public object Clone()
{
var newObj = new Gene(this.body.Length);
newObj.points = points;
newObj.keyValList = keyValList.ToList();
newObj.body = (int[]) body.Clone();
return newObj;
}
private void SimpleMutation(List<Gene> result, Gene gene)
{
var count = (int)(_random.NextDouble() * 10) + 1;
for (int n = 0; n < count; n++)
{
int pos = (int)(_random.NextDouble() * (gene.body.Length - 2)) + 1;
gene.body[pos] = 1 - gene.body[pos]; // zmiana 0 na 1 i odwrotnie
}
gene.points = genePointsValue(gene);
result.Add(gene);
}
private void OneHalfCrossPoint(List<Gene> result, Gene first, Gene second, int firstBodyLen, int secondBodyLen)
{
int crossPoint = (int)(_random.NextDouble() * (secondBodyLen - 1) + 1);
Gene firstCrossedGene = new Gene(firstBodyLen + (secondBodyLen - crossPoint));
Gene secondCrossedGene = new Gene(crossPoint + 1);
Array.Copy(first.body, firstCrossedGene.body, 0);
Array.Copy(second.body, crossPoint, firstCrossedGene.body, firstBodyLen, (secondBodyLen - crossPoint));
Array.Copy(second.body, secondCrossedGene.body, crossPoint + 1);
secondCrossedGene.body[crossPoint] = 1;
firstCrossedGene.points = genePointsValue(firstCrossedGene);
result.Add(firstCrossedGene);
secondCrossedGene.points = genePointsValue(secondCrossedGene);
result.Add(secondCrossedGene);
}
private void NoCutsChangeMutation(List<Gene> result, Gene gene)
{
int first_pos = (int)(_random.NextDouble() * (gene.body.Length - 2)) + 1;
int second_pos = (int)(_random.NextDouble() * (gene.body.Length - 2)) + 1;
gene.body[first_pos] = 1 - gene.body[first_pos]; // zmiana 0 na 1 i odwrotnie
gene.body[second_pos] = 1 - gene.body[second_pos]; // zmiana 0 na 1 i odwrotnie
gene.points = genePointsValue(gene);
result.Add(gene);
}
private void JoinCrossing(List<Gene> result, Gene first, Gene second, int firstBodyLen, int secondBodyLen)
{
var lenghtVal = firstBodyLen + secondBodyLen - 1;
Gene crossedGene = new Gene(lenghtVal);
first.body.CopyTo(crossedGene.body, 0);
second.body.CopyTo(crossedGene.body, first.body.Length - 1); // nadpisujemy ostatnia jedynke - ciecie
crossedGene.points = genePointsValue(crossedGene);
result.Add(crossedGene);
}
private void IncrementCutsMutation(List<Gene> result, Gene gene, int gene_cuts)
{
if (gene_cuts < gene.body.Length - 2) // jesli istnieje przynajmniej jedno "0"
{
for (var ni = 0; ni < try_n_times; ni++) // probuj kilka razy
{
int pos = (int)(_random.NextDouble() * (gene.body.Length - 2)) + 1;
if (gene.body[pos] == 0)
{
gene.body[pos] = 1;
gene.points = genePointsValue(gene);
break;
}
}
}
result.Add(gene);
}
private List<Gene> generatePopulation2()
{
int pos1, pos2;
for (int i = 0; i < _populationSize; i++)
{
int ok = 0;
if (segments.Count > 1)
{
do
{
pos1 = (int)(_random.NextDouble() * segments.Count);
pos2 = (int)(_random.NextDouble() * segments.Count);
if (pos1 != pos2)
{
foreach (var el in segments)
{
if (el == segments[pos1] + segments[pos2])
{
ok = 1;
Gene gene = new Gene(el + 1);
gene.body[segments[pos1]] = 1;
gene.points = genePointsValue(gene);
startPop.Add(gene);
break;
}
}
}
else // wylosowano 2 razy ten sam element, losuj dalej
ok = 0;
} while (ok == 0);
}
else // nie losuj, jest tylko 1 element
{
Gene gene = new Gene(segments[0] + 1);
gene.points = genePointsValue(gene);
startPop.Add(gene);
}
}
return startPop;
}
private int genePointsValue(Gene gene_to_evaluate)
{
List<int> cuts = new List<int>();
List<int> gene_segments = new List<int>();
List<int> temp_gene_segments = new List<int>();
List<int> temp_segments = segments.Where(e => true).ToList();
for (int i = 0; i < gene_to_evaluate.body.Length; i++)
{
if (gene_to_evaluate.body[i] == 1)
{
cuts.Add(i);
}
}
int cutsCount = cuts.Count;
for (int i = 0; i < cutsCount - 1; i++)
{
for (int j = i + 1; j < cutsCount; j++)
{
gene_segments.Add(cuts[j] - cuts[i]);
temp_gene_segments.Add(cuts[j] - cuts[i]);
}
}
int points = 0;
int gs_count = gene_segments.Count;
// jesli prawidlowa sekwencja to temp_gene_segments powinno byc puste a temp_segments jak najmneij elementów (zostana tylko FP)
for (int i = 0; i < gs_count; i++)
{
if (temp_segments.Contains(gene_segments[i]))
{
temp_segments.Remove(gene_segments[i]);
points++;
temp_gene_segments.Remove(gene_segments[i]);
}
}
//maxCurrentK = Gene.calculateMaxCutsFromCurrentData(gs_count);
var gte_len = gene_to_evaluate.body.Length;
if (gte_len - 1 > maxSegment) // kara za przekroczenie dlugosci
points -= (int)((gte_len - maxSegment)* _random.NextDouble());
var tgs_count = temp_gene_segments.Count;
if (tgs_count != 0)
{
points -= (int)((tgs_count) * maxK);
}
else
gene_to_evaluate.isValid = true;
var gte_count = gene_to_evaluate.CountCuts();
if (gte_count > maxK)
points -= (gte_count - maxK);// +(int)(maxK * _random.NextDouble());
return points;
}
private void DecrementCutsMutation(List<Gene> result, Gene gene)
{
for (var ni = 0; ni < try_n_times; ni++) // probuje kilka razy
{
int pos = (int)(_random.NextDouble() * (gene.body.Length - 2)) + 1;
if (gene.body[pos] == 1)
{
gene.body[pos] = 0;
gene.points = genePointsValue(gene);
break;
}
}
result.Add(gene);
}
private void CrossMutation(List<Gene> result, Gene gene)
{
var gene_len = gene.body.Length;
Gene crossedGene = new Gene(gene_len + 1); // wiekszy o jeden
int pos = (int)(_random.NextDouble() * (gene_len - 2)) + 1;
Array.Copy(gene.body, pos, crossedGene.body, 1, (gene_len - pos));
Array.Copy(gene.body, 1, crossedGene.body, gene_len - pos + 1, pos - 1);
crossedGene.body[0] = 1;
crossedGene.body[gene_len] = 1;
crossedGene.points = genePointsValue(crossedGene);
result.Add(crossedGene);
}
private void backgroundWorker1_DoWork(object sender, DoWorkEventArgs e)
{
BackgroundWorker bgw = sender as BackgroundWorker;
// inicjalizacja populacji v1.0
/*for (int i = 0; i < _populationSize; i++)
{
int val = (int)(_random.NextDouble() * segments.Count);
Gene gene = new Gene(segments[val] + 1);
gene.points = genePointsValue(gene);
genes.Add(gene);
//Invoke(new Action(() => listView1.Items.Add(gene.PrintBody() + " " + gene.points.ToString())));
//listView2.Items.Add(genes[i].CountCuts().ToString());
}*/
// inicjalizacja populacji v2.0
genes = generatePopulation2();
for (int generation = 0; generation < maxGenerationCount; generation++)
{
List<Gene> GenerationGeneList = new List<Gene>();
int maxGenePoint = -10000;
for (int j = 0; j < _populationSize - _takeBestCount; j++)
{
List<Gene> result = new List<Gene>();
double val = _random.NextDouble();
//losujem 2 dowolne geny
int one = (int)(_random.NextDouble() * _populationSize);
int two = (int)(_random.NextDouble() * _populationSize);
Gene first = (Gene)genes[one].Clone();
Gene second = (Gene)genes[two].Clone();
int firstBodyLen = first.body.Length;
int secondBodyLen = second.body.Length;
if (val < _crossingRatio) // krzyzowanie
{
var crossingType = _random.NextDouble();
// jednopunktowe losowe
if (crossingType < 0.5 && firstBodyLen > 2 && secondBodyLen > 2)
OneCrossPoint(result, first, second, firstBodyLen, secondBodyLen);
// jednopunktowe polowkowe
else if (crossingType >= 0.6 && crossingType < 0.95 && firstBodyLen > 2 && secondBodyLen > 2)
OneHalfCrossPoint(result, first, second, firstBodyLen, secondBodyLen);
// polacz oba geny
else JoinCrossing(result, first, second, firstBodyLen, secondBodyLen);
}
// nie krzyzuj
else { }
int points1 = first.points;
int points2 = second.points;
if (val < _mutationRatio) //mutacja
{
Gene gene;
if (points1 > points2) gene = first;
else gene = second;
int gene_cuts = gene.CountCuts();
// za duzo ciec, zmniejszamy k przez mutacje
if (gene_cuts > maxK) DecrementCutsMutation(result, gene);
else if (gene_cuts == maxK)
{
double mut_type = _random.NextDouble();
// utrzymujemy ta sama liczbe ciec
if (mut_type < 0.33) NoCutsChangeMutation(result, gene);
// zmniejszamy liczbe ciec
else DecrementCutsMutation(result, gene);
}
else // za malo ciec, zwiekszamy k przez mutacje
{
double mut_type = _random.NextDouble();
if (mut_type < 0.5) IncrementCutsMutation(result, gene, gene_cuts);
else if (mut_type >= 0.5 && mut_type < 0.9) NoCutsChangeMutation(result, gene);//SimpleMutation(result, gene);
// mutuj poprzez zamiane
else CrossMutation(result, gene);
}
}
if (result.Count == 0) // nie bylo ani mutacji ani krzyzowania
{
if (points1 > points2)
{
Array.Reverse(first.body);
result.Add(first);
}
else
{
Array.Reverse(second.body);
result.Add(second);
}
}
// wybieramy najlepszy gen z rezultatow
int resCount = result.Count;
if (resCount > 1)
{
for (int n = 0; n < resCount; n++)
{
int points = result[n].points;
if (maxGenePoint < points && result[n].isValid == true) maxGenePoint = points;
}
result = result.OrderByDescending(item => item.points).ToList();
if (result[1].points >0) otherResults.Add(result[1]);
}
else
{
int points = result[0].points;
if (maxGenePoint < points) maxGenePoint = points;
}
// najlepszy wynik dodajemy do glownej listy
GenerationGeneList.Add(result[0]);
}
genes = genes.OrderByDescending(item => item.points).ToList();
for (int j = 0; j < _takeBestCount; j++)
{
GenerationGeneList.Add(genes[j]); // dodanie 3 najlepszych z poprzedniego pokolenia
}
// utworzenie listy populacji dla nastepnego kroku
genes = GenerationGeneList;
genes = genes.OrderByDescending(item => item.points).ToList();
var bestGene = genes[0];
var bestVal = bestGene.points;
maxValues.Add(bestVal);
var restartVal = (int)(20000 / maxPointsFromData);
if (restartVal < 5) restartVal = 5;
int tmpTrackBarVal = 0;
if (bestGene.isValid == true)
{
bestValidGene = bestGene;
if (bestValidGenes.Count == 0)
{
bestValidGenes.Add(bestValidGene);
}
else if (bestValidGenes.Last().points < bestValidGene.points && !bestValidGenes.Any(item => item.body == bestValidGene.body))
{
bestValidGenes.Add(bestValidGene);
Invoke(new Action(() =>
resultsListView.Items.Add("P: "+ bestValidGene.points.ToString() + ", c: "
+ bestValidGene.CountCuts().ToString()
+ ", t: " + totalSecondLabelCount.Text + "[sek]")));
}
}
Invoke(new Action(() =>
{
tmpTrackBarVal = trackBar1.Value;
label2.Text = "Punkty: " + bestVal.ToString();
if (bestValidGene != null)
{
label1.Text = bestValidGenes.Last().PrintBody();
label6.Text = "Długość: " + (bestValidGenes.Last().body.Count() - 1).ToString();
label8.Text = "Zapis zbioru: " + bestValidGenes.Last().PrintBodyAsSet();
label10.Text = "Liczba ciec: " + bestValidGenes.Last().CountCuts().ToString();
}
label9.Text = "maxK: " + maxK.ToString();
label13.Text = "Restart po: " + restartVal.ToString();
toolStripProgressBar1.Increment(1);
refreshPlotGraph(maxValues);
}));
updateTime();
if ((generation % restartVal) == (restartVal - 1))
{
bool needChanges = checkIfNoChanges(restartVal - 1);
if (needChanges)
{
addToResultsNewGenes3(genes); ////////////////////////////////////////////////////////////////////////
}
}
if (maxGenePoint >= tmpTrackBarVal || bgw.CancellationPending)
{
if (bgw.CancellationPending)
e.Cancel = true;
Invoke(new Action(() => button1.Enabled = false));
break;
}
}
Invoke(new Action(() =>
{
if (bestValidGenes.Count > 0)
{
int bestVG = bestValidGenes.Last().CountCuts();
label17.Text = _optimalK.ToString();
label15.Text = bestVG.ToString();
label14.Text = Decimal.Round((decimal)bestVG / _optimalK * 100, 2).ToString() + " %";
}
else
{
label17.Text = _optimalK.ToString();
label15.Text = "Brak rozwiązania";
label14.Text = Decimal.Round((decimal)0 / _optimalK * 100, 2).ToString() + " %";
}
toolStripStatusLabel1.Text = "Obliczenia zakończono.";
toolStripProgressBar1.Value = 0;
beepPing();
}));
}
private void addToResultsNewGenes4(List<Gene> genes)
{
var g = bestValidGenes.Count - 1;
int count = 0;
if (g > -1)
{
Gene gene = (Gene)bestValidGenes[g].Clone();
int cuts = gene.CountCuts();
int size = gene.body.Length;
Array.Reverse(gene.body);
for (int i = 1; (i < size - 1) && (count < _populationSize); i++)
{
if (gene.body[i] == 1)
{
Gene tmpGene = new Gene(i+1);
Array.Copy(gene.body, 0, tmpGene.body, 0, i + 1);
tmpGene.points = genePointsValue(tmpGene);
genes[count++] = tmpGene;
tmpGene = new Gene(size - i);
Array.Copy(gene.body, i, tmpGene.body, 0, size - i);
tmpGene.points = genePointsValue(tmpGene);
genes[count++] = tmpGene;
}
}
}
}
