example, linear locusbased encodings for graph problems mostly map each vertex to a fixed corresponding position (locus) on the linear string =-=[11, 20, 18, 16, 21, 19]. The term-=- "geographical linkage" is somewhat vague; here we describe the term as it is used below. If two genes are more likely to move together in a crossover (from a parent to the offspring) than a... |

48 | Representing trees in genetic algorithms
- Palmer, Kershenbaum
- 1994
(Show Context)
Citation Context ...locus-based encodings with this chromosome structure. For example, linear locusbased encodings for graph problems mostly map each vertex to a fixed corresponding position (locus) on the linear string =-=[11, 20, 18, 16, 21, 19]. The term-=- "geographical linkage" is somewhat vague; here we describe the term as it is used below. If two genes are more likely to move together in a crossover (from a parent to the offspring) than a... |

18 |
Hyperplane synthesis for genetic algorithms
- Bui, Moon
- 1993
(Show Context)
Citation Context ... (%) Running time (%) Figure 8: Performances of Z3, block-uniform crossover, and 3-dimensional geographic crossover relative to traditional linear crossover performance when a linear encoding is used =-=[4, 5]-=-. We believe that this phenomenon will persist with high-dimensional encodings. Recently, Linial-LondonRabinovich [17] proved that we need a fairly highdimensional encoding to embed a graph with low d... |

14 |
On multidimensional encoding/crossover
- Bui, Moon
- 1995
(Show Context)
Citation Context ...as limiting GA performance [9, 2, 19]. For example, with graph problems it is inevitable that considerable adjacency is lost when mapping a graph (e.g., a multi-dimensional mesh) onto a linear string =-=[7]-=-. Cohoon and Paris [9] proposed a two-dimensional This work was supported by NSF YI award MIP9257982 and grant MIP-9223740. The authors are with the UCLA VLSI CAD Laboratory and the UCLA Commotion Lab... |

13 | Macro-cell and module placement by genetic adaptive search with bitmap-represented chromosome - Chan, Mazumder, et al. - 1991 |

11 |
A Hybrid Genetic Algorithm for Task Allocation in Multicomputers
- Mansour, Fox
- 1991
(Show Context)
Citation Context |

9 |
A twodimensional genetic algorithm for the ising problem
- Anderson, Jones, et al.
- 1991
(Show Context)
Citation Context ...ature of the genetic algorithm (GA) approach, and most GA implementations have been based on linear encodings [13]. At the same time, linear encodings have also been viewed as limiting GA performance =-=[9, 2, 19]-=-. For example, with graph problems it is inevitable that considerable adjacency is lost when mapping a graph (e.g., a multi-dimensional mesh) onto a linear string [7]. Cohoon and Paris [9] proposed a ... |

9 | Clique findingâ€” A genetic approach - Murthy, Parthasarathy, et al. - 1994 |

5 |
Vertex and edge partitions of graphs
- Jones
- 1992
(Show Context)
Citation Context ...raphs, random regular graphs with known small bisection widths, caterpillar graphs, and grid graphs. These have been extensively studied in the discrete algorithms and optimization literatures, e.g., =-=[14, 15, 6, 19]-=-. 2 All implementations in this paper used steady-state GAs with five cuts for crossover unless otherwise noted; we use the code of [6] except for the crossover part. 3 The GAs stop when 80% of the po... |

4 |
Graph partitioning and genetic algorithms
- Bui, Moon
- 1994
(Show Context)
Citation Context ...raphs, random regular graphs with known small bisection widths, caterpillar graphs, and grid graphs. These have been extensively studied in the discrete algorithms and optimization literatures, e.g., =-=[14, 15, 6, 19]-=-. 2 All implementations in this paper used steady-state GAs with five cuts for crossover unless otherwise noted; we use the code of [6] except for the crossover part. 3 The GAs stop when 80% of the po... |

3 |
Hybrid genetic algorithms with hyperplane synthesis: A theoretical and empirical study
- Moon
- 1994
(Show Context)
Citation Context ...ature of the genetic algorithm (GA) approach, and most GA implementations have been based on linear encodings [13]. At the same time, linear encodings have also been viewed as limiting GA performance =-=[9, 2, 19]-=-. For example, with graph problems it is inevitable that considerable adjacency is lost when mapping a graph (e.g., a multi-dimensional mesh) onto a linear string [7]. Cohoon and Paris [9] proposed a ... |