public static void gaussj(MatDoub a)
{
// Overloaded version with no right-hand sides. Replaces a by its
// inverse.
MatDoub b = new MatDoub(a.nrows(), 0); // Dummy vector with zero columns.
gaussj(a, b);
}
public static void printmatrix(MatDoub a)
{
for (int i = 0; i < a.nrows(); i++)
{
for (int j = 0; j < a.ncols(); j++)
{
System.Console.Write(a[i][j] + ",");
}
System.Console.WriteLine();
}
System.Console.WriteLine();
}
public static void fillMat(MatDoub a, Double[,] b)
{
int rows = a.nrows();
for (int i = 0; i < rows; i++)
{
for (int j = 0; j < rows; j++)
{
a[i][j] = b[i, j];
}
}
}
/// <summary>
/// Constructor. The single argument is A. The SVD computation is done by
/// decompose(), and the results are sorted by reorder().
/// </summary>
/// <param name="a">The matrix to decompose.</param>
public SVD(MatDoub a)
{
m = (a.nrows());
n = (a.ncols());
u = new MatDoub(a);
v = new MatDoub(n, n);
w = new VecDoub(n);
eps = double.Epsilon;
decompose();
reorder();
tsh = 0.5 * Math.Sqrt(m + n + 1.0) * w[0] * eps; // Default threshold for nonzero singular values;
}
public Sprsin(MatDoub a)//TODO: make a VecULong type to further save on space.
{
/************************************
* Converts a square matrix a[1..n][1..n] into row-indexed sparse storage mode.
* Only elements of a with magnitude ≥thresh are retained. Output is in two linear arrays
* with dimension nmax (an input parameter): sa[1..] contains array values, indexed by ija[1..].
* The number of elements filled of sa and ija on output are both ija[ija[1]-1]-1
************************************/
sa = new VecDoub(new double[nmax]);
ija = new VecLong(18, new long[nmax], 0);
n = a.nrows();
int i, j, k;
for (j = 0; j < n; j++)
{
sa[j] = a[j][j]; //Store the diagonal elements.
}
ija[0] = n + 1; //Index to the first off diagonal element if any.
k = n;
for (i = 0; i < n; i++)
{
for (j = 0; j < n; j++)
{
if (Math.Abs(a[i][j]) >= thresh && i != j)
{
if (++k > (int)nmax)
{
throw new Exception("sprsin: nmax too small");
}
sa[k] = a[i][j]; //Store off diagonal elements.
ija[k] = j; //And their column indices.
}
}
ija[i + 1] = k + 1; //As each row is completed, store index to next
}
this.numelements = k + 1;
}
/// <summary>
/// Solve m sets of n equations AX=B using the pseudoinverse of A.
/// The right-hand sides are input as b[0..n-1][0..m-1], while
/// x[0..n-1][0..m-1] return the solutions.
/// If positive, thresh is the threshold value below which singular values
/// as considered to be zero. If thresh is negative, a default based on rounodd error is used.
/// </summary>
/// <param name="b"></param>
/// <param name="x"></param>
/// <param name="thresh"></param>
//void solve(MatDoub_I &b, MatDoub_O &x, double thresh);
public void solve(MatDoub b, MatDoub x, double thresh)
{
int i, j, m = b.ncols();
if (b.nrows() != n || x.nrows() != n || b.ncols() != x.ncols())
{
throw new Exception("SVD::solve bad sizes");
}
VecDoub xx = new VecDoub(n);
for (j = 0; j < m; j++)
{
for (i = 0; i < n; i++)
{
xx[i] = b[i][j];
}
solve(xx, xx, thresh);
for (i = 0; i < n; i++)
{
x[i][j] = xx[i];
}
}
}
private bool sing; // Indicates whether A is singular.
public QRdcmp(MatDoub a)
{
n = a.nrows();
MatDoub qt = new MatDoub(n, n);
MatDoub r = new MatDoub(a);
sing = (false);
int i, j, k;
VecDoub c = new VecDoub(n);
VecDoub d = new VecDoub(n);
double scale, sigma, sum, tau;
for (k = 0; k < n - 1; k++)
{
scale = 0.0;
for (i = k; i < n; i++)
{
scale = Math.Max(scale, Math.Abs(r[i][k]));
}
if (scale == 0.0)
{ // Singular case.
sing = true;
c[k] = d[k] = 0.0;
}
else
{ // Form Qk and Qk A.
for (i = k; i < n; i++)
{
r[i][k] /= scale;
}
for (sum = 0.0, i = k; i < n; i++)
{
sum += r[i][k] * r[i][k];
}
sigma = NR.SIGN(Math.Sqrt(sum), r[k][k]);
r[k][k] += sigma;
c[k] = sigma * r[k][k];
d[k] = -scale * sigma;
for (j = k + 1; j < n; j++)
{
for (sum = 0.0, i = k; i < n; i++)
{
sum += r[i][k] * r[i][j];
}
tau = sum / c[k];
for (i = k; i < n; i++)
{
r[i][j] -= tau * r[i][k];
}
}
}
}
d[n - 1] = r[n - 1][n - 1];
if (d[n - 1] == 0.0)
{
sing = true;
}
///////////////////////////
for (i = 0; i < n; i++)
{ // Form QT explicitly.
for (j = 0; j < n; j++)
{
qt[i][j] = 0.0;
}
qt[i][i] = 1.0;
}
for (k = 0; k < n - 1; k++)
{
if (c[k] != 0.0)
{
for (j = 0; j < n; j++)
{
sum = 0.0;
for (i = k; i < n; i++)
{
sum += r[i][k] * qt[i][j];
}
sum /= c[k];
for (i = k; i < n; i++)
{
qt[i][j] -= sum * r[i][k];
}
}
}
}
for (i = 0; i < n; i++)
{ // Form R explicitly.
r[i][i] = d[i];
for (j = 0; j < i; j++)
{
r[i][j] = 0.0;
}
}
}
public static void gaussj(MatDoub a, MatDoub b)
{
// Linear equation solution by Gauss-Jordan elimination,
// equation (2.1.1) above. The input matrix is a[0..n-1][0..n-1].
// b[0..n-1][0..m-1] is input containing the m right-hand side vectors.
// On output, a is replaced by its matrix inverse, and b is replaced by
// the corresponding set of solution vectors.
int i, icol = 0, irow = 0, j, k, l, ll, n = a.nrows(), m = a.ncols();
double big, dum, pivinv;
VecInt indxc = new VecInt(n);
VecInt indxr = new VecInt(n);
VecInt ipiv = new VecInt(n); // These integer arrays are used
// for bookkeeping on the pivoting.
for (j = 0; j < n; j++)
{
ipiv[j] = 0;
}
for (i = 0; i < n; i++) // This is the main loop over the columns to
// be
{
big = 0.0; // reduced.
for (j = 0; j < n; j++)
{
// This is the outer loop of the search for a pivot over the entire matrix!
if (ipiv[j] != 1) // element.
{
for (k = 0; k < n; k++)
{
if (ipiv[k] == 0)
{
if (Math.Abs(a[j][k]) >= big)
{
big = Math.Abs(a[j][k]);
irow = j;
icol = k;
}
}
}
}
}
++(ipiv[icol]);
// We now have the pivot element, so we interchange rows, if needed,
// to put the pivot element on the diagonal. The columns are not
// physically interchanged, only relabeled: indxc[i], the column of the .iC1/th
// pivot element, is the .iC1/th column that is reduced, while indxr[i] is
// the row in which that pivot element was originally located.
// If indxr[i] indxc[i], there is an implied column interchange.
// With this form of bookkeeping, the solution bs will end up in the
// correct order, and the inverse matrix will be scrambled by columns.
if (irow != icol)
{
for (l = 0; l < n; l++)
{
NR.SWAP(a, irow, l, icol, l);
}
for (l = 0; l < m; l++)
{
NR.SWAP(b, irow, l, icol, l);
}
}
indxr[i] = irow; // We are now ready to divide the pivot row by the
// pivot element, located at irow and icol.
indxc[i] = icol;
if (a[icol][icol] == 0.0)
{
throw new Exception("gaussj: Singular Matrix");
}
pivinv = 1.0 / a[icol][icol];
a[icol][icol] = 1.0;
for (l = 0; l < n; l++)
{
a[icol][l] *= pivinv;
}
for (l = 0; l < m; l++)
{
b[icol][l] *= pivinv;
}
for (ll = 0; ll < n; ll++)
{
// Next, we reduce the rows...
if (ll != icol)
{ // ...except for the pivot one, of course.
dum = a[ll][icol];
a[ll][icol] = 0.0;
for (l = 0; l < n; l++)
{
a[ll][l] -= a[icol][l] * dum;
}
for (l = 0; l < m; l++)
{
b[ll][l] -= b[icol][l] * dum;
}
}
}
}
// This is the end of the main loop over columns of the reduction. It
// only remains to unscramble the solution in view of the column
// interchanges. We do this by interchanging pairs of columns in the
// reverse order that the permutation was built up.
for (l = n - 1; l >= 0; l--)
{
if (indxr[l] != indxc[l])
{
for (k = 0; k < n; k++)
{
NR.SWAP(a, k, indxr[l], k, indxc[l]);
}
}
}
}