root / tmp / org.txm.analec.rcp / src / JamaPlus / Matrix.java @ 2099
Historique | Voir | Annoter | Télécharger (33,16 ko)
1 |
package JamaPlus; |
---|---|
2 |
|
3 |
import java.text.NumberFormat; |
4 |
import java.text.DecimalFormat; |
5 |
import java.text.DecimalFormatSymbols; |
6 |
import java.util.Locale; |
7 |
import java.io.PrintWriter; |
8 |
import java.io.BufferedReader; |
9 |
import java.io.StreamTokenizer; |
10 |
import JamaPlus.util.*; |
11 |
import java.util.ArrayList; |
12 |
|
13 |
/**
|
14 |
Jama = Java Matrix class.
|
15 |
<P>
|
16 |
The Java Matrix Class provides the fundamental operations of numerical
|
17 |
linear algebra. Various constructors create Matrices from two dimensional
|
18 |
arrays of double precision floating point numbers. Various "gets" and
|
19 |
"sets" provide access to submatrices and matrix elements. Several methods
|
20 |
implement basic matrix arithmetic, including matrix addition and
|
21 |
multiplication, matrix norms, and element-by-element array operations.
|
22 |
Methods for reading and printing matrices are also included. All the
|
23 |
operations in this version of the Matrix Class involve real matrices.
|
24 |
Complex matrices may be handled in a future version.
|
25 |
<P>
|
26 |
Five fundamental matrix decompositions, which consist of pairs or triples
|
27 |
of matrices, permutation vectors, and the like, produce results in five
|
28 |
decomposition classes. These decompositions are accessed by the Matrix
|
29 |
class to compute solutions of simultaneous linear equations, determinants,
|
30 |
inverses and other matrix functions. The five decompositions are:
|
31 |
<P><UL>
|
32 |
<LI>Cholesky Decomposition of symmetric, positive definite matrices.
|
33 |
<LI>LU Decomposition of rectangular matrices.
|
34 |
<LI>QR Decomposition of rectangular matrices.
|
35 |
<LI>Singular Value Decomposition of rectangular matrices.
|
36 |
<LI>Eigenvalue Decomposition of both symmetric and nonsymmetric square matrices.
|
37 |
</UL>
|
38 |
<DL>
|
39 |
<DT><B>Example of use:</B></DT>
|
40 |
<P>
|
41 |
<DD>Solve a linear system A x = b and compute the residual norm, ||b - A x||.
|
42 |
<P><PRE>
|
43 |
double[][] vals = {{1.,2.,3},{4.,5.,6.},{7.,8.,10.}};
|
44 |
Matrix A = new Matrix(vals);
|
45 |
Matrix b = Matrix.random(3,1);
|
46 |
Matrix x = A.solve(b);
|
47 |
Matrix r = A.times(x).minus(b);
|
48 |
double rnorm = r.normInf();
|
49 |
</PRE></DD>
|
50 |
</DL>
|
51 |
|
52 |
@author The MathWorks, Inc. and the National Institute of Standards and Technology.
|
53 |
@version 5 August 1998
|
54 |
*/
|
55 |
public class Matrix implements Cloneable, java.io.Serializable { |
56 |
|
57 |
/* ------------------------
|
58 |
Class variables
|
59 |
* ------------------------ */
|
60 |
/** Array for internal storage of elements.
|
61 |
@serial internal array storage.
|
62 |
*/
|
63 |
private double[][] A; |
64 |
/** Row and column dimensions.
|
65 |
@serial row dimension.
|
66 |
@serial column dimension.
|
67 |
*/
|
68 |
private int m, n; |
69 |
|
70 |
/* ------------------------
|
71 |
Constructors
|
72 |
* ------------------------ */
|
73 |
/** Construct an m-by-n matrix of zeros.
|
74 |
@param m Number of rows.
|
75 |
@param n Number of colums.
|
76 |
*/
|
77 |
public Matrix(int m, int n) { |
78 |
this.m = m;
|
79 |
this.n = n;
|
80 |
A = new double[m][n]; |
81 |
} |
82 |
|
83 |
/** Construct an m-by-n constant matrix.
|
84 |
@param m Number of rows.
|
85 |
@param n Number of colums.
|
86 |
@param s Fill the matrix with this scalar value.
|
87 |
*/
|
88 |
public Matrix(int m, int n, double s) { |
89 |
this.m = m;
|
90 |
this.n = n;
|
91 |
A = new double[m][n]; |
92 |
for (int i = 0; i < m; i++) { |
93 |
for (int j = 0; j < n; j++) { |
94 |
A[i][j] = s; |
95 |
} |
96 |
} |
97 |
} |
98 |
|
99 |
/** Construct a matrix from a 2-D array.
|
100 |
@param A Two-dimensional array of doubles.
|
101 |
@exception IllegalArgumentException All rows must have the same length
|
102 |
@see #constructWithCopy
|
103 |
*/
|
104 |
public Matrix(double[][] A) { |
105 |
m = A.length; |
106 |
n = A[0].length;
|
107 |
for (int i = 0; i < m; i++) { |
108 |
if (A[i].length != n) {
|
109 |
throw new IllegalArgumentException("All rows must have the same length."); |
110 |
} |
111 |
} |
112 |
this.A = A;
|
113 |
} |
114 |
|
115 |
/** Construct a matrix quickly without checking arguments.
|
116 |
@param A Two-dimensional array of doubles.
|
117 |
@param m Number of rows.
|
118 |
@param n Number of colums.
|
119 |
*/
|
120 |
public Matrix(double[][] A, int m, int n) { |
121 |
this.A = A;
|
122 |
this.m = m;
|
123 |
this.n = n;
|
124 |
} |
125 |
|
126 |
/** Construct a matrix from a one-dimensional packed array
|
127 |
@param vals One-dimensional array of doubles, packed by columns (ala Fortran).
|
128 |
@param m Number of rows.
|
129 |
@exception IllegalArgumentException Array length must be a multiple of m.
|
130 |
*/
|
131 |
public Matrix(double vals[], int m) { |
132 |
this.m = m;
|
133 |
n = (m != 0 ? vals.length / m : 0); |
134 |
if (m * n != vals.length) {
|
135 |
throw new IllegalArgumentException("Array length must be a multiple of m."); |
136 |
} |
137 |
A = new double[m][n]; |
138 |
for (int i = 0; i < m; i++) { |
139 |
for (int j = 0; j < n; j++) { |
140 |
A[i][j] = vals[i + j * m]; |
141 |
} |
142 |
} |
143 |
} |
144 |
|
145 |
/* ------------------------
|
146 |
Public Methods
|
147 |
* ------------------------ */
|
148 |
/** Construct a matrix from a copy of a 2-D array.
|
149 |
@param A Two-dimensional array of doubles.
|
150 |
@exception IllegalArgumentException All rows must have the same length
|
151 |
*/
|
152 |
public static Matrix constructWithCopy(double[][] A) { |
153 |
int m = A.length;
|
154 |
int n = A[0].length; |
155 |
Matrix X = new Matrix(m, n);
|
156 |
double[][] C = X.getArray(); |
157 |
for (int i = 0; i < m; i++) { |
158 |
if (A[i].length != n) {
|
159 |
throw new IllegalArgumentException("All rows must have the same length."); |
160 |
} |
161 |
for (int j = 0; j < n; j++) { |
162 |
C[i][j] = A[i][j]; |
163 |
} |
164 |
} |
165 |
return X;
|
166 |
} |
167 |
|
168 |
/** Make a deep copy of a matrix
|
169 |
*/
|
170 |
public Matrix copy() {
|
171 |
Matrix X = new Matrix(m, n);
|
172 |
double[][] C = X.getArray(); |
173 |
for (int i = 0; i < m; i++) { |
174 |
System.arraycopy(A[i], 0, C[i], 0, n); |
175 |
} |
176 |
return X;
|
177 |
} |
178 |
|
179 |
/** Clone the Matrix object.
|
180 |
*/
|
181 |
public Object clone() { |
182 |
return this.copy(); |
183 |
} |
184 |
|
185 |
/** Access the internal two-dimensional array.
|
186 |
@return Pointer to the two-dimensional array of matrix elements.
|
187 |
*/
|
188 |
public double[][] getArray() { |
189 |
return A;
|
190 |
} |
191 |
|
192 |
/** Copy the internal two-dimensional array.
|
193 |
@return Two-dimensional array copy of matrix elements.
|
194 |
*/
|
195 |
public double[][] getArrayCopy() { |
196 |
double[][] C = new double[m][n]; |
197 |
for (int i = 0; i < m; i++) { |
198 |
System.arraycopy(A[i], 0, C[i], 0, n); |
199 |
} |
200 |
return C;
|
201 |
} |
202 |
|
203 |
/** Make a one-dimensional column packed copy of the internal array.
|
204 |
@return Matrix elements packed in a one-dimensional array by columns.
|
205 |
*/
|
206 |
public double[] getColumnPackedCopy() { |
207 |
double[] vals = new double[m * n]; |
208 |
for (int i = 0; i < m; i++) { |
209 |
for (int j = 0; j < n; j++) { |
210 |
vals[i + j * m] = A[i][j]; |
211 |
} |
212 |
} |
213 |
return vals;
|
214 |
} |
215 |
|
216 |
/** Make a one-dimensional row packed copy of the internal array.
|
217 |
@return Matrix elements packed in a one-dimensional array by rows.
|
218 |
*/
|
219 |
public double[] getRowPackedCopy() { |
220 |
double[] vals = new double[m * n]; |
221 |
for (int i = 0; i < m; i++) { |
222 |
for (int j = 0; j < n; j++) { |
223 |
vals[i * n + j] = A[i][j]; |
224 |
} |
225 |
} |
226 |
return vals;
|
227 |
} |
228 |
|
229 |
/** Get row dimension.
|
230 |
@return m, the number of rows.
|
231 |
*/
|
232 |
public int getRowDimension() { |
233 |
return m;
|
234 |
} |
235 |
|
236 |
/** Get column dimension.
|
237 |
@return n, the number of columns.
|
238 |
*/
|
239 |
public int getColumnDimension() { |
240 |
return n;
|
241 |
} |
242 |
|
243 |
/** Get a single element.
|
244 |
@param i Row index.
|
245 |
@param j Column index.
|
246 |
@return A(i,j)
|
247 |
@exception ArrayIndexOutOfBoundsException
|
248 |
*/
|
249 |
public double get(int i, int j) { |
250 |
return A[i][j];
|
251 |
} |
252 |
|
253 |
/** Get a submatrix.
|
254 |
@param i0 Initial row index
|
255 |
@param i1 Final row index
|
256 |
@param j0 Initial column index
|
257 |
@param j1 Final column index
|
258 |
@return A(i0:i1,j0:j1)
|
259 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
260 |
*/
|
261 |
public Matrix getMatrix(int i0, int i1, int j0, int j1) { |
262 |
Matrix X = new Matrix(i1 - i0 + 1, j1 - j0 + 1); |
263 |
double[][] B = X.getArray(); |
264 |
try {
|
265 |
for (int i = i0; i <= i1; i++) { |
266 |
for (int j = j0; j <= j1; j++) { |
267 |
B[i - i0][j - j0] = A[i][j]; |
268 |
} |
269 |
} |
270 |
} catch (ArrayIndexOutOfBoundsException e) { |
271 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
272 |
} |
273 |
return X;
|
274 |
} |
275 |
|
276 |
/** Get a submatrix.
|
277 |
@param r Array of row indices.
|
278 |
@param c Array of column indices.
|
279 |
@return A(r(:),c(:))
|
280 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
281 |
*/
|
282 |
public Matrix getMatrix(int[] r, int[] c) { |
283 |
Matrix X = new Matrix(r.length, c.length);
|
284 |
double[][] B = X.getArray(); |
285 |
try {
|
286 |
for (int i = 0; i < r.length; i++) { |
287 |
for (int j = 0; j < c.length; j++) { |
288 |
B[i][j] = A[r[i]][c[j]]; |
289 |
} |
290 |
} |
291 |
} catch (ArrayIndexOutOfBoundsException e) { |
292 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
293 |
} |
294 |
return X;
|
295 |
} |
296 |
|
297 |
/** Get a submatrix.
|
298 |
@param i0 Initial row index
|
299 |
@param i1 Final row index
|
300 |
@param c Array of column indices.
|
301 |
@return A(i0:i1,c(:))
|
302 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
303 |
*/
|
304 |
public Matrix getMatrix(int i0, int i1, int[] c) { |
305 |
Matrix X = new Matrix(i1 - i0 + 1, c.length); |
306 |
double[][] B = X.getArray(); |
307 |
try {
|
308 |
for (int i = i0; i <= i1; i++) { |
309 |
for (int j = 0; j < c.length; j++) { |
310 |
B[i - i0][j] = A[i][c[j]]; |
311 |
} |
312 |
} |
313 |
} catch (ArrayIndexOutOfBoundsException e) { |
314 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
315 |
} |
316 |
return X;
|
317 |
} |
318 |
|
319 |
/** Get a submatrix.
|
320 |
@param r Array of row indices.
|
321 |
@param i0 Initial column index
|
322 |
@param i1 Final column index
|
323 |
@return A(r(:),j0:j1)
|
324 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
325 |
*/
|
326 |
public Matrix getMatrix(int[] r, int j0, int j1) { |
327 |
Matrix X = new Matrix(r.length, j1 - j0 + 1); |
328 |
double[][] B = X.getArray(); |
329 |
try {
|
330 |
for (int i = 0; i < r.length; i++) { |
331 |
for (int j = j0; j <= j1; j++) { |
332 |
B[i][j - j0] = A[r[i]][j]; |
333 |
} |
334 |
} |
335 |
} catch (ArrayIndexOutOfBoundsException e) { |
336 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
337 |
} |
338 |
return X;
|
339 |
} |
340 |
|
341 |
/** Set a single element.
|
342 |
@param i Row index.
|
343 |
@param j Column index.
|
344 |
@param s A(i,j).
|
345 |
@exception ArrayIndexOutOfBoundsException
|
346 |
*/
|
347 |
public void set(int i, int j, double s) { |
348 |
A[i][j] = s; |
349 |
} |
350 |
|
351 |
/** Set a submatrix.
|
352 |
@param i0 Initial row index
|
353 |
@param i1 Final row index
|
354 |
@param j0 Initial column index
|
355 |
@param j1 Final column index
|
356 |
@param X A(i0:i1,j0:j1)
|
357 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
358 |
*/
|
359 |
public void setMatrix(int i0, int i1, int j0, int j1, Matrix X) { |
360 |
try {
|
361 |
for (int i = i0; i <= i1; i++) { |
362 |
for (int j = j0; j <= j1; j++) { |
363 |
A[i][j] = X.get(i - i0, j - j0); |
364 |
} |
365 |
} |
366 |
} catch (ArrayIndexOutOfBoundsException e) { |
367 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
368 |
} |
369 |
} |
370 |
|
371 |
/** Set a submatrix.
|
372 |
@param r Array of row indices.
|
373 |
@param c Array of column indices.
|
374 |
@param X A(r(:),c(:))
|
375 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
376 |
*/
|
377 |
public void setMatrix(int[] r, int[] c, Matrix X) { |
378 |
try {
|
379 |
for (int i = 0; i < r.length; i++) { |
380 |
for (int j = 0; j < c.length; j++) { |
381 |
A[r[i]][c[j]] = X.get(i, j); |
382 |
} |
383 |
} |
384 |
} catch (ArrayIndexOutOfBoundsException e) { |
385 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
386 |
} |
387 |
} |
388 |
|
389 |
/** Set a submatrix.
|
390 |
@param r Array of row indices.
|
391 |
@param j0 Initial column index
|
392 |
@param j1 Final column index
|
393 |
@param X A(r(:),j0:j1)
|
394 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
395 |
*/
|
396 |
public void setMatrix(int[] r, int j0, int j1, Matrix X) { |
397 |
try {
|
398 |
for (int i = 0; i < r.length; i++) { |
399 |
for (int j = j0; j <= j1; j++) { |
400 |
A[r[i]][j] = X.get(i, j - j0); |
401 |
} |
402 |
} |
403 |
} catch (ArrayIndexOutOfBoundsException e) { |
404 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
405 |
} |
406 |
} |
407 |
|
408 |
/** Set a submatrix.
|
409 |
@param i0 Initial row index
|
410 |
@param i1 Final row index
|
411 |
@param c Array of column indices.
|
412 |
@param X A(i0:i1,c(:))
|
413 |
@exception ArrayIndexOutOfBoundsException Submatrix indices
|
414 |
*/
|
415 |
public void setMatrix(int i0, int i1, int[] c, Matrix X) { |
416 |
try {
|
417 |
for (int i = i0; i <= i1; i++) { |
418 |
for (int j = 0; j < c.length; j++) { |
419 |
A[i][c[j]] = X.get(i - i0, j); |
420 |
} |
421 |
} |
422 |
} catch (ArrayIndexOutOfBoundsException e) { |
423 |
throw new ArrayIndexOutOfBoundsException("Submatrix indices"); |
424 |
} |
425 |
} |
426 |
|
427 |
/** Matrix transpose.
|
428 |
@return A'
|
429 |
*/
|
430 |
public Matrix transpose() {
|
431 |
Matrix X = new Matrix(n, m);
|
432 |
double[][] C = X.getArray(); |
433 |
for (int i = 0; i < m; i++) { |
434 |
for (int j = 0; j < n; j++) { |
435 |
C[j][i] = A[i][j]; |
436 |
} |
437 |
} |
438 |
return X;
|
439 |
} |
440 |
|
441 |
/** One norm
|
442 |
@return maximum column sum.
|
443 |
*/
|
444 |
public double norm1() { |
445 |
double f = 0; |
446 |
for (int j = 0; j < n; j++) { |
447 |
double s = 0; |
448 |
for (int i = 0; i < m; i++) { |
449 |
s += Math.abs(A[i][j]);
|
450 |
} |
451 |
f = Math.max(f, s);
|
452 |
} |
453 |
return f;
|
454 |
} |
455 |
|
456 |
/** Two norm
|
457 |
@return maximum singular value.
|
458 |
*/
|
459 |
public double norm2() { |
460 |
return (new SingularValueDecomposition(this).norm2()); |
461 |
} |
462 |
|
463 |
/** Infinity norm
|
464 |
@return maximum row sum.
|
465 |
*/
|
466 |
public double normInf() { |
467 |
double f = 0; |
468 |
for (int i = 0; i < m; i++) { |
469 |
double s = 0; |
470 |
for (int j = 0; j < n; j++) { |
471 |
s += Math.abs(A[i][j]);
|
472 |
} |
473 |
f = Math.max(f, s);
|
474 |
} |
475 |
return f;
|
476 |
} |
477 |
|
478 |
/** Frobenius norm
|
479 |
@return sqrt of sum of squares of all elements.
|
480 |
*/
|
481 |
public double normF() { |
482 |
double f = 0; |
483 |
for (int i = 0; i < m; i++) { |
484 |
for (int j = 0; j < n; j++) { |
485 |
f = Math.hypot(f, A[i][j]);
|
486 |
} |
487 |
} |
488 |
return f;
|
489 |
} |
490 |
|
491 |
/** Unary minus
|
492 |
@return -A
|
493 |
*/
|
494 |
public Matrix uminus() {
|
495 |
Matrix X = new Matrix(m, n);
|
496 |
double[][] C = X.getArray(); |
497 |
for (int i = 0; i < m; i++) { |
498 |
for (int j = 0; j < n; j++) { |
499 |
C[i][j] = -A[i][j]; |
500 |
} |
501 |
} |
502 |
return X;
|
503 |
} |
504 |
|
505 |
/** C = A + B
|
506 |
@param B another matrix
|
507 |
@return A + B
|
508 |
*/
|
509 |
public Matrix plus(Matrix B) {
|
510 |
checkMatrixDimensions(B); |
511 |
Matrix X = new Matrix(m, n);
|
512 |
double[][] C = X.getArray(); |
513 |
for (int i = 0; i < m; i++) { |
514 |
for (int j = 0; j < n; j++) { |
515 |
C[i][j] = A[i][j] + B.A[i][j]; |
516 |
} |
517 |
} |
518 |
return X;
|
519 |
} |
520 |
|
521 |
/** A = A + B
|
522 |
@param B another matrix
|
523 |
@return A + B
|
524 |
*/
|
525 |
public Matrix plusEquals(Matrix B) {
|
526 |
checkMatrixDimensions(B); |
527 |
for (int i = 0; i < m; i++) { |
528 |
for (int j = 0; j < n; j++) { |
529 |
A[i][j] = A[i][j] + B.A[i][j]; |
530 |
} |
531 |
} |
532 |
return this; |
533 |
} |
534 |
|
535 |
/** C = A - B
|
536 |
@param B another matrix
|
537 |
@return A - B
|
538 |
*/
|
539 |
public Matrix minus(Matrix B) {
|
540 |
checkMatrixDimensions(B); |
541 |
Matrix X = new Matrix(m, n);
|
542 |
double[][] C = X.getArray(); |
543 |
for (int i = 0; i < m; i++) { |
544 |
for (int j = 0; j < n; j++) { |
545 |
C[i][j] = A[i][j] - B.A[i][j]; |
546 |
} |
547 |
} |
548 |
return X;
|
549 |
} |
550 |
|
551 |
/** A = A - B
|
552 |
@param B another matrix
|
553 |
@return A - B
|
554 |
*/
|
555 |
public Matrix minusEquals(Matrix B) {
|
556 |
checkMatrixDimensions(B); |
557 |
for (int i = 0; i < m; i++) { |
558 |
for (int j = 0; j < n; j++) { |
559 |
A[i][j] = A[i][j] - B.A[i][j]; |
560 |
} |
561 |
} |
562 |
return this; |
563 |
} |
564 |
|
565 |
/** Element-by-element multiplication, C = A.*B
|
566 |
@param B another matrix
|
567 |
@return A.*B
|
568 |
*/
|
569 |
public Matrix arrayTimes(Matrix B) {
|
570 |
checkMatrixDimensions(B); |
571 |
Matrix X = new Matrix(m, n);
|
572 |
double[][] C = X.getArray(); |
573 |
for (int i = 0; i < m; i++) { |
574 |
for (int j = 0; j < n; j++) { |
575 |
C[i][j] = A[i][j] * B.A[i][j]; |
576 |
} |
577 |
} |
578 |
return X;
|
579 |
} |
580 |
|
581 |
/** Element-by-element multiplication in place, A = A.*B
|
582 |
@param B another matrix
|
583 |
@return A.*B
|
584 |
*/
|
585 |
public Matrix arrayTimesEquals(Matrix B) {
|
586 |
checkMatrixDimensions(B); |
587 |
for (int i = 0; i < m; i++) { |
588 |
for (int j = 0; j < n; j++) { |
589 |
A[i][j] = A[i][j] * B.A[i][j]; |
590 |
} |
591 |
} |
592 |
return this; |
593 |
} |
594 |
|
595 |
/** Element-by-element right division, C = A./B
|
596 |
@param B another matrix
|
597 |
@return A./B
|
598 |
*/
|
599 |
public Matrix arrayRightDivide(Matrix B) {
|
600 |
checkMatrixDimensions(B); |
601 |
Matrix X = new Matrix(m, n);
|
602 |
double[][] C = X.getArray(); |
603 |
for (int i = 0; i < m; i++) { |
604 |
for (int j = 0; j < n; j++) { |
605 |
C[i][j] = A[i][j] / B.A[i][j]; |
606 |
} |
607 |
} |
608 |
return X;
|
609 |
} |
610 |
|
611 |
/** Element-by-element right division in place, A = A./B
|
612 |
@param B another matrix
|
613 |
@return A./B
|
614 |
*/
|
615 |
public Matrix arrayRightDivideEquals(Matrix B) {
|
616 |
checkMatrixDimensions(B); |
617 |
for (int i = 0; i < m; i++) { |
618 |
for (int j = 0; j < n; j++) { |
619 |
A[i][j] = A[i][j] / B.A[i][j]; |
620 |
} |
621 |
} |
622 |
return this; |
623 |
} |
624 |
|
625 |
/** Element-by-element left division, C = A.\B
|
626 |
@param B another matrix
|
627 |
@return A.\B
|
628 |
*/
|
629 |
public Matrix arrayLeftDivide(Matrix B) {
|
630 |
checkMatrixDimensions(B); |
631 |
Matrix X = new Matrix(m, n);
|
632 |
double[][] C = X.getArray(); |
633 |
for (int i = 0; i < m; i++) { |
634 |
for (int j = 0; j < n; j++) { |
635 |
C[i][j] = B.A[i][j] / A[i][j]; |
636 |
} |
637 |
} |
638 |
return X;
|
639 |
} |
640 |
|
641 |
/** Element-by-element left division in place, A = A.\B
|
642 |
@param B another matrix
|
643 |
@return A.\B
|
644 |
*/
|
645 |
public Matrix arrayLeftDivideEquals(Matrix B) {
|
646 |
checkMatrixDimensions(B); |
647 |
for (int i = 0; i < m; i++) { |
648 |
for (int j = 0; j < n; j++) { |
649 |
A[i][j] = B.A[i][j] / A[i][j]; |
650 |
} |
651 |
} |
652 |
return this; |
653 |
} |
654 |
|
655 |
/** Multiply a matrix by a scalar, C = s*A
|
656 |
@param s scalar
|
657 |
@return s*A
|
658 |
*/
|
659 |
public Matrix times(double s) { |
660 |
Matrix X = new Matrix(m, n);
|
661 |
double[][] C = X.getArray(); |
662 |
for (int i = 0; i < m; i++) { |
663 |
for (int j = 0; j < n; j++) { |
664 |
C[i][j] = s * A[i][j]; |
665 |
} |
666 |
} |
667 |
return X;
|
668 |
} |
669 |
|
670 |
/** Multiply a matrix by a scalar in place, A = s*A
|
671 |
@param s scalar
|
672 |
@return replace A by s*A
|
673 |
*/
|
674 |
public Matrix timesEquals(double s) { |
675 |
for (int i = 0; i < m; i++) { |
676 |
for (int j = 0; j < n; j++) { |
677 |
A[i][j] = s * A[i][j]; |
678 |
} |
679 |
} |
680 |
return this; |
681 |
} |
682 |
|
683 |
/** Linear algebraic matrix multiplication, A * B
|
684 |
@param B another matrix
|
685 |
@return Matrix product, A * B
|
686 |
@exception IllegalArgumentException Matrix inner dimensions must agree.
|
687 |
*/
|
688 |
public Matrix times(Matrix B) {
|
689 |
if (B.m != n) {
|
690 |
throw new IllegalArgumentException("Matrix inner dimensions must agree."); |
691 |
} |
692 |
Matrix X = new Matrix(m, B.n);
|
693 |
double[][] C = X.getArray(); |
694 |
double[] Bcolj = new double[n]; |
695 |
for (int j = 0; j < B.n; j++) { |
696 |
for (int k = 0; k < n; k++) { |
697 |
Bcolj[k] = B.A[k][j]; |
698 |
} |
699 |
for (int i = 0; i < m; i++) { |
700 |
double[] Arowi = A[i]; |
701 |
double s = 0; |
702 |
for (int k = 0; k < n; k++) { |
703 |
s += Arowi[k] * Bcolj[k]; |
704 |
} |
705 |
C[i][j] = s; |
706 |
} |
707 |
} |
708 |
return X;
|
709 |
} |
710 |
|
711 |
/** LU Decomposition
|
712 |
@return LUDecomposition
|
713 |
@see LUDecomposition
|
714 |
*/
|
715 |
public LUDecomposition lu() {
|
716 |
return new LUDecomposition(this); |
717 |
} |
718 |
|
719 |
/** QR Decomposition
|
720 |
@return QRDecomposition
|
721 |
@see QRDecomposition
|
722 |
*/
|
723 |
public QRDecomposition qr() {
|
724 |
return new QRDecomposition(this); |
725 |
} |
726 |
|
727 |
/** Cholesky Decomposition
|
728 |
@return CholeskyDecomposition
|
729 |
@see CholeskyDecomposition
|
730 |
*/
|
731 |
public CholeskyDecomposition chol() {
|
732 |
return new CholeskyDecomposition(this); |
733 |
} |
734 |
|
735 |
/** Singular Value Decomposition
|
736 |
@return SingularValueDecomposition
|
737 |
@see SingularValueDecomposition
|
738 |
*/
|
739 |
public SingularValueDecomposition svd() {
|
740 |
return new SingularValueDecomposition(this); |
741 |
} |
742 |
|
743 |
/** Eigenvalue Decomposition
|
744 |
@return EigenvalueDecomposition
|
745 |
@see EigenvalueDecomposition
|
746 |
*/
|
747 |
public EigenvalueDecomposition eig() {
|
748 |
return new EigenvalueDecomposition(this); |
749 |
} |
750 |
|
751 |
/** Solve A*X = B
|
752 |
@param B right hand side
|
753 |
@return solution if A is square, least squares solution otherwise
|
754 |
*/
|
755 |
public Matrix solve(Matrix B) {
|
756 |
return (m == n ? (new LUDecomposition(this)).solve(B) : (new QRDecomposition(this)).solve(B)); |
757 |
} |
758 |
|
759 |
/** Solve X*A = B, which is also A'*X' = B'
|
760 |
@param B right hand side
|
761 |
@return solution if A is square, least squares solution otherwise.
|
762 |
*/
|
763 |
public Matrix solveTranspose(Matrix B) {
|
764 |
return transpose().solve(B.transpose());
|
765 |
} |
766 |
|
767 |
/** Matrix inverse or pseudoinverse
|
768 |
@return inverse(A) if A is square, pseudoinverse otherwise.
|
769 |
*/
|
770 |
public Matrix inverse() {
|
771 |
return solve(identity(m, m));
|
772 |
} |
773 |
|
774 |
/** Matrix determinant
|
775 |
@return determinant
|
776 |
*/
|
777 |
public double det() { |
778 |
return new LUDecomposition(this).det(); |
779 |
} |
780 |
|
781 |
/** Matrix rank
|
782 |
@return effective numerical rank, obtained from SVD.
|
783 |
*/
|
784 |
public int rank() { |
785 |
return new SingularValueDecomposition(this).rank(); |
786 |
} |
787 |
|
788 |
/** Matrix condition (2 norm)
|
789 |
@return ratio of largest to smallest singular value.
|
790 |
*/
|
791 |
public double cond() { |
792 |
return new SingularValueDecomposition(this).cond(); |
793 |
} |
794 |
|
795 |
/** Matrix trace.
|
796 |
@return sum of the diagonal elements.
|
797 |
*/
|
798 |
public double trace() { |
799 |
double t = 0; |
800 |
for (int i = 0; i < Math.min(m, n); i++) { |
801 |
t += A[i][i]; |
802 |
} |
803 |
return t;
|
804 |
} |
805 |
|
806 |
/** Generate matrix with random elements
|
807 |
@param m Number of rows.
|
808 |
@param n Number of colums.
|
809 |
@return An m-by-n matrix with uniformly distributed random elements.
|
810 |
*/
|
811 |
public static Matrix random(int m, int n) { |
812 |
Matrix A = new Matrix(m, n);
|
813 |
double[][] X = A.getArray(); |
814 |
for (int i = 0; i < m; i++) { |
815 |
for (int j = 0; j < n; j++) { |
816 |
X[i][j] = Math.random();
|
817 |
} |
818 |
} |
819 |
return A;
|
820 |
} |
821 |
|
822 |
/** Generate identity matrix
|
823 |
@param m Number of rows.
|
824 |
@param n Number of colums.
|
825 |
@return An m-by-n matrix with ones on the diagonal and zeros elsewhere.
|
826 |
*/
|
827 |
public static Matrix identity(int m, int n) { |
828 |
Matrix A = new Matrix(m, n);
|
829 |
double[][] X = A.getArray(); |
830 |
for (int i = 0; i < m; i++) { |
831 |
for (int j = 0; j < n; j++) { |
832 |
X[i][j] = (i == j ? 1.0 : 0.0); |
833 |
} |
834 |
} |
835 |
return A;
|
836 |
} |
837 |
|
838 |
/** Print the matrix to stdout. Line the elements up in columns
|
839 |
* with a Fortran-like 'Fw.d' style format.
|
840 |
@param w Column width.
|
841 |
@param d Number of digits after the decimal.
|
842 |
*/
|
843 |
public void print(int w, int d) { |
844 |
print(new PrintWriter(System.out, true), w, d); |
845 |
} |
846 |
|
847 |
/** Print the matrix to the output stream. Line the elements up in
|
848 |
* columns with a Fortran-like 'Fw.d' style format.
|
849 |
@param output Output stream.
|
850 |
@param w Column width.
|
851 |
@param d Number of digits after the decimal.
|
852 |
*/
|
853 |
public void print(PrintWriter output, int w, int d) { |
854 |
DecimalFormat format = new DecimalFormat(); |
855 |
format.setDecimalFormatSymbols(new DecimalFormatSymbols(Locale.US)); |
856 |
format.setMinimumIntegerDigits(1);
|
857 |
format.setMaximumFractionDigits(d); |
858 |
format.setMinimumFractionDigits(d); |
859 |
format.setGroupingUsed(false);
|
860 |
print(output, format, w + 2);
|
861 |
} |
862 |
|
863 |
/** Print the matrix to stdout. Line the elements up in columns.
|
864 |
* Use the format object, and right justify within columns of width
|
865 |
* characters.
|
866 |
* Note that is the matrix is to be read back in, you probably will want
|
867 |
* to use a NumberFormat that is set to US Locale.
|
868 |
@param format A Formatting object for individual elements.
|
869 |
@param width Field width for each column.
|
870 |
@see java.text.DecimalFormat#setDecimalFormatSymbols
|
871 |
*/
|
872 |
public void print(NumberFormat format, int width) { |
873 |
print(new PrintWriter(System.out, true), format, width); |
874 |
} |
875 |
|
876 |
// DecimalFormat is a little disappointing coming from Fortran or C's printf.
|
877 |
// Since it doesn't pad on the left, the elements will come out different
|
878 |
// widths. Consequently, we'll pass the desired column width in as an
|
879 |
// argument and do the extra padding ourselves.
|
880 |
/** Print the matrix to the output stream. Line the elements up in columns.
|
881 |
* Use the format object, and right justify within columns of width
|
882 |
* characters.
|
883 |
* Note that is the matrix is to be read back in, you probably will want
|
884 |
* to use a NumberFormat that is set to US Locale.
|
885 |
@param output the output stream.
|
886 |
@param format A formatting object to format the matrix elements
|
887 |
@param width Column width.
|
888 |
@see java.text.DecimalFormat#setDecimalFormatSymbols
|
889 |
*/
|
890 |
public void print(PrintWriter output, NumberFormat format, int width) { |
891 |
output.println(); // start on new line.
|
892 |
for (int i = 0; i < m; i++) { |
893 |
for (int j = 0; j < n; j++) { |
894 |
String s = format.format(A[i][j]); // format the number |
895 |
int padding = Math.max(1, width - s.length()); // At _least_ 1 space |
896 |
for (int k = 0; k < padding; k++) { |
897 |
output.print(' ');
|
898 |
} |
899 |
output.print(s); |
900 |
} |
901 |
output.println(); |
902 |
} |
903 |
output.println(); // end with blank line.
|
904 |
} |
905 |
|
906 |
/** Read a matrix from a stream. The format is the same the print method,
|
907 |
* so printed matrices can be read back in (provided they were printed using
|
908 |
* US Locale). Elements are separated by
|
909 |
* whitespace, all the elements for each row appear on a single line,
|
910 |
* the last row is followed by a blank line.
|
911 |
@param input the input stream.
|
912 |
*/
|
913 |
@SuppressWarnings("unchecked") |
914 |
public static Matrix read(BufferedReader input) throws java.io.IOException { |
915 |
StreamTokenizer tokenizer = new StreamTokenizer(input); |
916 |
|
917 |
// Although StreamTokenizer will parse numbers, it doesn't recognize
|
918 |
// scientific notation (E or D); however, Double.valueOf does.
|
919 |
// The strategy here is to disable StreamTokenizer's number parsing.
|
920 |
// We'll only get whitespace delimited words, EOL's and EOF's.
|
921 |
// These words should all be numbers, for Double.valueOf to parse.
|
922 |
|
923 |
tokenizer.resetSyntax(); |
924 |
tokenizer.wordChars(0, 255); |
925 |
tokenizer.whitespaceChars(0, ' '); |
926 |
tokenizer.eolIsSignificant(true);
|
927 |
java.util.Vector v = new java.util.Vector();
|
928 |
|
929 |
// Ignore initial empty lines
|
930 |
while (tokenizer.nextToken() == StreamTokenizer.TT_EOL); |
931 |
if (tokenizer.ttype == StreamTokenizer.TT_EOF) { |
932 |
throw new java.io.IOException("Unexpected EOF on matrix read."); |
933 |
} |
934 |
do {
|
935 |
v.addElement(Double.valueOf(tokenizer.sval)); // Read & store 1st row. |
936 |
} while (tokenizer.nextToken() == StreamTokenizer.TT_WORD); |
937 |
|
938 |
int n = v.size(); // Now we've got the number of columns! |
939 |
double row[] = new double[n]; |
940 |
for (int j = 0; j < n; j++) // extract the elements of the 1st row. |
941 |
{ |
942 |
row[j] = ((Double) v.elementAt(j)).doubleValue();
|
943 |
} |
944 |
v.removeAllElements(); |
945 |
v.addElement(row); // Start storing rows instead of columns.
|
946 |
while (tokenizer.nextToken() == StreamTokenizer.TT_WORD) { |
947 |
// While non-empty lines
|
948 |
v.addElement(row = new double[n]); |
949 |
int j = 0; |
950 |
do {
|
951 |
if (j >= n) {
|
952 |
throw new java.io.IOException("Row " + v.size() + " is too long."); |
953 |
} |
954 |
row[j++] = Double.valueOf(tokenizer.sval).doubleValue();
|
955 |
} while (tokenizer.nextToken() == StreamTokenizer.TT_WORD); |
956 |
if (j < n) {
|
957 |
throw new java.io.IOException("Row " + v.size() + " is too short."); |
958 |
} |
959 |
} |
960 |
int m = v.size(); // Now we've got the number of rows. |
961 |
double[][] A = new double[m][]; |
962 |
v.copyInto(A); // copy the rows out of the vector
|
963 |
return new Matrix(A); |
964 |
} |
965 |
|
966 |
|
967 |
/* ------------------------
|
968 |
Private Methods
|
969 |
* ------------------------ */
|
970 |
/** Check if size(A) == size(B) **/
|
971 |
private void checkMatrixDimensions(Matrix B) { |
972 |
if (B.m != m || B.n != n) {
|
973 |
throw new IllegalArgumentException("Matrix dimensions must agree."); |
974 |
} |
975 |
} |
976 |
|
977 |
/* ---------------------------------
|
978 |
Quelques méthodes supplémentaires
|
979 |
* ---------------------------------*/
|
980 |
private void checkLineDimension(Matrix B) { |
981 |
if (B.m != m) {
|
982 |
throw new IllegalArgumentException("Matrix line dimension must agree."); |
983 |
} |
984 |
} |
985 |
|
986 |
private void checkColumnDimension(Matrix B) { |
987 |
if (B.n != n) {
|
988 |
throw new IllegalArgumentException("Matrix column dimension must agree."); |
989 |
} |
990 |
} |
991 |
|
992 |
public long getNumberNonZeroValues() { |
993 |
long nnz = 0; |
994 |
for (int i = 0; i < m; i++) { |
995 |
for (int j = 0; j < n; j++) { |
996 |
if (A[i][j] != 0) nnz++; |
997 |
} |
998 |
} |
999 |
return nnz;
|
1000 |
} |
1001 |
|
1002 |
public double minColumn(int j) { |
1003 |
double min = Double.POSITIVE_INFINITY; |
1004 |
for (int i = 0; i < m; i++) { |
1005 |
min = Math.min(min, A[i][j]);
|
1006 |
} |
1007 |
return min;
|
1008 |
} |
1009 |
|
1010 |
public double maxColumn(int j) { |
1011 |
double max = Double.NEGATIVE_INFINITY; |
1012 |
for (int i = 0; i < m; i++) { |
1013 |
max = Math.max(max, A[i][j]);
|
1014 |
} |
1015 |
return max;
|
1016 |
} |
1017 |
|
1018 |
public double elementSum() { |
1019 |
double s = 0; |
1020 |
for (int i = 0; i < m; i++) { |
1021 |
for (int j = 0; j < n; j++) { |
1022 |
s += A[i][j]; |
1023 |
} |
1024 |
} |
1025 |
return s;
|
1026 |
} |
1027 |
|
1028 |
public Matrix lineSum() {
|
1029 |
Matrix S = new Matrix(m, 1); |
1030 |
double s[][] = S.getArray(); |
1031 |
for (int i = 0; i < m; i++) { |
1032 |
for (int j = 0; j < n; j++) { |
1033 |
s[i][0] += A[i][j];
|
1034 |
} |
1035 |
} |
1036 |
return S;
|
1037 |
} |
1038 |
|
1039 |
public Matrix lineMean() {
|
1040 |
Matrix S = new Matrix(m, 1); |
1041 |
double s[][] = S.getArray(); |
1042 |
for (int i = 0; i < m; i++) { |
1043 |
for (int j = 0; j < n; j++) { |
1044 |
s[i][0] += A[i][j];
|
1045 |
} |
1046 |
s[i][0] /= n;
|
1047 |
} |
1048 |
return S;
|
1049 |
} |
1050 |
|
1051 |
public Matrix columnSum() {
|
1052 |
Matrix S = new Matrix(1, n); |
1053 |
double s[][] = S.getArray(); |
1054 |
for (int j = 0; j < n; j++) { |
1055 |
for (int i = 0; i < m; i++) { |
1056 |
s[0][j] += A[i][j];
|
1057 |
} |
1058 |
} |
1059 |
return S;
|
1060 |
} |
1061 |
|
1062 |
public Matrix columnMean() {
|
1063 |
Matrix S = new Matrix(1, n); |
1064 |
double s[][] = S.getArray(); |
1065 |
for (int j = 0; j < n; j++) { |
1066 |
for (int i = 0; i < m; i++) { |
1067 |
s[0][j] += A[i][j];
|
1068 |
} |
1069 |
s[0][j] /= m;
|
1070 |
} |
1071 |
return S;
|
1072 |
} |
1073 |
|
1074 |
public Matrix columnCovariance() {
|
1075 |
double mean[] = new double[n]; |
1076 |
for (int j = 0; j < n; j++) { |
1077 |
for (int i = 0; i < m; i++) { |
1078 |
mean[j] += A[i][j]; |
1079 |
} |
1080 |
mean[j] /= m; |
1081 |
} |
1082 |
Matrix X = new Matrix(n, n);
|
1083 |
double[][] B = X.getArray(); |
1084 |
double[] Acolj = new double[m]; |
1085 |
int coef = m > 2 ? m-1 : 1; |
1086 |
for (int j = 0; j < n; j++) { |
1087 |
for (int k = 0; k < m; k++) { |
1088 |
Acolj[k] = A[k][j] - mean[j]; |
1089 |
} |
1090 |
for (int i = 0; i < n; i++) { |
1091 |
double[] Atrowi = new double[m]; |
1092 |
for (int k = 0; k < m; k++) { |
1093 |
Atrowi[k] = A[k][i] - mean[i]; |
1094 |
} |
1095 |
double s = 0; |
1096 |
for (int k = 0; k < m; k++) { |
1097 |
s += Atrowi[k] * Acolj[k]; |
1098 |
} |
1099 |
B[i][j] = s / coef; |
1100 |
} |
1101 |
} |
1102 |
return X;
|
1103 |
} |
1104 |
|
1105 |
public Matrix firstLineMinus(Matrix B) {
|
1106 |
checkColumnDimension(B); |
1107 |
Matrix X = new Matrix(m, n);
|
1108 |
double[][] C = X.getArray(); |
1109 |
for (int i = 0; i < m; i++) { |
1110 |
for (int j = 0; j < n; j++) { |
1111 |
C[i][j] = A[i][j] - B.A[0][j];
|
1112 |
} |
1113 |
} |
1114 |
return X;
|
1115 |
|
1116 |
} |
1117 |
public Matrix nulColumnSuppression() {
|
1118 |
ArrayList<Integer> cols = new ArrayList<Integer>(); |
1119 |
for (int j = 0; j < n; j++) { |
1120 |
for (int i = 0; i < m; i++) { |
1121 |
if (A[i][j] != 0) { |
1122 |
cols.add(j); |
1123 |
break;
|
1124 |
} |
1125 |
} |
1126 |
} |
1127 |
int[] c = new int[cols.size()]; |
1128 |
for (int i = 0; i < c.length; i++) { |
1129 |
c[i] = cols.get(i); |
1130 |
} |
1131 |
return getMatrix(0, m - 1, c); |
1132 |
} |
1133 |
|
1134 |
public Matrix arrayInverse() {
|
1135 |
Matrix X = new Matrix(m, n);
|
1136 |
double[][] C = X.getArray(); |
1137 |
for (int i = 0; i < m; i++) { |
1138 |
for (int j = 0; j < n; j++) { |
1139 |
C[i][j] = 1 / A[i][j];
|
1140 |
} |
1141 |
} |
1142 |
return X;
|
1143 |
} |
1144 |
|
1145 |
public Matrix arraySqrt() {
|
1146 |
Matrix X = new Matrix(m, n);
|
1147 |
double[][] C = X.getArray(); |
1148 |
for (int i = 0; i < m; i++) { |
1149 |
for (int j = 0; j < n; j++) { |
1150 |
C[i][j] = Math.sqrt(A[i][j]);
|
1151 |
} |
1152 |
} |
1153 |
return X;
|
1154 |
} |
1155 |
|
1156 |
public Matrix firstColumTimes(Matrix B) {
|
1157 |
checkLineDimension(B); |
1158 |
Matrix X = new Matrix(m, n);
|
1159 |
double[][] C = X.getArray(); |
1160 |
for (int i = 0; i < m; i++) { |
1161 |
for (int j = 0; j < n; j++) { |
1162 |
|
1163 |
C[i][j] = A[i][j] * B.A[i][0];
|
1164 |
} |
1165 |
} |
1166 |
return X;
|
1167 |
} |
1168 |
|
1169 |
public Matrix firstColumnTimesEqual(Matrix B) {
|
1170 |
checkLineDimension(B); |
1171 |
for (int i = 0; i < m; i++) { |
1172 |
for (int j = 0; j < n; j++) { |
1173 |
|
1174 |
A[i][j] *= B.A[i][0];
|
1175 |
} |
1176 |
} |
1177 |
return this; |
1178 |
} |
1179 |
|
1180 |
public Matrix firstLineTimesEqual(Matrix B) {
|
1181 |
checkColumnDimension(B); |
1182 |
for (int i = 0; i < m; i++) { |
1183 |
for (int j = 0; j < n; j++) { |
1184 |
A[i][j] *= B.A[0][j];
|
1185 |
} |
1186 |
} |
1187 |
return this; |
1188 |
} |
1189 |
|
1190 |
public Matrix firstLineTimes(Matrix B) {
|
1191 |
checkColumnDimension(B); |
1192 |
Matrix X = new Matrix(m, n);
|
1193 |
double[][] C = X.getArray(); |
1194 |
for (int i = 0; i < m; i++) { |
1195 |
for (int j = 0; j < n; j++) { |
1196 |
C[i][j] = A[i][j] * B.A[0][j];
|
1197 |
} |
1198 |
} |
1199 |
return X;
|
1200 |
} |
1201 |
|
1202 |
public void fixColumnSigns() { |
1203 |
for (int j = 0; j < n; j++) { |
1204 |
int i = 0; |
1205 |
while (i<m && A[i][j] == 0) i++; |
1206 |
if (i==m || A[i][j] > 0) continue; |
1207 |
while (i < m) {
|
1208 |
A[i][j] = -A[i][j]; |
1209 |
i++; |
1210 |
} |
1211 |
} |
1212 |
} |
1213 |
|
1214 |
/** Analyse factorielle de correspondances
|
1215 |
@return Analyse factorielle de correspondances
|
1216 |
@see Analyse factorielle de correspondances
|
1217 |
*/
|
1218 |
public AFC afc() {
|
1219 |
return new AFC(this); |
1220 |
} |
1221 |
} |