Circuitry Routing Using a Genetic Algorithm-3

ALife Home

1 Introduction

2 Automata

3 Genetic Algorithms

4 Swarm Automata

5 Neural Networks

6 Experiments

7 People

8 Timeline

9 References

This Genetic Algorithm geverned population does converge but not as quickly as the previous ones, because a better state-transition model of the Ant automata is used.

The Ant chromosome consists of a linear string, read 8 bits at a time.
Each 8 bit field contains...
- 2 bits for movement in 4 horizontal directions.
- 1 bit to signal whether to move or not.
- 1 bit to signal whether to attempt to switch circuit board layers.
- 1 bit signaling whether to turn towards the Ant's target node.

So at every time step in the simulation, the Ant looks at its proximity values, and expresses the current allele of its genotype.
Learning is not a factor in this GA because the solution must be constant and non-ambiguous.
The GA Router enforces rules such as no path overlapping of traces, co-operation between Ants with the same electrical net ID.

See Genetic Algorithm Circuitry Router. Sun Java Applet for other versions.

© 2000 F. Michael O'Brien V.2000.12.03
Last Updated: Ottawa, Canada

Genetic Algorithm Router Applet Genetic Algorithm Router Applet GA Router Applet Version 3 Fast GA Router Applet Version 3 Fast Artificial Life Home Artificial Life Home