root / tmp / org.txm.analec.rcp / src / JamaPlus / SparseMatrix.java @ 2250
Historique | Voir | Annoter | Télécharger (7,7 ko)
| 1 |
/*
|
|---|---|
| 2 |
* To change this template, choose Tools | Templates
|
| 3 |
* and open the template in the editor.
|
| 4 |
*/
|
| 5 |
package JamaPlus; |
| 6 |
|
| 7 |
import JamaPlus.SparseSVD.SparseSVDException; |
| 8 |
import java.util.*; |
| 9 |
import java.util.logging.Level; |
| 10 |
import java.util.logging.Logger; |
| 11 |
|
| 12 |
/**
|
| 13 |
*
|
| 14 |
* @author Bernard
|
| 15 |
*/
|
| 16 |
public class SparseMatrix { |
| 17 |
private int m, n, nnz; |
| 18 |
private int[] colpts, rowinds; |
| 19 |
private double[] vals; |
| 20 |
private double[] tempRow; // stockage temporaire pour certains calculs (opb) |
| 21 |
|
| 22 |
public SparseMatrix(int m, int n) { // null matrix |
| 23 |
this.m = m;
|
| 24 |
this.n = n;
|
| 25 |
nnz = 0;
|
| 26 |
colpts = new int[n+1]; |
| 27 |
rowinds = new int[0]; |
| 28 |
vals = new double[0]; |
| 29 |
} |
| 30 |
|
| 31 |
private static class Triple implements Comparable<Triple> { |
| 32 |
int row;
|
| 33 |
int col;
|
| 34 |
double val;
|
| 35 |
|
| 36 |
private Triple(int row, int col, double val) { |
| 37 |
this.row = row;
|
| 38 |
this.col = col;
|
| 39 |
this.val = val;
|
| 40 |
} |
| 41 |
private Triple(int row, int col) { |
| 42 |
this.row = row;
|
| 43 |
this.col = col;
|
| 44 |
} |
| 45 |
public int compareTo(Triple t) { |
| 46 |
int res = col - t.col;
|
| 47 |
if(res==0) res = row - t.row; |
| 48 |
return res;
|
| 49 |
} |
| 50 |
} |
| 51 |
|
| 52 |
public SparseMatrix(int m, int n, ArrayList<Integer> rows, ArrayList<Integer> cols, |
| 53 |
ArrayList<Double> values) { |
| 54 |
this.m = m;
|
| 55 |
this.n = n;
|
| 56 |
nnz = rows.size(); |
| 57 |
if (nnz!=cols.size()||nnz!=rows.size())
|
| 58 |
throw new ArrayIndexOutOfBoundsException( |
| 59 |
"Size of lines, cols and values don't agree");
|
| 60 |
colpts = new int[n+1]; |
| 61 |
rowinds = new int[nnz]; |
| 62 |
vals = new double[nnz]; |
| 63 |
Triple[] triples = new Triple[nnz]; |
| 64 |
for(int i=0; i<nnz; i++) |
| 65 |
triples[i] = new Triple(rows.get(i), cols.get(i), values.get(i));
|
| 66 |
Arrays.sort(triples);
|
| 67 |
colpts[0] = 0; |
| 68 |
int c = 0; |
| 69 |
for(int i=0; i<nnz; i++) { |
| 70 |
if(i>0 && triples[i].row == triples[i-1].row && triples[i].col == triples[i-1].col) |
| 71 |
throw new ArrayIndexOutOfBoundsException( |
| 72 |
"Two values with same row and column numbers");
|
| 73 |
while(c<triples[i].col) colpts[++c] = i;
|
| 74 |
rowinds[i] = triples[i].row; |
| 75 |
vals[i] = triples[i].val; |
| 76 |
} |
| 77 |
while(c<n) colpts[++c] = nnz;
|
| 78 |
} |
| 79 |
|
| 80 |
public SparseMatrix(int m, int n, ArrayList<Integer> rows, ArrayList<Integer> cols, |
| 81 |
double value) {
|
| 82 |
this.m = m;
|
| 83 |
this.n = n;
|
| 84 |
nnz = rows.size(); |
| 85 |
if (nnz!=cols.size()||nnz!=rows.size())
|
| 86 |
throw new ArrayIndexOutOfBoundsException( |
| 87 |
"Size of lines, cols and values don't agree");
|
| 88 |
colpts = new int[n+1]; |
| 89 |
rowinds = new int[nnz]; |
| 90 |
vals = new double[nnz]; |
| 91 |
Triple[] triples = new Triple[nnz]; |
| 92 |
for(int i=0; i<nnz; i++) |
| 93 |
triples[i] = new Triple(rows.get(i), cols.get(i));
|
| 94 |
Arrays.sort(triples);
|
| 95 |
colpts[0] = 0; |
| 96 |
int c = 0; |
| 97 |
for(int i=0; i<nnz; i++) { |
| 98 |
if(i>0 && triples[i].row == triples[i-1].row && triples[i].col == triples[i-1].col) |
| 99 |
throw new ArrayIndexOutOfBoundsException( |
| 100 |
"Two values with same row and column numbers");
|
| 101 |
while(c<triples[i].col) colpts[++c] = i;
|
| 102 |
rowinds[i] = triples[i].row; |
| 103 |
vals[i] = value; |
| 104 |
} |
| 105 |
while(c<n) colpts[++c] = nnz;
|
| 106 |
} |
| 107 |
|
| 108 |
private SparseMatrix(int m, int n, int nnz, int[] colpts, int[] rowinds, double[] vals) { |
| 109 |
this.m = m;
|
| 110 |
this.n = n;
|
| 111 |
this.nnz = nnz;
|
| 112 |
this.colpts = colpts;
|
| 113 |
this.rowinds = rowinds;
|
| 114 |
this.vals = vals;
|
| 115 |
} |
| 116 |
|
| 117 |
public SparseMatrix(Matrix mat) { // convert a dense matrix |
| 118 |
m = mat.getRowDimension(); |
| 119 |
n = mat.getColumnDimension(); |
| 120 |
long nbval = mat.getNumberNonZeroValues();
|
| 121 |
if (nbval>Integer.MAX_VALUE) throw new ArrayIndexOutOfBoundsException( |
| 122 |
"Number of non-zero values exceeds 32 bits integer");
|
| 123 |
nnz = (int) nbval;
|
| 124 |
colpts = new int[n+1]; |
| 125 |
rowinds = new int[nnz]; |
| 126 |
vals = new double[nnz]; |
| 127 |
int nval = 0; |
| 128 |
for (int j = 0; j<n; j++) { |
| 129 |
colpts[j] = nval; |
| 130 |
for (int i = 0; i<m; i++) if (mat.get(i, j)!=0) { |
| 131 |
rowinds[nval] = i; |
| 132 |
vals[nval++] = mat.get(i, j); |
| 133 |
} |
| 134 |
} |
| 135 |
colpts[n] = nnz; |
| 136 |
} |
| 137 |
|
| 138 |
public SparseMatrix copy() {
|
| 139 |
int[] colpts1 = new int[n+1]; |
| 140 |
int[] rowinds1 = new int[nnz]; |
| 141 |
double[] vals1 = new double[nnz]; |
| 142 |
System.arraycopy(colpts, 0, colpts1, 0, n+1); |
| 143 |
System.arraycopy(rowinds, 0, rowinds1, 0, nnz); |
| 144 |
System.arraycopy(vals, 0, vals1, 0, nnz); |
| 145 |
return new SparseMatrix(m, n, nnz, colpts1, rowinds1, vals1); |
| 146 |
} |
| 147 |
|
| 148 |
public int getRowDimension() { |
| 149 |
return m;
|
| 150 |
} |
| 151 |
|
| 152 |
public int getColumnDimension() { |
| 153 |
return n;
|
| 154 |
} |
| 155 |
|
| 156 |
public double get(int i, int j) { |
| 157 |
try {
|
| 158 |
for (int nval = colpts[j]; nval<colpts[j+1]; nval++) |
| 159 |
if (rowinds[nval]==i) return vals[nval]; |
| 160 |
return 0; |
| 161 |
} catch (ArrayIndexOutOfBoundsException ex) { |
| 162 |
throw new ArrayIndexOutOfBoundsException("Sparse matrix : " |
| 163 |
+ex.getLocalizedMessage()); |
| 164 |
} |
| 165 |
} |
| 166 |
|
| 167 |
public SparseMatrix transpose() {
|
| 168 |
int[] colpts1 = new int[m+1]; |
| 169 |
int[] rowinds1 = new int[nnz]; |
| 170 |
double[] vals1 = new double[nnz]; |
| 171 |
for (int v = 0; v<nnz; v++) colpts1[rowinds[v]]++; // nnz in each row |
| 172 |
colpts1[m] = nnz-colpts1[m-1];
|
| 173 |
for (int r = m-1; r>0; r--) |
| 174 |
colpts1[r] = colpts1[r+1]-colpts1[r-1]; // starting point of the previous row |
| 175 |
colpts1[0] = 0; |
| 176 |
int i = 0; |
| 177 |
for (int c = 0; c<n; c++) while (i<colpts[c+1]) { |
| 178 |
int j = colpts1[rowinds[i]+1]++; // j : index of the value in the new vectors |
| 179 |
rowinds1[j] = c; |
| 180 |
vals1[j] = vals[i++]; |
| 181 |
} |
| 182 |
return new SparseMatrix(n, m, nnz, colpts1, rowinds1, vals1); |
| 183 |
} |
| 184 |
|
| 185 |
public void nulColumnSuppression() { |
| 186 |
ArrayList<Integer> colpts1 = new ArrayList<Integer>(); |
| 187 |
colpts1.add(0);
|
| 188 |
for (int i = 0; i<n; i++) if (colpts[i+1]>colpts[i]) |
| 189 |
colpts1.add(colpts[i+1]);
|
| 190 |
n = colpts1.size()-1;
|
| 191 |
colpts = new int[n+1]; |
| 192 |
for (int i = 0; i<=n; i++) colpts[i] = colpts1.get(i); |
| 193 |
} |
| 194 |
|
| 195 |
public double elementSum() { |
| 196 |
double s = 0; |
| 197 |
for (int i = 0; i<nnz; i++) s += vals[i]; |
| 198 |
|
| 199 |
return s;
|
| 200 |
} |
| 201 |
|
| 202 |
public Matrix lineSum() {
|
| 203 |
double[] S = new double[m]; |
| 204 |
for (int i = 0; i<nnz; i++) S[rowinds[i]] += vals[i]; |
| 205 |
|
| 206 |
return new Matrix(S, m); |
| 207 |
} |
| 208 |
|
| 209 |
public Matrix columnSum() {
|
| 210 |
double[] S = new double[n]; |
| 211 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 212 |
S[i] += vals[j]; |
| 213 |
|
| 214 |
return new Matrix(S, 1); |
| 215 |
} |
| 216 |
|
| 217 |
public SparseMatrix timesEquals(double s) { |
| 218 |
for (int i = 0; i<nnz; i++) { |
| 219 |
vals[i] *= s; |
| 220 |
} |
| 221 |
return this; |
| 222 |
} |
| 223 |
|
| 224 |
public SparseMatrix firstColumnTimesEqual(Matrix B) {
|
| 225 |
double[][] A = B.getArray(); |
| 226 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 227 |
vals[j] *= A[rowinds[j]][0];
|
| 228 |
return this; |
| 229 |
} |
| 230 |
|
| 231 |
public SparseMatrix firstLineTimesEqual(Matrix B) {
|
| 232 |
double[][] A = B.getArray(); |
| 233 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 234 |
vals[j] *= A[0][i];
|
| 235 |
return this; |
| 236 |
} |
| 237 |
|
| 238 |
public void opa(double x[], double y[]) { |
| 239 |
// Computes y = Ax (x et y : length n)
|
| 240 |
Arrays.fill(y, 0); |
| 241 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 242 |
y[rowinds[j]] += vals[j]*x[i]; |
| 243 |
} |
| 244 |
|
| 245 |
public void opb(double x[], double y[]) { |
| 246 |
// Computes y = A'Ax (x et y : length n)
|
| 247 |
if (tempRow==null) tempRow = new double[m]; |
| 248 |
else Arrays.fill(tempRow, 0); |
| 249 |
Arrays.fill(y, 0); |
| 250 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 251 |
tempRow[rowinds[j]] += vals[j]*x[i]; |
| 252 |
for (int i = 0; i<n; i++) for (int j = colpts[i]; j<colpts[i+1]; j++) |
| 253 |
y[i] += vals[j]*tempRow[rowinds[j]]; |
| 254 |
} |
| 255 |
|
| 256 |
public SparseSVD svd(int dim) throws SparseSVD.SparseSVDException { |
| 257 |
return new SparseSVD(this, dim); |
| 258 |
} |
| 259 |
|
| 260 |
/** Analyse factorielle de correspondances
|
| 261 |
@return Analyse factorielle de correspondances
|
| 262 |
@see Analyse factorielle de correspondances
|
| 263 |
*/
|
| 264 |
public SparseAFC afc() {
|
| 265 |
try {
|
| 266 |
return new SparseAFC(this); |
| 267 |
} catch (SparseSVDException ex) {
|
| 268 |
return null; |
| 269 |
} |
| 270 |
} |
| 271 |
} |