summaryrefslogtreecommitdiff
path: root/jni/ruby/lib/matrix/lup_decomposition.rb
blob: 30f3276253545bc4a928ac1ea4ce535b2d2ac069 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
class Matrix
  # Adapted from JAMA: http://math.nist.gov/javanumerics/jama/

  #
  # For an m-by-n matrix A with m >= n, the LU decomposition is an m-by-n
  # unit lower triangular matrix L, an n-by-n upper triangular matrix U,
  # and a m-by-m permutation matrix P so that L*U = P*A.
  # If m < n, then L is m-by-m and U is m-by-n.
  #
  # The LUP decomposition with pivoting always exists, even if the matrix is
  # singular, so the constructor will never fail.  The primary use of the
  # LU decomposition is in the solution of square systems of simultaneous
  # linear equations.  This will fail if singular? returns true.
  #

  class LUPDecomposition
    # Returns the lower triangular factor +L+

    include Matrix::ConversionHelper

    def l
      Matrix.build(@row_count, [@column_count, @row_count].min) do |i, j|
        if (i > j)
          @lu[i][j]
        elsif (i == j)
          1
        else
          0
        end
      end
    end

    # Returns the upper triangular factor +U+

    def u
      Matrix.build([@column_count, @row_count].min, @column_count) do |i, j|
        if (i <= j)
          @lu[i][j]
        else
          0
        end
      end
    end

    # Returns the permutation matrix +P+

    def p
      rows = Array.new(@row_count){Array.new(@row_count, 0)}
      @pivots.each_with_index{|p, i| rows[i][p] = 1}
      Matrix.send :new, rows, @row_count
    end

    # Returns +L+, +U+, +P+ in an array

    def to_ary
      [l, u, p]
    end
    alias_method :to_a, :to_ary

    # Returns the pivoting indices

    attr_reader :pivots

    # Returns +true+ if +U+, and hence +A+, is singular.

    def singular? ()
      @column_count.times do |j|
        if (@lu[j][j] == 0)
          return true
        end
      end
      false
    end

    # Returns the determinant of +A+, calculated efficiently
    # from the factorization.

    def det
      if (@row_count != @column_count)
        Matrix.Raise Matrix::ErrDimensionMismatch
      end
      d = @pivot_sign
      @column_count.times do |j|
        d *= @lu[j][j]
      end
      d
    end
    alias_method :determinant, :det

    # Returns +m+ so that <tt>A*m = b</tt>,
    # or equivalently so that <tt>L*U*m = P*b</tt>
    # +b+ can be a Matrix or a Vector

    def solve b
      if (singular?)
        Matrix.Raise Matrix::ErrNotRegular, "Matrix is singular."
      end
      if b.is_a? Matrix
        if (b.row_count != @row_count)
          Matrix.Raise Matrix::ErrDimensionMismatch
        end

        # Copy right hand side with pivoting
        nx = b.column_count
        m = @pivots.map{|row| b.row(row).to_a}

        # Solve L*Y = P*b
        @column_count.times do |k|
          (k+1).upto(@column_count-1) do |i|
            nx.times do |j|
              m[i][j] -= m[k][j]*@lu[i][k]
            end
          end
        end
        # Solve U*m = Y
        (@column_count-1).downto(0) do |k|
          nx.times do |j|
            m[k][j] = m[k][j].quo(@lu[k][k])
          end
          k.times do |i|
            nx.times do |j|
              m[i][j] -= m[k][j]*@lu[i][k]
            end
          end
        end
        Matrix.send :new, m, nx
      else # same algorithm, specialized for simpler case of a vector
        b = convert_to_array(b)
        if (b.size != @row_count)
          Matrix.Raise Matrix::ErrDimensionMismatch
        end

        # Copy right hand side with pivoting
        m = b.values_at(*@pivots)

        # Solve L*Y = P*b
        @column_count.times do |k|
          (k+1).upto(@column_count-1) do |i|
            m[i] -= m[k]*@lu[i][k]
          end
        end
        # Solve U*m = Y
        (@column_count-1).downto(0) do |k|
          m[k] = m[k].quo(@lu[k][k])
          k.times do |i|
            m[i] -= m[k]*@lu[i][k]
          end
        end
        Vector.elements(m, false)
      end
    end

    def initialize a
      raise TypeError, "Expected Matrix but got #{a.class}" unless a.is_a?(Matrix)
      # Use a "left-looking", dot-product, Crout/Doolittle algorithm.
      @lu = a.to_a
      @row_count = a.row_count
      @column_count = a.column_count
      @pivots = Array.new(@row_count)
      @row_count.times do |i|
         @pivots[i] = i
      end
      @pivot_sign = 1
      lu_col_j = Array.new(@row_count)

      # Outer loop.

      @column_count.times do |j|

        # Make a copy of the j-th column to localize references.

        @row_count.times do |i|
          lu_col_j[i] = @lu[i][j]
        end

        # Apply previous transformations.

        @row_count.times do |i|
          lu_row_i = @lu[i]

          # Most of the time is spent in the following dot product.

          kmax = [i, j].min
          s = 0
          kmax.times do |k|
            s += lu_row_i[k]*lu_col_j[k]
          end

          lu_row_i[j] = lu_col_j[i] -= s
        end

        # Find pivot and exchange if necessary.

        p = j
        (j+1).upto(@row_count-1) do |i|
          if (lu_col_j[i].abs > lu_col_j[p].abs)
            p = i
          end
        end
        if (p != j)
          @column_count.times do |k|
            t = @lu[p][k]; @lu[p][k] = @lu[j][k]; @lu[j][k] = t
          end
          k = @pivots[p]; @pivots[p] = @pivots[j]; @pivots[j] = k
          @pivot_sign = -@pivot_sign
        end

        # Compute multipliers.

        if (j < @row_count && @lu[j][j] != 0)
          (j+1).upto(@row_count-1) do |i|
            @lu[i][j] = @lu[i][j].quo(@lu[j][j])
          end
        end
      end
    end
  end
end