mdarray 0.5.4-java → 0.5.5-java
Sign up to get free protection for your applications and to get access to all the features.
- data/LICENSE.txt +35 -16
- data/README.md +180 -234
- data/lib/colt/matrix/algebra.rb +466 -0
- data/lib/colt/matrix/colt_matrix.rb +161 -220
- data/lib/colt/matrix/creation.rb +374 -0
- data/lib/colt/matrix/{matrix_hierarchy.rb → hierarchy.rb} +78 -99
- data/lib/colt/matrix/property.rb +363 -0
- data/lib/mdarray/views.rb +46 -1
- data/test/colt/matrix/test_complete.rb +29 -0
- data/test/colt/{test_matrix.rb → matrix/test_creation.rb} +26 -3
- data/test/colt/{test_double_matrix2d.rb → matrix/test_matrix1d_floatingalgebra.rb} +86 -84
- data/test/colt/matrix/test_matrix2d_fixpointalgebra.rb +531 -0
- data/test/colt/matrix/test_matrix2d_floatingalgebra.rb +693 -0
- data/test/colt/matrix/test_operations.rb +120 -0
- data/test/colt/matrix/test_properties.rb +649 -0
- data/test/colt/test_complete.rb +1 -1
- data/test/complete.rb +12 -7
- data/test/mdarray/test_views.rb +55 -0
- data/vendor/{incanter.jar → mdarray.jar} +0 -0
- data/version.rb +1 -1
- metadata +22 -16
- data/lib/colt/matrix/matrix2D_floating_algebra.rb +0 -325
- data/test/colt/test_float_matrix2d.rb +0 -171
- data/vendor/commons-compiler.jar +0 -0
- data/vendor/janino.jar +0 -0
@@ -164,9 +164,32 @@ class MDArrayTest < Test::Unit::TestCase
|
|
164
164
|
|
165
165
|
end
|
166
166
|
|
167
|
+
#-------------------------------------------------------------------------------------
|
168
|
+
#
|
169
|
+
#-------------------------------------------------------------------------------------
|
170
|
+
|
171
|
+
should "create matrices from high order sliced MDArray" do
|
172
|
+
|
173
|
+
d1 = MDArray.typed_arange("double", 0, 420)
|
174
|
+
d1.reshape!([5, 4, 3, 7])
|
175
|
+
|
176
|
+
p "three dimensional array"
|
177
|
+
d1.print
|
178
|
+
matrix = MDMatrix.from_mdarray(d1.slice(0, 0))
|
179
|
+
p "two dimensional matrix from given array"
|
180
|
+
matrix.print
|
181
|
+
p " "
|
182
|
+
|
183
|
+
p "from 4 to two"
|
184
|
+
matrix = MDMatrix.from_mdarray(d1.region(:spec => "0:0, 0:0, 0:2, 0:6").reduce)
|
185
|
+
matrix.print
|
186
|
+
p " "
|
187
|
+
matrix.mdarray.print
|
188
|
+
|
189
|
+
# matrix = MDMatrix.from_mdarray(
|
190
|
+
|
191
|
+
end
|
192
|
+
|
167
193
|
end
|
168
194
|
|
169
195
|
end
|
170
|
-
|
171
|
-
require_relative 'test_double_matrix2d'
|
172
|
-
# require_relative 'test_float_matrix2d'
|
@@ -38,105 +38,122 @@ class MDArrayTest < Test::Unit::TestCase
|
|
38
38
|
#
|
39
39
|
#-------------------------------------------------------------------------------------
|
40
40
|
|
41
|
-
should "test matrix functions" do
|
41
|
+
should "test 1d double matrix functions" do
|
42
42
|
|
43
43
|
b = MDMatrix.double([3], [1.5, 1, 1.3])
|
44
44
|
|
45
|
-
pos = MDArray.double([
|
46
|
-
|
47
|
-
|
48
|
-
p "
|
49
|
-
|
45
|
+
pos = MDArray.double([9], [2, -1, 0, -1, 2, -1, 0, -1, 2])
|
46
|
+
matrix1 = MDMatrix.from_mdarray(pos)
|
47
|
+
|
48
|
+
p "getting regions from the above matrix"
|
49
|
+
p "specification is '0:6'"
|
50
|
+
matrix1.region(:spec => "0:6").print
|
50
51
|
printf("\n\n")
|
51
52
|
|
52
|
-
|
53
|
-
|
54
|
-
p "eigenvalue matrix"
|
55
|
-
eig[0].print
|
53
|
+
p "specification is '0:9:2'"
|
54
|
+
matrix1.region(:spec => "0:9:2").print
|
56
55
|
printf("\n\n")
|
57
|
-
|
58
|
-
|
56
|
+
|
57
|
+
p "flipping dim 0 of the matrix"
|
58
|
+
matrix1.flip(0).print
|
59
59
|
printf("\n\n")
|
60
|
-
|
61
|
-
|
60
|
+
|
61
|
+
m = MDArray.typed_arange("double", 16, 32)
|
62
|
+
matrix2 = MDMatrix.from_mdarray(m).region(:spec => "0:9")
|
63
|
+
matrix2.print
|
62
64
|
printf("\n\n")
|
63
|
-
|
64
|
-
|
65
|
+
|
66
|
+
p "matrix multiplication - dot product of two vectors"
|
67
|
+
result = matrix1 * matrix2
|
68
|
+
p result
|
65
69
|
printf("\n\n")
|
66
70
|
|
67
|
-
|
68
|
-
p "
|
69
|
-
|
70
|
-
p "determinant: #{lu[1]}"
|
71
|
-
p "pivot vector: #{lu[2]}"
|
72
|
-
p "lower triangular matrix"
|
73
|
-
lu[3].print
|
71
|
+
result = matrix1.kron(matrix2)
|
72
|
+
p "Kronecker multiplication"
|
73
|
+
result.print
|
74
74
|
printf("\n\n")
|
75
|
-
|
76
|
-
|
75
|
+
|
76
|
+
p "norm1"
|
77
|
+
p result.norm1
|
78
|
+
|
79
|
+
p "norm2"
|
80
|
+
p result.norm2
|
81
|
+
|
82
|
+
p "Returns the Frobenius norm of matrix A, which is Sqrt(Sum(A[i,j]^2))"
|
83
|
+
p result.normF
|
84
|
+
|
85
|
+
p "Returns the infinity norm of matrix A, which is the maximum absolute row sum."
|
86
|
+
p result.norm_infinity
|
87
|
+
|
88
|
+
result.normalize!
|
89
|
+
result.print
|
77
90
|
printf("\n\n")
|
78
91
|
|
79
|
-
|
80
|
-
|
81
|
-
|
82
|
-
p matrix.cond
|
92
|
+
p "summing all values of result: #{result.sum}"
|
93
|
+
|
94
|
+
result.mdarray.print
|
83
95
|
|
84
|
-
#
|
85
|
-
|
86
|
-
|
87
|
-
|
96
|
+
# val1 is not a matrix2d... wrong test file. Need to actually work with division on
|
97
|
+
# Matrix2D.
|
98
|
+
p "dividing two matrices... actually multiply by the inverse"
|
99
|
+
val1 = MDMatrix.double([2], [118.4, 135.2])
|
100
|
+
val2 = MDMatrix.double([2, 2], [3, 3.5, 3.2, 3.6])
|
101
|
+
val1.print
|
102
|
+
printf("\n\n")
|
103
|
+
val2.print
|
88
104
|
printf("\n\n")
|
89
105
|
|
90
|
-
|
91
|
-
p "
|
92
|
-
|
93
|
-
solve.print
|
106
|
+
div = val1 / val2
|
107
|
+
p "result of division"
|
108
|
+
div.print
|
94
109
|
printf("\n\n")
|
95
110
|
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
p "Orthogonal factor Q:"
|
100
|
-
qr[1].print
|
111
|
+
p "multiply by the inverse"
|
112
|
+
r1 = val2.inverse
|
113
|
+
r1.print
|
101
114
|
printf("\n\n")
|
102
|
-
|
103
|
-
|
115
|
+
|
116
|
+
val1.print
|
104
117
|
printf("\n\n")
|
105
118
|
|
106
|
-
|
107
|
-
|
108
|
-
p "operation success? #{svd[0]}" # 0 success; < 0 ith value is illegal; > 0 not converge
|
109
|
-
p "cond: #{svd[1]}"
|
110
|
-
p "norm2: #{svd[2]}"
|
111
|
-
p "rank: #{svd[3]}"
|
112
|
-
p "singular values"
|
113
|
-
p svd[4]
|
114
|
-
p "Diagonal matrix of singular values"
|
115
|
-
# svd[5].print
|
119
|
+
mult = r1 * val1
|
120
|
+
mult.print
|
116
121
|
printf("\n\n")
|
117
|
-
|
118
|
-
|
122
|
+
|
123
|
+
end
|
124
|
+
|
125
|
+
#-------------------------------------------------------------------------------------
|
126
|
+
#
|
127
|
+
#-------------------------------------------------------------------------------------
|
128
|
+
|
129
|
+
should "test 1D float matrix functions" do
|
130
|
+
|
131
|
+
b = MDMatrix.float([3], [1.5, 1, 1.3])
|
132
|
+
|
133
|
+
pos = MDArray.float([9], [2, -1, 0, -1, 2, -1, 0, -1, 2])
|
134
|
+
matrix1 = MDMatrix.from_mdarray(pos)
|
135
|
+
|
136
|
+
p "getting regions from the above matrix"
|
137
|
+
p "specification is '0:6'"
|
138
|
+
matrix1.region(:spec => "0:6").print
|
119
139
|
printf("\n\n")
|
120
|
-
|
121
|
-
|
140
|
+
|
141
|
+
p "specification is '0:9:2'"
|
142
|
+
matrix1.region(:spec => "0:9:2").print
|
122
143
|
printf("\n\n")
|
123
144
|
|
124
|
-
|
125
|
-
|
126
|
-
matrix1 = MDMatrix.from_mdarray(m)
|
127
|
-
# mat2 = matrix.chol
|
128
|
-
matrix1.print
|
145
|
+
p "flipping dim 0 of the matrix"
|
146
|
+
matrix1.flip(0).print
|
129
147
|
printf("\n\n")
|
130
148
|
|
131
|
-
m = MDArray.typed_arange("
|
132
|
-
m.
|
133
|
-
matrix2 = MDMatrix.from_mdarray(m)
|
149
|
+
m = MDArray.typed_arange("float", 16, 32)
|
150
|
+
matrix2 = MDMatrix.from_mdarray(m).region(:spec => "0:9")
|
134
151
|
matrix2.print
|
135
152
|
printf("\n\n")
|
136
153
|
|
154
|
+
p "matrix multiplication - dot product of two vectors"
|
137
155
|
result = matrix1 * matrix2
|
138
|
-
p
|
139
|
-
result.print
|
156
|
+
p result
|
140
157
|
printf("\n\n")
|
141
158
|
|
142
159
|
result = matrix1.kron(matrix2)
|
@@ -144,9 +161,6 @@ class MDArrayTest < Test::Unit::TestCase
|
|
144
161
|
result.print
|
145
162
|
printf("\n\n")
|
146
163
|
|
147
|
-
print "determinant is: #{result.det}"
|
148
|
-
printf("\n\n")
|
149
|
-
|
150
164
|
p "norm1"
|
151
165
|
p result.norm1
|
152
166
|
|
@@ -159,18 +173,6 @@ class MDArrayTest < Test::Unit::TestCase
|
|
159
173
|
p "Returns the infinity norm of matrix A, which is the maximum absolute row sum."
|
160
174
|
p result.norm_infinity
|
161
175
|
|
162
|
-
power3 = result ** 3
|
163
|
-
power3.print
|
164
|
-
printf("\n\n")
|
165
|
-
|
166
|
-
p result.trace
|
167
|
-
|
168
|
-
trap_lower = result.trapezoidal_lower
|
169
|
-
trap_lower.print
|
170
|
-
printf("\n\n")
|
171
|
-
|
172
|
-
p result.vector_norm2
|
173
|
-
|
174
176
|
result.normalize!
|
175
177
|
result.print
|
176
178
|
printf("\n\n")
|
@@ -180,7 +182,7 @@ class MDArrayTest < Test::Unit::TestCase
|
|
180
182
|
result.mdarray.print
|
181
183
|
|
182
184
|
end
|
183
|
-
|
185
|
+
|
184
186
|
end
|
185
187
|
|
186
188
|
end
|
@@ -0,0 +1,531 @@
|
|
1
|
+
# -*- coding: utf-8 -*-
|
2
|
+
|
3
|
+
##########################################################################################
|
4
|
+
# Copyright © 2013 Rodrigo Botafogo. All Rights Reserved. Permission to use, copy, modify,
|
5
|
+
# and distribute this software and its documentation, without fee and without a signed
|
6
|
+
# licensing agreement, is hereby granted, provided that the above copyright notice, this
|
7
|
+
# paragraph and the following two paragraphs appear in all copies, modifications, and
|
8
|
+
# distributions.
|
9
|
+
#
|
10
|
+
# IN NO EVENT SHALL RODRIGO BOTAFOGO BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL,
|
11
|
+
# INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF
|
12
|
+
# THIS SOFTWARE AND ITS DOCUMENTATION, EVEN IF RODRIGO BOTAFOGO HAS BEEN ADVISED OF THE
|
13
|
+
# POSSIBILITY OF SUCH DAMAGE.
|
14
|
+
#
|
15
|
+
# RODRIGO BOTAFOGO SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
|
16
|
+
# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
|
17
|
+
# SOFTWARE AND ACCOMPANYING DOCUMENTATION, IF ANY, PROVIDED HEREUNDER IS PROVIDED "AS IS".
|
18
|
+
# RODRIGO BOTAFOGO HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS,
|
19
|
+
# OR MODIFICATIONS.
|
20
|
+
##########################################################################################
|
21
|
+
|
22
|
+
require 'rubygems'
|
23
|
+
require "test/unit"
|
24
|
+
require 'shoulda'
|
25
|
+
|
26
|
+
require 'mdarray'
|
27
|
+
|
28
|
+
class MDArrayTest < Test::Unit::TestCase
|
29
|
+
|
30
|
+
context "Colt Matrix" do
|
31
|
+
|
32
|
+
setup do
|
33
|
+
|
34
|
+
|
35
|
+
end
|
36
|
+
|
37
|
+
#-------------------------------------------------------------------------------------
|
38
|
+
#
|
39
|
+
#-------------------------------------------------------------------------------------
|
40
|
+
|
41
|
+
should "get and set values for long Matrix" do
|
42
|
+
|
43
|
+
a = MDMatrix.long([4, 4])
|
44
|
+
a[0, 0] = 1
|
45
|
+
assert_equal(1, a[0, 0])
|
46
|
+
assert_equal(0.0, a[0, 1])
|
47
|
+
|
48
|
+
a.fill(2)
|
49
|
+
assert_equal(2, a[3, 3])
|
50
|
+
|
51
|
+
b = MDMatrix.long([4, 4])
|
52
|
+
b.fill(a)
|
53
|
+
assert_equal(2, b[1, 3])
|
54
|
+
|
55
|
+
# fill the matrix with the value of a Proc evaluation. The argument to the
|
56
|
+
# Proc is the content of the array at the given index, i.e, x = b[i] for all i.
|
57
|
+
func = Proc.new { |x| x ** 2 }
|
58
|
+
b.fill(func)
|
59
|
+
assert_equal(4, b[0, 3])
|
60
|
+
b.print
|
61
|
+
printf("\n\n")
|
62
|
+
|
63
|
+
# fill the Matrix with the value of method apply from a given class.
|
64
|
+
# In general this solution is more efficient than the above solution with
|
65
|
+
# Proc.
|
66
|
+
class LongFunc
|
67
|
+
def self.apply(x)
|
68
|
+
x/2
|
69
|
+
end
|
70
|
+
end
|
71
|
+
|
72
|
+
b.fill(LongFunc)
|
73
|
+
assert_equal(2, b[2,0])
|
74
|
+
b.print
|
75
|
+
printf("\n\n")
|
76
|
+
|
77
|
+
# defines a class with a method apply with two arguments
|
78
|
+
class LongFunc2
|
79
|
+
def self.apply(x, y)
|
80
|
+
(x + y) ** 2
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
# fill array a with the value the result of a function to each cell;
|
85
|
+
# x[row,col] = function(x[row,col],y[row,col]).
|
86
|
+
a.fill(b, LongFunc2)
|
87
|
+
a.print
|
88
|
+
printf("\n\n")
|
89
|
+
|
90
|
+
tens = MDMatrix.init_with("long", [5, 3], 10.0)
|
91
|
+
tens.print
|
92
|
+
printf("\n\n")
|
93
|
+
|
94
|
+
typed_arange = MDMatrix.typed_arange("long", 0, 20, 2)
|
95
|
+
typed_arange.print
|
96
|
+
printf("\n\n")
|
97
|
+
|
98
|
+
linspace = MDMatrix.linspace("long", 0, 10, 50)
|
99
|
+
linspace.print
|
100
|
+
printf("\n\n")
|
101
|
+
|
102
|
+
# set the value of all cells that are bigger than 5 to 1.0
|
103
|
+
linspace.fill_cond(Proc.new { |x| x > 5 }, 1.0)
|
104
|
+
linspace.print
|
105
|
+
printf("\n\n")
|
106
|
+
|
107
|
+
# set the value of all cells that are smaller than 5 to the square value
|
108
|
+
linspace.fill_cond(Proc.new { |x| x < 5 }, Proc.new { |x| x * x })
|
109
|
+
linspace.print
|
110
|
+
printf("\n\n")
|
111
|
+
|
112
|
+
ones = MDMatrix.ones("long", [3, 5])
|
113
|
+
ones.print
|
114
|
+
printf("\n\n")
|
115
|
+
|
116
|
+
arange = MDMatrix.arange(0, 10)
|
117
|
+
arange.print
|
118
|
+
printf("\n\n")
|
119
|
+
|
120
|
+
end
|
121
|
+
|
122
|
+
#-------------------------------------------------------------------------------------
|
123
|
+
#
|
124
|
+
#-------------------------------------------------------------------------------------
|
125
|
+
|
126
|
+
should "test 2d long matrix functions" do
|
127
|
+
|
128
|
+
m = MDArray.typed_arange("long", 0, 16)
|
129
|
+
m.reshape!([4, 4])
|
130
|
+
matrix1 = MDMatrix.from_mdarray(m)
|
131
|
+
# mat2 = matrix.chol
|
132
|
+
matrix1.print
|
133
|
+
printf("\n\n")
|
134
|
+
|
135
|
+
p "Transposing the above matrix"
|
136
|
+
matrix1.transpose.print
|
137
|
+
printf("\n\n")
|
138
|
+
|
139
|
+
p "getting regions from the above matrix"
|
140
|
+
p "specification is '0:0, 1:3'"
|
141
|
+
matrix1.region(:spec => "0:0, 1:3").print
|
142
|
+
printf("\n\n")
|
143
|
+
|
144
|
+
p "specification is '0:3:2, 1:3'"
|
145
|
+
matrix1.region(:spec => "0:3:2, 1:3").print
|
146
|
+
printf("\n\n")
|
147
|
+
|
148
|
+
p "flipping dim 0 of the matrix"
|
149
|
+
matrix1.flip(0).print
|
150
|
+
printf("\n\n")
|
151
|
+
|
152
|
+
m = MDArray.typed_arange("long", 16, 32)
|
153
|
+
m.reshape!([4, 4])
|
154
|
+
matrix2 = MDMatrix.from_mdarray(m)
|
155
|
+
matrix2.print
|
156
|
+
printf("\n\n")
|
157
|
+
|
158
|
+
result = matrix1 * matrix2
|
159
|
+
p "matrix multiplication of square matrices"
|
160
|
+
result.print
|
161
|
+
printf("\n\n")
|
162
|
+
|
163
|
+
p "matrix multiplication of rec matrices"
|
164
|
+
array1 = MDMatrix.long([2, 3], [1, 2, 3, 4, 5, 6])
|
165
|
+
array2 = MDMatrix.long([3, 2], [1, 2, 3, 4, 5, 6])
|
166
|
+
mult = array1 * array2
|
167
|
+
mult.print
|
168
|
+
printf("\n\n")
|
169
|
+
|
170
|
+
p "matrix multiplication of rec matrices passing alpha and beta parameters."
|
171
|
+
p "C = alpha * A x B + beta*C"
|
172
|
+
mult = array1.mult(array2, 2, 2)
|
173
|
+
mult.print
|
174
|
+
printf("\n\n")
|
175
|
+
|
176
|
+
p "matrix multiplication of rec matrices passing alpha, beta and return (C) parameters."
|
177
|
+
p "C = alpha * A x B + beta*C"
|
178
|
+
result = MDMatrix.long([2, 2], [2, 2, 2, 2])
|
179
|
+
mult = array1.mult(array2, 2, 2, false, false, result)
|
180
|
+
mult.print
|
181
|
+
printf("\n\n")
|
182
|
+
|
183
|
+
p "matrix multiplication by vector"
|
184
|
+
array1 = MDMatrix.long([2, 3], [1, 2, 3, 4, 5, 6])
|
185
|
+
array2 = MDMatrix.long([3], [4, 5, 6])
|
186
|
+
mult = array1 * array2
|
187
|
+
mult.print
|
188
|
+
printf("\n\n")
|
189
|
+
|
190
|
+
end
|
191
|
+
|
192
|
+
#-------------------------------------------------------------------------------------
|
193
|
+
#
|
194
|
+
#-------------------------------------------------------------------------------------
|
195
|
+
|
196
|
+
should "get and set values for int Matrix" do
|
197
|
+
|
198
|
+
a = MDMatrix.int([4, 4])
|
199
|
+
a[0, 0] = 1
|
200
|
+
assert_equal(1, a[0, 0])
|
201
|
+
assert_equal(0.0, a[0, 1])
|
202
|
+
|
203
|
+
a.fill(2)
|
204
|
+
assert_equal(2, a[3, 3])
|
205
|
+
|
206
|
+
b = MDMatrix.int([4, 4])
|
207
|
+
b.fill(a)
|
208
|
+
assert_equal(2, b[1, 3])
|
209
|
+
|
210
|
+
# fill the matrix with the value of a Proc evaluation. The argument to the
|
211
|
+
# Proc is the content of the array at the given index, i.e, x = b[i] for all i.
|
212
|
+
func = Proc.new { |x| x ** 2 }
|
213
|
+
b.fill(func)
|
214
|
+
assert_equal(4, b[0, 3])
|
215
|
+
b.print
|
216
|
+
printf("\n\n")
|
217
|
+
|
218
|
+
# fill the Matrix with the value of method apply from a given class.
|
219
|
+
# In general this solution is more efficient than the above solution with
|
220
|
+
# Proc.
|
221
|
+
class IntFunc
|
222
|
+
def self.apply(x)
|
223
|
+
x/2
|
224
|
+
end
|
225
|
+
end
|
226
|
+
|
227
|
+
b.fill(IntFunc)
|
228
|
+
assert_equal(2, b[2,0])
|
229
|
+
b.print
|
230
|
+
printf("\n\n")
|
231
|
+
|
232
|
+
# defines a class with a method apply with two arguments
|
233
|
+
class IntFunc2
|
234
|
+
def self.apply(x, y)
|
235
|
+
(x + y) ** 2
|
236
|
+
end
|
237
|
+
end
|
238
|
+
|
239
|
+
# fill array a with the value the result of a function to each cell;
|
240
|
+
# x[row,col] = function(x[row,col],y[row,col]).
|
241
|
+
a.fill(b, IntFunc2)
|
242
|
+
a.print
|
243
|
+
printf("\n\n")
|
244
|
+
|
245
|
+
tens = MDMatrix.init_with("int", [5, 3], 10.0)
|
246
|
+
tens.print
|
247
|
+
printf("\n\n")
|
248
|
+
|
249
|
+
typed_arange = MDMatrix.typed_arange("int", 0, 20, 2)
|
250
|
+
typed_arange.print
|
251
|
+
printf("\n\n")
|
252
|
+
|
253
|
+
linspace = MDMatrix.linspace("int", 0, 10, 50)
|
254
|
+
linspace.print
|
255
|
+
printf("\n\n")
|
256
|
+
|
257
|
+
# set the value of all cells that are bigger than 5 to 1.0
|
258
|
+
linspace.fill_cond(Proc.new { |x| x > 5 }, 1.0)
|
259
|
+
linspace.print
|
260
|
+
printf("\n\n")
|
261
|
+
|
262
|
+
# set the value of all cells that are smaller than 5 to the square value
|
263
|
+
linspace.fill_cond(Proc.new { |x| x < 5 }, Proc.new { |x| x * x })
|
264
|
+
linspace.print
|
265
|
+
printf("\n\n")
|
266
|
+
|
267
|
+
ones = MDMatrix.ones("int", [3, 5])
|
268
|
+
ones.print
|
269
|
+
printf("\n\n")
|
270
|
+
|
271
|
+
arange = MDMatrix.arange(0, 10)
|
272
|
+
arange.print
|
273
|
+
printf("\n\n")
|
274
|
+
|
275
|
+
end
|
276
|
+
|
277
|
+
#-------------------------------------------------------------------------------------
|
278
|
+
#
|
279
|
+
#-------------------------------------------------------------------------------------
|
280
|
+
|
281
|
+
should "test 2d int matrix functions" do
|
282
|
+
|
283
|
+
m = MDArray.typed_arange("int", 0, 16)
|
284
|
+
m.reshape!([4, 4])
|
285
|
+
matrix1 = MDMatrix.from_mdarray(m)
|
286
|
+
# mat2 = matrix.chol
|
287
|
+
matrix1.print
|
288
|
+
printf("\n\n")
|
289
|
+
|
290
|
+
p "Transposing the above matrix"
|
291
|
+
matrix1.transpose.print
|
292
|
+
printf("\n\n")
|
293
|
+
|
294
|
+
p "getting regions from the above matrix"
|
295
|
+
p "specification is '0:0, 1:3'"
|
296
|
+
matrix1.region(:spec => "0:0, 1:3").print
|
297
|
+
printf("\n\n")
|
298
|
+
|
299
|
+
p "specification is '0:3:2, 1:3'"
|
300
|
+
matrix1.region(:spec => "0:3:2, 1:3").print
|
301
|
+
printf("\n\n")
|
302
|
+
|
303
|
+
p "flipping dim 0 of the matrix"
|
304
|
+
matrix1.flip(0).print
|
305
|
+
printf("\n\n")
|
306
|
+
|
307
|
+
m = MDArray.typed_arange("int", 16, 32)
|
308
|
+
m.reshape!([4, 4])
|
309
|
+
matrix2 = MDMatrix.from_mdarray(m)
|
310
|
+
matrix2.print
|
311
|
+
printf("\n\n")
|
312
|
+
|
313
|
+
result = matrix1 * matrix2
|
314
|
+
p "matrix multiplication of square matrices"
|
315
|
+
result.print
|
316
|
+
printf("\n\n")
|
317
|
+
|
318
|
+
p "matrix multiplication of rec matrices"
|
319
|
+
array1 = MDMatrix.int([2, 3], [1, 2, 3, 4, 5, 6])
|
320
|
+
array2 = MDMatrix.int([3, 2], [1, 2, 3, 4, 5, 6])
|
321
|
+
mult = array1 * array2
|
322
|
+
mult.print
|
323
|
+
printf("\n\n")
|
324
|
+
|
325
|
+
p "matrix multiplication of rec matrices passing alpha and beta parameters."
|
326
|
+
p "C = alpha * A x B + beta*C"
|
327
|
+
mult = array1.mult(array2, 2, 2)
|
328
|
+
mult.print
|
329
|
+
printf("\n\n")
|
330
|
+
|
331
|
+
p "matrix multiplication of rec matrices passing alpha, beta and return (C) parameters."
|
332
|
+
p "C = alpha * A x B + beta*C"
|
333
|
+
result = MDMatrix.int([2, 2], [2, 2, 2, 2])
|
334
|
+
mult = array1.mult(array2, 2, 2, false, false, result)
|
335
|
+
mult.print
|
336
|
+
printf("\n\n")
|
337
|
+
|
338
|
+
p "matrix multiplication by vector"
|
339
|
+
array1 = MDMatrix.int([2, 3], [1, 2, 3, 4, 5, 6])
|
340
|
+
array2 = MDMatrix.int([3], [4, 5, 6])
|
341
|
+
mult = array1 * array2
|
342
|
+
mult.print
|
343
|
+
printf("\n\n")
|
344
|
+
|
345
|
+
end
|
346
|
+
|
347
|
+
#=end
|
348
|
+
end
|
349
|
+
|
350
|
+
end
|
351
|
+
|
352
|
+
=begin
|
353
|
+
|
354
|
+
Example of SVD decomposition with PColt. Check if working!
|
355
|
+
|
356
|
+
A =2.0 0.0 8.0 6.0 0.0
|
357
|
+
1.0 6.0 0.0 1.0 7.0
|
358
|
+
5.0 0.0 7.0 4.0 0.0
|
359
|
+
7.0 0.0 8.0 5.0 0.0
|
360
|
+
0.0 10.0 0.0 0.0 7.0
|
361
|
+
|
362
|
+
U = -0.542255 0.0649957 0.821617 0.105747 -0.124490
|
363
|
+
-0.101812 -0.593461 -0.112552 0.788123 0.0602700
|
364
|
+
-0.524953 0.0593817 -0.212969 -0.115742 0.813724
|
365
|
+
-0.644870 0.0704063 -0.508744 -0.0599027 -0.562829
|
366
|
+
-0.0644952 -0.796930 0.0900097 -0.592195 -0.0441263
|
367
|
+
|
368
|
+
VT =-0.464617 0.0215065 -0.868509 0.000799554 -0.171349
|
369
|
+
-0.0700860 -0.759988 0.0630715 -0.601346 -0.227841
|
370
|
+
-0.735094 0.0987971 0.284009 -0.223485 0.565040
|
371
|
+
-0.484392 0.0254474 0.398866 0.332684 -0.703523
|
372
|
+
-0.0649698 -0.641520 -0.0442743 0.691201 0.323284
|
373
|
+
|
374
|
+
S =
|
375
|
+
(00) 17.91837085874625
|
376
|
+
(11) 15.17137188041607
|
377
|
+
(22) 3.5640020352605677
|
378
|
+
(33) 1.9842281528992616
|
379
|
+
(44) 0.3495556671751232
|
380
|
+
|
381
|
+
|
382
|
+
|
383
|
+
|
384
|
+
|
385
|
+
|
386
|
+
|
387
|
+
Dot product (or inner product of scalar product) of 2 vectors
|
388
|
+
|
389
|
+
>>> import scipy as sp
|
390
|
+
>>> x = sp.array([1,-2, .0])
|
391
|
+
>>> y = sp.array([2, 5, -1.0])
|
392
|
+
>>> dotproduct = sp.dot(x,y)
|
393
|
+
>>> print dotproduct
|
394
|
+
-8.0
|
395
|
+
Matrix product.
|
396
|
+
|
397
|
+
>>> import scipy as sp
|
398
|
+
>>> A = sp.array([[3.2, -1, 2],[2,-2,4],[1.5,-1,-4]])
|
399
|
+
>>> B = sp.array([[1., 2.],[-1., 2.0],[2, -1.5]])
|
400
|
+
|
401
|
+
>>> Matrixproduct = sp.dot(A,B)
|
402
|
+
>>> print Matrixproduct
|
403
|
+
[[ 8.2, 1.4]
|
404
|
+
[ 12., -6. ]
|
405
|
+
[ -5.5, 7. ]]
|
406
|
+
Matrix-vector product
|
407
|
+
|
408
|
+
>>> import scipy as sp
|
409
|
+
>>> x = sp.array([1,-2, .0])
|
410
|
+
>>> A = sp.array([[3.2, -1, 2],[2,-2,4],[1.5,-1,-4]])
|
411
|
+
>>> b = sp.dot(A,x)
|
412
|
+
>>> print b
|
413
|
+
[ 5.2, 6., 3.5]
|
414
|
+
Diagonal, transpose and trace
|
415
|
+
|
416
|
+
>>> print sp.diagonal(A) #returns the diagonal of A
|
417
|
+
array([ 3.2, -2., -4.])
|
418
|
+
>>> D = sp.diag((1, 2, 3)) # return a diagonal matrix
|
419
|
+
[[1 0 0]
|
420
|
+
[0 2 0]
|
421
|
+
[0 0 3]]
|
422
|
+
>>> print sp.transpose(A)
|
423
|
+
[[ 3.2 2. 1.5]
|
424
|
+
[-1. -2. -1. ]
|
425
|
+
[ 2. 4. -4. ]]
|
426
|
+
>>> print sp.trace(A)
|
427
|
+
-2.7999999999999998
|
428
|
+
Determinant and inverse. To compute the determinant or the inverse of a matrix, we need the numpy linear algebra submodule linalg
|
429
|
+
|
430
|
+
>>> import numpy
|
431
|
+
>>> import numpy.linalg
|
432
|
+
>>> A = numpy.array([[3.2, -1, 2],[2,-2,4],[1.5,-1,-4]])
|
433
|
+
>>> determinant = numpy.linalg.det(A)
|
434
|
+
>>> print determinant
|
435
|
+
26.4
|
436
|
+
>>> inverse = numpy.linalg.inv(A)
|
437
|
+
[[ 0.45454545 -0.22727273 0. ]
|
438
|
+
[ 0.53030303 -0.59848485 -0.33333333]
|
439
|
+
[ 0.03787879 0.06439394 -0.16666667]]
|
440
|
+
1.2 Eigenvalue and Eigenvectors
|
441
|
+
|
442
|
+
|
443
|
+
>>> import numpy as np
|
444
|
+
>>> import np.linalg
|
445
|
+
>>> A = np.array([[3.2, -1, 2],[2,-2,4],[1.5,-1,-4]])
|
446
|
+
>>> e_values, e_vectors = np.linalg.eig(A) #returns a list e_values of eigenvalues and a matrix e_vector whose row i corresponds
|
447
|
+
#to the eigenvectors associated with eigenvalue i
|
448
|
+
>>> print e_values
|
449
|
+
[ 2.965908+0.j, -2.882954+0.76793809j, -2.882954-0.76793809j ]
|
450
|
+
>>> print e_vectors # row i corresponds to the eigenvector
|
451
|
+
# associated with eigenvalues i
|
452
|
+
[[ 0.88158576+0.j, -0.25687900+0.02847007j, -0.25687900-0.02847007j]
|
453
|
+
[ 0.45531661+0.j, -0.89064266+0.j, -0.89064266+0.j ]
|
454
|
+
[ 0.12447222+0.j, 0.32503863-0.18522464j, 0.32503863+0.18522464j]]
|
455
|
+
1.3 Matrix factorization
|
456
|
+
|
457
|
+
LU factorization (LU_Decomposition.py).
|
458
|
+
Given a matrix A, A can be written as A = PLU where L lower triangular matrix U upper triangular matrix P is the matrix whose row i is a permutation of the identity matrix row i
|
459
|
+
LU_Decomposition.py
|
460
|
+
"""
|
461
|
+
LU decomposition: M = PLU
|
462
|
+
"""
|
463
|
+
import scipy
|
464
|
+
import scipy.linalg
|
465
|
+
|
466
|
+
if __name__ == "__main__":
|
467
|
+
M = scipy.array([ [1,-2, 3], [0,2,-1], [1,3,-2] ])
|
468
|
+
P, L, U = scipy.linalg.lu(M)
|
469
|
+
|
470
|
+
print "P = ",P
|
471
|
+
print "L = ",L
|
472
|
+
print 'U = ', U
|
473
|
+
QR factorization (QR_Decomposition.py).
|
474
|
+
Given a mxn matrix M, find a mxm unitary matrix Q - that is where is the adjoint of Q - and a mxn upper triangular matrix R such that M = QR.
|
475
|
+
"""
|
476
|
+
QR decomposition: M = QR
|
477
|
+
"""
|
478
|
+
import scipy
|
479
|
+
import scipy.linalg
|
480
|
+
|
481
|
+
if __name__ == "__main__":
|
482
|
+
M = scipy.array([ [1,-2,3], [2,1,-1], [1,0,2], [0,-2,-1]])
|
483
|
+
Q, R = scipy.linalg.qr(M)
|
484
|
+
|
485
|
+
print "Q = ",Q
|
486
|
+
print "R = ", R
|
487
|
+
Singular values decomposition:(SVD.py)
|
488
|
+
For a given mxn matrix A. and are Hermitians - a matrix H is Hermitian if it is equal to its adjoint, i.e. - so their eigenvalues are all real positive numbers. Further there are at most Min(m,n) non-zero identical eigenvalues for and . The square roots of these are called singular values. A can be decomposed as where U is a mxm matrix of eigenvectors of , V is a nxn matrix of eigenvectors of and is mxn diagonal matrix whose entries are the singular values. The command linalg.svd will return U, V, and a list of singular values. To obtain the matrix use linalg.diagsvd.
|
489
|
+
|
490
|
+
"""
|
491
|
+
SVD, A = U \Sigma V
|
492
|
+
"""
|
493
|
+
import scipy
|
494
|
+
import scipy.linalg
|
495
|
+
if __name__ == "__main__":
|
496
|
+
A = scipy.array([[1,-2],[2,0],[-1,3]])
|
497
|
+
U,s,V = scipy.linalg.svd(A) #s is the list os singular values
|
498
|
+
|
499
|
+
print "U = ",U
|
500
|
+
print "V = ", V
|
501
|
+
print "singular values list = ",s
|
502
|
+
Sigma = scipy.mat(scipy.linalg.diagsvd(s,3,2))
|
503
|
+
print "Sigma = ", Sigma
|
504
|
+
Cholesky factorization (CholeskyDecomposition.py): Given a Hermitian matrix M. Find a decomposition as M = where U is a upper triangular matrix and is the adjoint of U
|
505
|
+
"""
|
506
|
+
Cholesky decomposition. M = VU where
|
507
|
+
U is a upper triangular matrix and V is the adjoint of U
|
508
|
+
"""
|
509
|
+
import scipy
|
510
|
+
import scipy.linalg
|
511
|
+
|
512
|
+
if __name__ == "__main__":
|
513
|
+
M = scipy.array([[1,0,-1],[1,2,1],[0,-1,2]])
|
514
|
+
U = scipy.linalg.cholesky(M)
|
515
|
+
|
516
|
+
print "U = ", U
|
517
|
+
Schur factorization (SchurDecomposition.py): For a square nxn matrix M, find a unitary matrix Z and a upper-triangular (or quasi-triangular) matrix T such that
|
518
|
+
"""
|
519
|
+
Schur decomposition: M = ZTZh, where Zh is the adjoint of Z
|
520
|
+
"""
|
521
|
+
import scipy
|
522
|
+
import scipy.linalg
|
523
|
+
|
524
|
+
if __name__ == "__main__":
|
525
|
+
M = scipy.array([[1,0,-1],[1,2,1],[0,-1,2]])
|
526
|
+
Z, T = scipy.linalg.schur(M)
|
527
|
+
|
528
|
+
print "Z = ", Z
|
529
|
+
print 'T = ', T
|
530
|
+
|
531
|
+
=end
|