JavaGenes: Evolving Graphs

1
JavaGenes Evolving Graphs

• Al Globus, Veridian MRJ Technology Solutions,
  Inc.
• globus_at_nas.nasa.gov
• http//www.nas.nasa.gov/globus/home.html
• John Lawton, UCSC
• Todd Wipke, UCSC
• Sean Atsatt, Sierra Imaging,Inc.

2
Crossover
Graphs
Strings
Trees
abcd wxyz
abcd wxyz
abyz wxcd
3
Graph Crossover
Combine into a Child
Rip Two Parents Apart
4
Molecule Fitness Function

• All-pairs-shortest-path (APSP) distance
• Assign extended types to each atom (element,
  single bonds, double bonds, triple bonds)
• Find shortest bond path between each pair of
  atoms
• Create bag with one item per pair of atoms
• item (type1, type2, path length)
• bag set with repeated items
• Tanimoto distance intersection / union

5
Evolving Molecules
size median shortest failures()
benzene purine cubane diazepam cholesterol
morphine
100 100 100 200 200 200
3 51 30 NA NA NA
0 0 0 71 74 97
0 6 4 453 1414 9769
6
Circuits

• Directed graphs
• More node types (and, nand, or, nor, xor, nxor,
  plus initial 0 or 1 for each type)
• Exactly one input and one output node per graph
• Crossover generates unconnected circuits on very
  rare occasion
• Fitness function number of right answers on 100
  randomly generated inputs
• Serial logic (one bit input and output at each
  step)
• Logic simulator assumes all devices require unit
  time

7
Circuits
name size median
shortest failures()
8
Conclusions

• Evolving graphs using genetic operators is
  possible
• Molecules and circuits have been evolved
• Molecular evolution is reasonably successful
• Non-trivial circuits have yet to be evolved
• This is a new technique that will require
  substantial refinement
• open source in the near future
• Where self-assembly problems can be cast as
  graphs JavaGenes may be of some utility