[DM-dev] thoughts on modelling genetic data

[stephan b. <st...@s1...>]

Hi again!

i just got main.cpp building under toc and was poking around the code a 
bit...

The Config type contains a variable (or container) for all of the 
various genes. As part of a different project i wrote up a class 
template for modelling genes in a generic manner, plus a container 
(Genome) to hold series of genes in. Attached are a couple of files 
which demonstrate how it works. It's based around two types, 
Gene<ValueType> and Genome<GeneType,KeyType>, and provides a consistent 
API for writing "breeding functors." These functors are used to breed 
genes using arbitrary algorithms/heuristics.

If the genetic data is moved to such a model, classes like Config don't 
have to be updated for new gene types. The Genome type acts like a map, 
so to add new types you simply refer to them in your client code, 
thereby automatically adding them. The genetic data is also 
serializable using the s11n framework.

A short demo of how to use it:

typedef gene::Gene<int> Gene;
typedef gene::Genome<Gene> Genome;

Genome him;
him["geekness"] = 5;
him["taste_in_clothes"] = -2;

Genome her;
her["geekness"] = 0;
her["taste_in_clothes"] = 7;

Genome them = gene::breed_genomes( him, her, gene::breed_average_f() );

The results in this case would look like:
geekness=2
taste_in_closes=2

but arbitrary breeder functors can be used to breed genes in arbitrary 
ways.

-- 
----- st...@s1...   http://s11n.net
"...pleasure is a grace and is not obedient to the commands
of the will." -- Alan W. Watts