public PopulationItem mutate(PopulationItem p)
{
var clone = new PopulationItem();
clone.placements = p.placements.ToArray().ToList();
clone.Rotation = p.Rotation.Clone() as float[];
for (int i = 0; i < clone.placements.Count(); i++)
{
var rand = r.NextDouble();
if (rand < 0.01 * Config.mutationRate)
{
var j = i + 1;
if (j < clone.placements.Count)
{
var temp = clone.placements[i];
clone.placements[i] = clone.placements[j];
clone.placements[j] = temp;
}
}
rand = r.NextDouble();
if (rand < 0.01 * Config.mutationRate)
{
clone.Rotation[i] = (float)Math.Floor(r.NextDouble() * Config.rotations) * (360f / Config.rotations);
}
}
return(clone);
}
// returns a random individual from the population, weighted to the front of the list (lower fitness value is more likely to be selected)
public PopulationItem randomWeightedIndividual(PopulationItem exclude = null)
{
//var pop = this.population.slice(0);
var pop = this.population.ToArray();
if (exclude != null && Array.IndexOf(pop, exclude) >= 0)
{
pop.splice(Array.IndexOf(pop, exclude), 1);
}
var rand = r.NextDouble();
float lower = 0;
var weight = 1 / (float)pop.Length;
float upper = weight;
for (var i = 0; i < pop.Length; i++)
{
// if the random number falls between lower and upper bounds, select this individual
if (rand > lower && rand < upper)
{
return(pop[i]);
}
lower = upper;
upper += 2 * weight * ((pop.Length - i) / (float)pop.Length);
}
return(pop[0]);
}
// single point crossover
public PopulationItem[] mate(PopulationItem male, PopulationItem female)
{
var cutpoint = (int)Math.Round(Math.Min(Math.Max(r.NextDouble(), 0.1), 0.9) * (male.placements.Count - 1));
var gene1 = new List <NFP>(male.placements.Take(cutpoint).ToArray());
var rot1 = new List <float>(male.Rotation.Take(cutpoint).ToArray());
var gene2 = new List <NFP>(female.placements.Take(cutpoint).ToArray());
var rot2 = new List <float>(female.Rotation.Take(cutpoint).ToArray());
int i = 0;
for (i = 0; i < female.placements.Count; i++)
{
if (!gene1.Any(z => z.id == female.placements[i].id))
{
gene1.Add(female.placements[i]);
rot1.Add(female.Rotation[i]);
}
}
for (i = 0; i < male.placements.Count; i++)
{
if (!gene2.Any(z => z.id == male.placements[i].id))
{
gene2.Add(male.placements[i]);
rot2.Add(male.Rotation[i]);
}
}
return(new[] { new PopulationItem()
{
placements = gene1, Rotation = rot1.ToArray()
},
new PopulationItem()
{
placements = gene2, Rotation = rot2.ToArray()
} });
}
public void launchWorkers(NestItem[] parts)
{
background.ResponseAction = ResponseProcessor;
if (ga == null)
{
List <NFP> adam = new List <NFP>();
var id = 0;
for (int i = 0; i < parts.Count(); i++)
{
if (!parts[i].IsSheet)
{
for (int j = 0; j < parts[i].Quanity; j++)
{
var poly = cloneTree(parts[i].Polygon); // deep copy
poly.id = id; // id is the unique id of all parts that will be nested, including cloned duplicates
poly.source = i; // source is the id of each unique part from the main part list
adam.Add(poly);
id++;
}
}
}
adam = adam.OrderByDescending(z => Math.Abs(GeometryUtil.polygonArea(z))).ToList();
/*List<NFP> shuffle = new List<NFP>();
* Random r = new Random(DateTime.Now.Millisecond);
* while (adam.Any())
* {
* var rr = r.Next(adam.Count);
* shuffle.Add(adam[rr]);
* adam.RemoveAt(rr);
* }
* adam = shuffle;*/
/*#region special case
* var temp = adam[1];
* adam.RemoveAt(1);
* adam.Insert(9, temp);
*
#endregion*/
ga = new GeneticAlgorithm(adam.ToArray(), Config);
}
individual = null;
// check if current generation is finished
var finished = true;
for (int i = 0; i < ga.population.Count; i++)
{
if (ga.population[i].fitness == null)
{
finished = false;
break;
}
}
if (finished)
{
//console.log('new generation!');
// all individuals have been evaluated, start next generation
ga.generation();
}
var running = ga.population.Where((p) =>
{
return(p.processing != null);
}).Count();
List <NFP> sheets = new List <NFP>();
List <int> sheetids = new List <int>();
List <int> sheetsources = new List <int>();
List <List <NFP> > sheetchildren = new List <List <NFP> >();
var sid = 0;
for (int i = 0; i < parts.Count(); i++)
{
if (parts[i].IsSheet)
{
var poly = parts[i].Polygon;
for (int j = 0; j < parts[i].Quanity; j++)
{
var cln = cloneTree(poly);
cln.id = sid; // id is the unique id of all parts that will be nested, including cloned duplicates
cln.source = poly.source; // source is the id of each unique part from the main part list
sheets.Add(cln);
sheetids.Add(sid);
sheetsources.Add(i);
sheetchildren.Add(poly.children);
sid++;
}
}
}
for (int i = 0; i < ga.population.Count; i++)
{
//if(running < config.threads && !GA.population[i].processing && !GA.population[i].fitness){
// only one background window now...
if (running < 1 && ga.population[i].processing == null && ga.population[i].fitness == null)
{
ga.population[i].processing = true;
// hash values on arrays don't make it across ipc, store them in an array and reassemble on the other side....
List <int> ids = new List <int>();
List <int> sources = new List <int>();
List <List <NFP> > children = new List <List <NFP> >();
for (int j = 0; j < ga.population[i].placements.Count; j++)
{
var id = ga.population[i].placements[j].id;
var source = ga.population[i].placements[j].source;
var child = ga.population[i].placements[j].children;
//ids[j] = id;
ids.Add(id);
//sources[j] = source;
sources.Add(source.Value);
//children[j] = child;
children.Add(child);
}
DataInfo data = new DataInfo()
{
index = i,
sheets = sheets,
sheetids = sheetids.ToArray(),
sheetsources = sheetsources.ToArray(),
sheetchildren = sheetchildren,
individual = ga.population[i],
config = Config,
ids = ids.ToArray(),
sources = sources.ToArray(),
children = children
};
background.BackgroundStart(data);
//ipcRenderer.send('background-start', { index: i, sheets: sheets, sheetids: sheetids, sheetsources: sheetsources, sheetchildren: sheetchildren, individual: GA.population[i], config: config, ids: ids, sources: sources, children: children});
running++;
}
}
}