[C#] Linear Programming with Revised Simplex Method for Prof

using System;

namespace Revised_Simplex
{
public class profit
{
public static void revisedSimplex(bool maximize, int n, int m, double[,] a, double epsilon, int[] basicvar)
{
int i, j, k, m2, p, idx = 0;
double[] objcoeff = new double[n + 1];
double[] varsum = new double[n + 1];
double[] optbasicval = new double[m + 3];
double[] aux = new double[m + 3];
double[,] work = new double[m + 3, m + 3];
double part, sum;
bool infeasible, unbound, abort, outres, iterate;
if (maximize)
for (j = 1; j <= n; j++)
a[0, j] = -a[0, j];
infeasible = false;
unbound = false;
m2 = m + 2;
p = m + 2;
outres = true;
k = m + 1;
for (j = 1; j <= n; j++)
{
objcoeff[j] = a[0, j];
sum = 0.0;
for (i = 1; i <= m; i++)
sum -= a[i, j];
varsum[j] = sum;
}
sum = 0.0;
for (i = 1; i <= m; i++)
{
basicvar[i] = n + i;
optbasicval[i] = a[i, 0];
sum -= a[i, 0];
}
optbasicval[k] = 0.0;
optbasicval[m2] = sum;
for (i = 1; i <= m2; i++)
{
for (j = 1; j <= m2; j++)
work[i, j] = 0.0;
work[i, i] = 1.0;
}
iterate = true;
do
{
if ((optbasicval[m2] >= -epsilon) && outres)
{
outres = false;
p = m + 1;
}
part = 0.0;
for (j = 1; j <= n; j++)
{
sum = work[p,m+1] * objcoeff[j] + work[p,m+2] * varsum[j];
for (i = 1; i <= m; i++)
sum += work[p, i] * a[i, j];
if (part > sum)
{
part = sum;
k = j;
}
}
if (part > -epsilon)
{
iterate = false;
if (outres)
infeasible = true;
else
a[0, 0] = -optbasicval[p];
}
else
{
for (i = 1; i <= p; i++)
{
sum = work[i,m+1] * objcoeff[k] + work[i,m+2] * varsum[k];
for (j = 1; j <= m; j++)
sum += work[i, j] * a[j, k];
aux[i] = sum;
}
abort = true;
for (i = 1; i <= m; i++)
if (aux[i] >= epsilon)
{
sum = optbasicval[i] / aux[i];
if (abort || (sum < part))
{
part = sum;
idx = i;
}
abort = false;
}
if (abort)
{
unbound = true;
iterate = false;
}
else
{
basicvar[idx] = k;
sum = 1.0 / aux[idx];
for (j = 1; j <= m; j++)
work[idx, j] *= sum;
i = ((idx == 1) ? 2 : 1);
do
{
sum = aux[i];
optbasicval[i] -= part * sum;
for (j = 1; j <= m; j++)
work[i, j] -= work[idx, j] * sum;
i += ((i == idx - 1) ? 2 : 1);
} while (i <= p);
optbasicval[idx] = part;
}
}
} while (iterate);
// return results
basicvar[m + 1] = (infeasible ? 1 : 0);
basicvar[m + 2] = (unbound ? 1 : 0);
for (i = 1; i <= m; i++)
a[i, 0] = optbasicval[i];
if (maximize)
{
for (j = 1; j <= n; j++)
a[0, j] = -a[0, j];
a[0, 0] = -a[0, 0];
}
}


static public void Main ()
{
int n = 5;
int m = 3;
double eps = 1.0e-5;
int[] basicvar = new int[m + 3];

double[,] a = { { 0, .42, .35, 0, 0, 0},
{100, 6, 0, 1, 0, 0},
{ 60, 2, 7, 0, 1, 0},
{ 90, 2, 3, 0, 0, 1}};

revisedSimplex(true, n, m, a, eps, basicvar);
Console.WriteLine("Suche optimale Mischung für maximalen Gewinn\n");

if (basicvar[m + 1] > 0)
Console.WriteLine("Keine Lösung gefunden.");
else
{
if (basicvar[m + 2] > 0)
Console.WriteLine("Keine Lösung gefunden.");
else
{
Console.WriteLine("Optimale Lösung gefunden.\nWerte von CHEWY und NUTTY und Slack-Variablen");
for (int i = 1; i <= m; i++)
Console.WriteLine("{0,10}\t {1,-10}", basicvar[i], a[i, 0]);
Console.WriteLine("Optimaler Wert der Funktion = {0}\n", a[0, 0]);
}
}
}
}
}