Clover icon

Coverage Report

  1. Project Clover database Thu Nov 7 2024 13:01:17 GMT
  2. Package jalview.math

File SparseMatrixTest.java

 

Code metrics

34
157
15
1
446
283
32
0.2
10.47
15
2.13

Classes

Class Line # Actions
SparseMatrixTest 33 157 32
0.8689320786.9%
 

Contributing tests

This file is covered by 11 tests. .

Source view

1    /*
2    * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3    * Copyright (C) $$Year-Rel$$ The Jalview Authors
4    *
5    * This file is part of Jalview.
6    *
7    * Jalview is free software: you can redistribute it and/or
8    * modify it under the terms of the GNU General Public License
9    * as published by the Free Software Foundation, either version 3
10    * of the License, or (at your option) any later version.
11    *
12    * Jalview is distributed in the hope that it will be useful, but
13    * WITHOUT ANY WARRANTY; without even the implied warranty
14    * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15    * PURPOSE. See the GNU General Public License for more details.
16    *
17    * You should have received a copy of the GNU General Public License
18    * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19    * The Jalview Authors are detailed in the 'AUTHORS' file.
20    */
21    package jalview.math;
22   
23    import static org.testng.Assert.assertEquals;
24    import static org.testng.Assert.assertFalse;
25    import static org.testng.Assert.assertTrue;
26    import static org.testng.Assert.fail;
27   
28    import java.util.Random;
29   
30    import org.testng.annotations.Test;
31    import org.testng.internal.junit.ArrayAsserts;
32   
 
33    public class SparseMatrixTest
34    {
35    final static double DELTA = 0.0001d;
36   
37    Random r = new Random(1729);
38   
 
39  1 toggle @Test(groups = "Functional")
40    public void testConstructor()
41    {
42  1 MatrixI m1 = new SparseMatrix(
43    new double[][]
44    { { 2, 0, 4 }, { 0, 6, 0 } });
45  1 assertEquals(m1.getValue(0, 0), 2d);
46  1 assertEquals(m1.getValue(0, 1), 0d);
47  1 assertEquals(m1.getValue(0, 2), 4d);
48  1 assertEquals(m1.getValue(1, 0), 0d);
49  1 assertEquals(m1.getValue(1, 1), 6d);
50  1 assertEquals(m1.getValue(1, 2), 0d);
51    }
52   
 
53  1 toggle @Test(groups = "Functional")
54    public void testTranspose()
55    {
56  1 MatrixI m1 = new SparseMatrix(
57    new double[][]
58    { { 2, 0, 4 }, { 5, 6, 0 } });
59  1 MatrixI m2 = m1.transpose();
60  1 assertTrue(m2 instanceof SparseMatrix);
61  1 assertEquals(m2.height(), 3);
62  1 assertEquals(m2.width(), 2);
63  1 assertEquals(m2.getValue(0, 0), 2d);
64  1 assertEquals(m2.getValue(0, 1), 5d);
65  1 assertEquals(m2.getValue(1, 0), 0d);
66  1 assertEquals(m2.getValue(1, 1), 6d);
67  1 assertEquals(m2.getValue(2, 0), 4d);
68  1 assertEquals(m2.getValue(2, 1), 0d);
69    }
70   
 
71  1 toggle @Test(groups = "Functional")
72    public void testPreMultiply()
73    {
74  1 MatrixI m1 = new SparseMatrix(new double[][] { { 2, 3, 4 } }); // 1x3
75  1 MatrixI m2 = new SparseMatrix(new double[][] { { 5 }, { 6 }, { 7 } }); // 3x1
76   
77    /*
78    * 1x3 times 3x1 is 1x1
79    * 2x5 + 3x6 + 4*7 = 56
80    */
81  1 MatrixI m3 = m2.preMultiply(m1);
82  1 assertFalse(m3 instanceof SparseMatrix);
83  1 assertEquals(m3.height(), 1);
84  1 assertEquals(m3.width(), 1);
85  1 assertEquals(m3.getValue(0, 0), 56d);
86   
87    /*
88    * 3x1 times 1x3 is 3x3
89    */
90  1 m3 = m1.preMultiply(m2);
91  1 assertEquals(m3.height(), 3);
92  1 assertEquals(m3.width(), 3);
93  1 assertEquals(m3.getValue(0, 0), 10d);
94  1 assertEquals(m3.getValue(0, 1), 15d);
95  1 assertEquals(m3.getValue(0, 2), 20d);
96  1 assertEquals(m3.getValue(1, 0), 12d);
97  1 assertEquals(m3.getValue(1, 1), 18d);
98  1 assertEquals(m3.getValue(1, 2), 24d);
99  1 assertEquals(m3.getValue(2, 0), 14d);
100  1 assertEquals(m3.getValue(2, 1), 21d);
101  1 assertEquals(m3.getValue(2, 2), 28d);
102    }
103   
 
104  1 toggle @Test(
105    groups = "Functional",
106    expectedExceptions =
107    { IllegalArgumentException.class })
108    public void testPreMultiply_tooManyColumns()
109    {
110  1 Matrix m1 = new SparseMatrix(
111    new double[][]
112    { { 2, 3, 4 }, { 3, 4, 5 } }); // 2x3
113   
114    /*
115    * 2x3 times 2x3 invalid operation -
116    * multiplier has more columns than multiplicand has rows
117    */
118  1 m1.preMultiply(m1);
119  0 fail("Expected exception");
120    }
121   
 
122  1 toggle @Test(
123    groups = "Functional",
124    expectedExceptions =
125    { IllegalArgumentException.class })
126    public void testPreMultiply_tooFewColumns()
127    {
128  1 Matrix m1 = new SparseMatrix(
129    new double[][]
130    { { 2, 3, 4 }, { 3, 4, 5 } }); // 2x3
131   
132    /*
133    * 3x2 times 3x2 invalid operation -
134    * multiplier has more columns than multiplicand has row
135    */
136  1 m1.preMultiply(m1);
137  0 fail("Expected exception");
138    }
139   
 
140  1 toggle @Test(groups = "Functional")
141    public void testPostMultiply()
142    {
143    /*
144    * Square matrices
145    * (2 3) . (10 100)
146    * (4 5) (1000 10000)
147    * =
148    * (3020 30200)
149    * (5040 50400)
150    */
151  1 MatrixI m1 = new SparseMatrix(new double[][] { { 2, 3 }, { 4, 5 } });
152  1 MatrixI m2 = new SparseMatrix(
153    new double[][]
154    { { 10, 100 }, { 1000, 10000 } });
155  1 MatrixI m3 = m1.postMultiply(m2);
156  1 assertEquals(m3.getValue(0, 0), 3020d);
157  1 assertEquals(m3.getValue(0, 1), 30200d);
158  1 assertEquals(m3.getValue(1, 0), 5040d);
159  1 assertEquals(m3.getValue(1, 1), 50400d);
160   
161    /*
162    * also check m2.preMultiply(m1) - should be same as m1.postMultiply(m2)
163    */
164  1 MatrixI m4 = m2.preMultiply(m1);
165  1 assertMatricesMatch(m3, m4, 0.00001d);
166   
167    /*
168    * m1 has more rows than columns
169    * (2).(10 100 1000) = (20 200 2000)
170    * (3) (30 300 3000)
171    */
172  1 m1 = new SparseMatrix(new double[][] { { 2 }, { 3 } });
173  1 m2 = new SparseMatrix(new double[][] { { 10, 100, 1000 } });
174  1 m3 = m1.postMultiply(m2);
175  1 assertEquals(m3.height(), 2);
176  1 assertEquals(m3.width(), 3);
177  1 assertEquals(m3.getValue(0, 0), 20d);
178  1 assertEquals(m3.getValue(0, 1), 200d);
179  1 assertEquals(m3.getValue(0, 2), 2000d);
180  1 assertEquals(m3.getValue(1, 0), 30d);
181  1 assertEquals(m3.getValue(1, 1), 300d);
182  1 assertEquals(m3.getValue(1, 2), 3000d);
183   
184  1 m4 = m2.preMultiply(m1);
185  1 assertMatricesMatch(m3, m4, 0.00001d);
186   
187    /*
188    * m1 has more columns than rows
189    * (2 3 4) . (5 4) = (56 25)
190    * (6 3)
191    * (7 2)
192    * [0, 0] = 2*5 + 3*6 + 4*7 = 56
193    * [0, 1] = 2*4 + 3*3 + 4*2 = 25
194    */
195  1 m1 = new SparseMatrix(new double[][] { { 2, 3, 4 } });
196  1 m2 = new SparseMatrix(new double[][] { { 5, 4 }, { 6, 3 }, { 7, 2 } });
197  1 m3 = m1.postMultiply(m2);
198  1 assertEquals(m3.height(), 1);
199  1 assertEquals(m3.width(), 2);
200  1 assertEquals(m3.getValue(0, 0), 56d);
201  1 assertEquals(m3.getValue(0, 1), 25d);
202   
203    /*
204    * and check premultiply equivalent
205    */
206  1 m4 = m2.preMultiply(m1);
207  1 assertMatricesMatch(m3, m4, 0.00001d);
208    }
209   
 
210  0 toggle @Test(groups = "Timing")
211    public void testSign()
212    {
213  0 assertEquals(Matrix.sign(-1, -2), -1d);
214  0 assertEquals(Matrix.sign(-1, 2), 1d);
215  0 assertEquals(Matrix.sign(-1, 0), 1d);
216  0 assertEquals(Matrix.sign(1, -2), -1d);
217  0 assertEquals(Matrix.sign(1, 2), 1d);
218  0 assertEquals(Matrix.sign(1, 0), 1d);
219    }
220   
221    /**
222    * Verify that the results of method tred() are the same for SparseMatrix as
223    * they are for Matrix (i.e. a regression test rather than an absolute test of
224    * correctness of results)
225    */
 
226  1 toggle @Test(groups = "Functional")
227    public void testTred_matchesMatrix()
228    {
229    /*
230    * make a pseudo-random symmetric matrix as required for tred/tqli
231    */
232  1 int rows = 10;
233  1 int cols = rows;
234  1 double[][] d = getSparseValues(rows, cols, 3);
235   
236    /*
237    * make a copy of the values so m1, m2 are not
238    * sharing arrays!
239    */
240  1 double[][] d1 = new double[rows][cols];
241  11 for (int row = 0; row < rows; row++)
242    {
243  110 for (int col = 0; col < cols; col++)
244    {
245  100 d1[row][col] = d[row][col];
246    }
247    }
248  1 Matrix m1 = new Matrix(d);
249  1 Matrix m2 = new SparseMatrix(d1);
250  1 assertMatricesMatch(m1, m2, 0.00001d); // sanity check
251  1 m1.tred();
252  1 m2.tred();
253  1 assertMatricesMatch(m1, m2, 0.00001d);
254    }
255   
 
256  9 toggle private void assertMatricesMatch(MatrixI m1, MatrixI m2, double delta)
257    {
258  9 if (m1.height() != m2.height())
259    {
260  0 fail("height mismatch");
261    }
262  9 if (m1.width() != m2.width())
263    {
264  0 fail("width mismatch");
265    }
266  66 for (int row = 0; row < m1.height(); row++)
267    {
268  541 for (int col = 0; col < m1.width(); col++)
269    {
270  484 double v2 = m2.getValue(row, col);
271  484 double v1 = m1.getValue(row, col);
272  484 if (Math.abs(v1 - v2) > DELTA)
273    {
274  0 fail(String.format("At [%d, %d] %f != %f", row, col, v1, v2));
275    }
276    }
277    }
278  9 ArrayAsserts.assertArrayEquals(m1.getD(), m2.getD(), delta);
279  9 ArrayAsserts.assertArrayEquals(m1.getE(), m2.getE(), 0.00001d);
280    }
281   
 
282  0 toggle @Test
283    public void testGetValue()
284    {
285  0 double[][] d = new double[][] { { 0, 0, 1, 0, 0 }, { 2, 3, 0, 0, 0 },
286    { 4, 0, 0, 0, 5 } };
287  0 MatrixI m = new SparseMatrix(d);
288  0 for (int row = 0; row < 3; row++)
289    {
290  0 for (int col = 0; col < 5; col++)
291    {
292  0 assertEquals(m.getValue(row, col), d[row][col],
293    String.format("At [%d, %d]", row, col));
294    }
295    }
296    }
297   
298    /**
299    * Verify that the results of method tqli() are the same for SparseMatrix as
300    * they are for Matrix (i.e. a regression test rather than an absolute test of
301    * correctness of results)
302    *
303    * @throws Exception
304    */
 
305  1 toggle @Test(groups = "Functional")
306    public void testTqli_matchesMatrix() throws Exception
307    {
308    /*
309    * make a pseudo-random symmetric matrix as required for tred
310    */
311  1 int rows = 6;
312  1 int cols = rows;
313  1 double[][] d = getSparseValues(rows, cols, 3);
314   
315    /*
316    * make a copy of the values so m1, m2 are not
317    * sharing arrays!
318    */
319  1 double[][] d1 = new double[rows][cols];
320  7 for (int row = 0; row < rows; row++)
321    {
322  42 for (int col = 0; col < cols; col++)
323    {
324  36 d1[row][col] = d[row][col];
325    }
326    }
327  1 Matrix m1 = new Matrix(d);
328  1 Matrix m2 = new SparseMatrix(d1);
329   
330    // have to do tred() before doing tqli()
331  1 m1.tred();
332  1 m2.tred();
333  1 assertMatricesMatch(m1, m2, 0.00001d);
334   
335  1 m1.tqli();
336  1 m2.tqli();
337  1 assertMatricesMatch(m1, m2, 0.00001d);
338    }
339   
340    /**
341    * Helper method to make values for a sparse, pseudo-random symmetric matrix
342    *
343    * @param rows
344    * @param cols
345    * @param occupancy
346    * one in 'occupancy' entries will be non-zero
347    * @return
348    */
 
349  3 toggle public double[][] getSparseValues(int rows, int cols, int occupancy)
350    {
351    /*
352    * generate whole number values between -12 and +12
353    * (to mimic score matrices used in Jalview)
354    */
355  3 double[][] d = new double[rows][cols];
356  3 int m = 0;
357  29 for (int i = 0; i < rows; i++)
358    {
359  26 if (++m % occupancy == 0)
360    {
361  0 d[i][i] = r.nextInt() % 13; // diagonal
362    }
363  131 for (int j = 0; j < i; j++)
364    {
365  105 if (++m % occupancy == 0)
366    {
367  43 d[i][j] = r.nextInt() % 13;
368  43 d[j][i] = d[i][j];
369    }
370    }
371    }
372  3 return d;
373   
374    }
375   
376    /**
377    * Test that verifies that the result of preMultiply is a SparseMatrix if more
378    * than 80% zeroes, else a Matrix
379    */
 
380  1 toggle @Test(groups = "Functional")
381    public void testPreMultiply_sparseProduct()
382    {
383  1 MatrixI m1 = new SparseMatrix(
384    new double[][]
385    { { 1 }, { 0 }, { 0 }, { 0 }, { 0 } }); // 5x1
386  1 MatrixI m2 = new SparseMatrix(new double[][] { { 1, 1, 1, 1 } }); // 1x4
387   
388    /*
389    * m1.m2 makes a row of 4 1's, and 4 rows of zeros
390    * 20% non-zero so not 'sparse'
391    */
392  1 MatrixI m3 = m2.preMultiply(m1);
393  1 assertFalse(m3 instanceof SparseMatrix);
394   
395    /*
396    * replace a 1 with a 0 in the product:
397    * it is now > 80% zero so 'sparse'
398    */
399  1 m2 = new SparseMatrix(new double[][] { { 1, 1, 1, 0 } });
400  1 m3 = m2.preMultiply(m1);
401  1 assertTrue(m3 instanceof SparseMatrix);
402    }
403   
 
404  1 toggle @Test(groups = "Functional")
405    public void testFillRatio()
406    {
407  1 SparseMatrix m1 = new SparseMatrix(
408    new double[][]
409    { { 2, 0, 4, 1, 0 }, { 0, 6, 0, 0, 0 } });
410  1 assertEquals(m1.getFillRatio(), 0.4f);
411    }
412   
413    /**
414    * Verify that the results of method tred() are the same if the calculation is
415    * redone
416    */
 
417  1 toggle @Test(groups = "Functional")
418    public void testTred_reproducible()
419    {
420    /*
421    * make a pseudo-random symmetric matrix as required for tred/tqli
422    */
423  1 int rows = 10;
424  1 int cols = rows;
425  1 double[][] d = getSparseValues(rows, cols, 3);
426   
427    /*
428    * make a copy of the values so m1, m2 are not
429    * sharing arrays!
430    */
431  1 double[][] d1 = new double[rows][cols];
432  11 for (int row = 0; row < rows; row++)
433    {
434  110 for (int col = 0; col < cols; col++)
435    {
436  100 d1[row][col] = d[row][col];
437    }
438    }
439  1 Matrix m1 = new SparseMatrix(d);
440  1 Matrix m2 = new SparseMatrix(d1);
441  1 assertMatricesMatch(m1, m2, 1.0e16); // sanity check
442  1 m1.tred();
443  1 m2.tred();
444  1 assertMatricesMatch(m1, m2, 0.00001d);
445    }
446    }