line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Math::Polynomial::Solve; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
require 5.010001; |
4
|
17
|
|
|
17
|
|
907412
|
use strict; |
|
17
|
|
|
|
|
175
|
|
|
17
|
|
|
|
|
469
|
|
5
|
17
|
|
|
17
|
|
79
|
use warnings; |
|
17
|
|
|
|
|
28
|
|
|
17
|
|
|
|
|
360
|
|
6
|
17
|
|
|
17
|
|
5584
|
use utf8; |
|
17
|
|
|
|
|
202
|
|
|
17
|
|
|
|
|
70
|
|
7
|
17
|
|
|
17
|
|
477
|
use Carp; |
|
17
|
|
|
|
|
29
|
|
|
17
|
|
|
|
|
786
|
|
8
|
|
|
|
|
|
|
|
9
|
17
|
|
|
17
|
|
2171
|
use Math::Complex; |
|
17
|
|
|
|
|
54549
|
|
|
17
|
|
|
|
|
2289
|
|
10
|
17
|
|
|
17
|
|
4640
|
use Math::Utils qw(:polynomial :utility); |
|
17
|
|
|
|
|
33214
|
|
|
17
|
|
|
|
|
2850
|
|
11
|
17
|
|
|
17
|
|
112
|
use Exporter; |
|
17
|
|
|
|
|
27
|
|
|
17
|
|
|
|
|
3829
|
|
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
our $VERSION = '2.81'; |
14
|
|
|
|
|
|
|
our @ISA = qw(Exporter); |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
# |
17
|
|
|
|
|
|
|
# Three # for "I am here" messages. |
18
|
|
|
|
|
|
|
# Four # for variable dumps. |
19
|
|
|
|
|
|
|
# Five # for a dump of the companion matrix. |
20
|
|
|
|
|
|
|
# Six # for sturm structs (sign chain, etc). |
21
|
|
|
|
|
|
|
# |
22
|
|
|
|
|
|
|
#use Smart::Comments q(######); |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
# |
25
|
|
|
|
|
|
|
# Export only on request. |
26
|
|
|
|
|
|
|
# |
27
|
|
|
|
|
|
|
our %EXPORT_TAGS = ( |
28
|
|
|
|
|
|
|
'classical' => [ qw( |
29
|
|
|
|
|
|
|
linear_roots |
30
|
|
|
|
|
|
|
quadratic_roots |
31
|
|
|
|
|
|
|
cubic_roots |
32
|
|
|
|
|
|
|
quartic_roots |
33
|
|
|
|
|
|
|
) ], |
34
|
|
|
|
|
|
|
'numeric' => [ qw(poly_roots |
35
|
|
|
|
|
|
|
poly_option |
36
|
|
|
|
|
|
|
build_companion |
37
|
|
|
|
|
|
|
balance_matrix |
38
|
|
|
|
|
|
|
hqr_eigen_hessenberg |
39
|
|
|
|
|
|
|
) ], |
40
|
|
|
|
|
|
|
'sturm' => [ qw( |
41
|
|
|
|
|
|
|
poly_real_root_count |
42
|
|
|
|
|
|
|
poly_sturm_chain |
43
|
|
|
|
|
|
|
sturm_real_root_range_count |
44
|
|
|
|
|
|
|
sturm_bisection |
45
|
|
|
|
|
|
|
sturm_bisection_roots |
46
|
|
|
|
|
|
|
sturm_sign_count |
47
|
|
|
|
|
|
|
sturm_sign_chain |
48
|
|
|
|
|
|
|
sturm_sign_minus_inf |
49
|
|
|
|
|
|
|
sturm_sign_plus_inf |
50
|
|
|
|
|
|
|
) ], |
51
|
|
|
|
|
|
|
'utility' => [ qw( |
52
|
|
|
|
|
|
|
epsilon |
53
|
|
|
|
|
|
|
laguerre |
54
|
|
|
|
|
|
|
newtonraphson |
55
|
|
|
|
|
|
|
poly_iteration |
56
|
|
|
|
|
|
|
poly_nonzero_term_count |
57
|
|
|
|
|
|
|
poly_tolerance |
58
|
|
|
|
|
|
|
) ], |
59
|
|
|
|
|
|
|
); |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
our @EXPORT_OK = ( |
62
|
|
|
|
|
|
|
@{ $EXPORT_TAGS{'classical'} }, |
63
|
|
|
|
|
|
|
@{ $EXPORT_TAGS{'numeric'} }, |
64
|
|
|
|
|
|
|
@{ $EXPORT_TAGS{'sturm'} }, |
65
|
|
|
|
|
|
|
@{ $EXPORT_TAGS{'utility'} } ); |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
our @EXPORT = qw( ascending_order ); |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
# |
70
|
|
|
|
|
|
|
# Add an :all tag automatically. |
71
|
|
|
|
|
|
|
# |
72
|
|
|
|
|
|
|
$EXPORT_TAGS{all} = [@EXPORT_OK, @EXPORT]; |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
# |
75
|
|
|
|
|
|
|
# Options to set or unset to force poly_roots() to use different |
76
|
|
|
|
|
|
|
# methods of solving. |
77
|
|
|
|
|
|
|
# |
78
|
|
|
|
|
|
|
# hessenberg (default 1): set to 1 to force poly_roots() to use the QR |
79
|
|
|
|
|
|
|
# Hessenberg method regardless of the degree of the polynomial. Set to zero |
80
|
|
|
|
|
|
|
# to force poly_roots() uses one of the specialized routines (linerar_roots(), |
81
|
|
|
|
|
|
|
# quadratic_roots(), etc) if the degree of the polynomial is less than five. |
82
|
|
|
|
|
|
|
# |
83
|
|
|
|
|
|
|
# root_function (default 0): set to 1 to force poly_roots() to use |
84
|
|
|
|
|
|
|
# Math::Complex's root(-c/a, n) function if the polynomial is of the form |
85
|
|
|
|
|
|
|
# ax**n + c. |
86
|
|
|
|
|
|
|
# |
87
|
|
|
|
|
|
|
# varsubst (default 0): try to reduce the degree of the polynomial through |
88
|
|
|
|
|
|
|
# variable substitution before solving. |
89
|
|
|
|
|
|
|
# |
90
|
|
|
|
|
|
|
my %option = ( |
91
|
|
|
|
|
|
|
hessenberg => 1, |
92
|
|
|
|
|
|
|
root_function => 0, |
93
|
|
|
|
|
|
|
varsubst => 0, |
94
|
|
|
|
|
|
|
); |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
# |
97
|
|
|
|
|
|
|
# Iteration limits. The Hessenberg matrix method and the Laguerre method run |
98
|
|
|
|
|
|
|
# continuously until they converge upon an answer. The iteration limits are |
99
|
|
|
|
|
|
|
# there to prevent the loops from running forever if they fail to converge. |
100
|
|
|
|
|
|
|
# |
101
|
|
|
|
|
|
|
my %iteration = ( |
102
|
|
|
|
|
|
|
hessenberg => 60, |
103
|
|
|
|
|
|
|
newtonraphson => 60, |
104
|
|
|
|
|
|
|
laguerre => 60, |
105
|
|
|
|
|
|
|
sturm_bisection => 100, |
106
|
|
|
|
|
|
|
); |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
# |
109
|
|
|
|
|
|
|
# Some values here are placeholders only, and will get |
110
|
|
|
|
|
|
|
# replaced in the BEGIN block. |
111
|
|
|
|
|
|
|
# |
112
|
|
|
|
|
|
|
my %tolerance = ( |
113
|
|
|
|
|
|
|
newtonraphson => 1e-14, |
114
|
|
|
|
|
|
|
laguerre => 1e-14, |
115
|
|
|
|
|
|
|
); |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
# |
118
|
|
|
|
|
|
|
# Set up the epsilon variable, the value that is, in the floating-point |
119
|
|
|
|
|
|
|
# math of the computer, the smallest value a variable can have before |
120
|
|
|
|
|
|
|
# it is indistinguishable from zero when adding it to one. |
121
|
|
|
|
|
|
|
# |
122
|
|
|
|
|
|
|
my $epsilon; |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
BEGIN |
125
|
|
|
|
|
|
|
{ |
126
|
17
|
|
|
17
|
|
63
|
$epsilon = 0.125; |
127
|
17
|
|
|
|
|
45
|
my $epsilon2 = $epsilon/2.0; |
128
|
|
|
|
|
|
|
|
129
|
17
|
|
|
|
|
70
|
while (1.0 + $epsilon2 > 1.0) |
130
|
|
|
|
|
|
|
{ |
131
|
833
|
|
|
|
|
921
|
$epsilon = $epsilon2; |
132
|
833
|
|
|
|
|
1218
|
$epsilon2 /= 2.0; |
133
|
|
|
|
|
|
|
} |
134
|
|
|
|
|
|
|
|
135
|
17
|
|
|
|
|
57
|
$tolerance{laguerre} = 2 * $epsilon; |
136
|
17
|
|
|
|
|
57563
|
$tolerance{newtonraphson} = 2 * $epsilon; |
137
|
|
|
|
|
|
|
} |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
# |
140
|
|
|
|
|
|
|
# Flag to determine whether calling order is |
141
|
|
|
|
|
|
|
# ($an_1, $an_2, $an_3, ...) or if it is |
142
|
|
|
|
|
|
|
# ($a0, $a1, $a2, $a3, ...) |
143
|
|
|
|
|
|
|
# |
144
|
|
|
|
|
|
|
my $ascending_flag = 0; # default 0, in a later version it will be 1. |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
# |
147
|
|
|
|
|
|
|
# See the END block. |
148
|
|
|
|
|
|
|
# |
149
|
|
|
|
|
|
|
my $ascending_order_called = 0; |
150
|
|
|
|
|
|
|
my %called_by; |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
=pod |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
=encoding UTF-8 |
155
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
=head1 NAME |
157
|
|
|
|
|
|
|
|
158
|
|
|
|
|
|
|
Math::Polynomial::Solve - Find the roots of polynomial equations. |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
=head1 SYNOPSIS |
161
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
use Math::Complex; # The roots may be complex numbers. |
163
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(poly_roots); |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
my @x = poly_roots(@coefficients); |
166
|
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
or |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
use Math::Complex; # The roots may be complex numbers. |
170
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(:numeric); # See the EXPORT section |
171
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
# |
173
|
|
|
|
|
|
|
# Find roots using the matrix method. |
174
|
|
|
|
|
|
|
# |
175
|
|
|
|
|
|
|
my @x = poly_roots(@coefficients); |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
# |
178
|
|
|
|
|
|
|
# Alternatively, use the classical methods instead of the matrix |
179
|
|
|
|
|
|
|
# method if the polynomial degree is less than five. |
180
|
|
|
|
|
|
|
# |
181
|
|
|
|
|
|
|
poly_option(hessenberg => 0); |
182
|
|
|
|
|
|
|
@x = poly_roots(@coefficients); |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
or |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
use Math::Complex; # The roots may be complex numbers. |
187
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(:classical); # See the EXPORT section |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
# |
190
|
|
|
|
|
|
|
# Find the polynomial roots using the classical methods. |
191
|
|
|
|
|
|
|
# |
192
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
# Find the roots of ax + b |
194
|
|
|
|
|
|
|
my @x1 = linear_roots($a, $b); |
195
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
# Find the roots of ax**2 + bx +c |
197
|
|
|
|
|
|
|
my @x2 = quadratic_roots($a, $b, $c); |
198
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
# Find the roots of ax**3 + bx**2 +cx + d |
200
|
|
|
|
|
|
|
my @x3 = cubic_roots($a, $b, $c, $d); |
201
|
|
|
|
|
|
|
|
202
|
|
|
|
|
|
|
# Find the roots of ax**4 + bx**3 +cx**2 + dx + e |
203
|
|
|
|
|
|
|
my @x4 = quartic_roots($a, $b, $c, $d, $e); |
204
|
|
|
|
|
|
|
|
205
|
|
|
|
|
|
|
or |
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
use Math::Complex; # The roots may be complex numbers. |
208
|
|
|
|
|
|
|
use Math::Polynomial; |
209
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(:classical ascending_order); |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
ascending_order(1); # Change default coefficient order for M::P::S. |
212
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
# |
214
|
|
|
|
|
|
|
# Form 8*x**3 - 6*x - 1 |
215
|
|
|
|
|
|
|
# |
216
|
|
|
|
|
|
|
my $p1 = Math::Polynomial->new(-1, -6, 0, 8); |
217
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
# |
219
|
|
|
|
|
|
|
# Use Math::Polynomial's coefficient order. |
220
|
|
|
|
|
|
|
# If ascending_order() had not been called, |
221
|
|
|
|
|
|
|
# the statement would be: |
222
|
|
|
|
|
|
|
# |
223
|
|
|
|
|
|
|
# my @roots = poly_roots(reverse $p1->coefficients); |
224
|
|
|
|
|
|
|
# |
225
|
|
|
|
|
|
|
my @roots = poly_roots($p1->coefficients); |
226
|
|
|
|
|
|
|
|
227
|
|
|
|
|
|
|
or |
228
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(:sturm); |
230
|
|
|
|
|
|
|
|
231
|
|
|
|
|
|
|
# |
232
|
|
|
|
|
|
|
# Find the number of unique real roots of the polynomial. |
233
|
|
|
|
|
|
|
# |
234
|
|
|
|
|
|
|
my $no_of_unique_roots = poly_real_root_count(@coefficients); |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
=head1 DESCRIPTION |
237
|
|
|
|
|
|
|
|
238
|
|
|
|
|
|
|
This package supplies a set of functions that find the roots of |
239
|
|
|
|
|
|
|
polynomials, along with some utility functions. |
240
|
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
Roots will be either real or of type L. |
242
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
Functions making use of the Sturm sequence are also available, letting you |
244
|
|
|
|
|
|
|
find the number of real roots present in a range of X values. |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
In addition to the root-finding functions, the internal functions have |
247
|
|
|
|
|
|
|
also been exported for your use. |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
=cut |
250
|
|
|
|
|
|
|
|
251
|
|
|
|
|
|
|
# |
252
|
|
|
|
|
|
|
# $asending = ascending_order(); |
253
|
|
|
|
|
|
|
# $oldorder = ascending_order($neworder); |
254
|
|
|
|
|
|
|
# |
255
|
|
|
|
|
|
|
# |
256
|
|
|
|
|
|
|
sub ascending_order |
257
|
|
|
|
|
|
|
{ |
258
|
14
|
|
|
14
|
1
|
1236
|
my $ascend = $ascending_flag; |
259
|
|
|
|
|
|
|
|
260
|
14
|
50
|
|
|
|
66
|
if (scalar @_ > 0) |
261
|
|
|
|
|
|
|
{ |
262
|
14
|
50
|
|
|
|
53
|
$ascending_flag = ($_[0] == 0)? 0: 1; |
263
|
14
|
|
|
|
|
27
|
$ascending_order_called = 1; |
264
|
|
|
|
|
|
|
} |
265
|
|
|
|
|
|
|
|
266
|
14
|
|
|
|
|
30
|
return $ascend; |
267
|
|
|
|
|
|
|
} |
268
|
|
|
|
|
|
|
|
269
|
|
|
|
|
|
|
# |
270
|
|
|
|
|
|
|
# ($new_coefficients_ref, $varsubst) = poly_analysis(@coefficients); |
271
|
|
|
|
|
|
|
# |
272
|
|
|
|
|
|
|
# If the polynomial has evenly spaced gaps of zero coefficients, reduce |
273
|
|
|
|
|
|
|
# the polynomial through variable substitution. |
274
|
|
|
|
|
|
|
# |
275
|
|
|
|
|
|
|
# For example, a degree-6 polynomial like 9x**6 + 128x**3 + 7 |
276
|
|
|
|
|
|
|
# can be reduced to a polynomial 9y**2 + 128y + 7, where y = x**3. |
277
|
|
|
|
|
|
|
# |
278
|
|
|
|
|
|
|
# After solving a quadratic instead of a sextic, the actual roots of |
279
|
|
|
|
|
|
|
# the original equation are found by taking the cube roots of each |
280
|
|
|
|
|
|
|
# root of the quadratic. |
281
|
|
|
|
|
|
|
# |
282
|
|
|
|
|
|
|
# Not exported. Coefficients are always in ascending order. |
283
|
|
|
|
|
|
|
sub poly_analysis |
284
|
|
|
|
|
|
|
{ |
285
|
44
|
|
|
44
|
0
|
84
|
my(@coefficients) = @_; |
286
|
44
|
|
|
|
|
57
|
my @czp; |
287
|
44
|
|
|
|
|
58
|
my $m = 1; |
288
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
# |
290
|
|
|
|
|
|
|
# Is the count of coefficients a multiple of any of the primes? |
291
|
|
|
|
|
|
|
# |
292
|
|
|
|
|
|
|
# Realistically I don't expect any gaps that can't be handled by |
293
|
|
|
|
|
|
|
# the first three prime numbers, but it's not much of a waste of |
294
|
|
|
|
|
|
|
# space to check the first dozen. |
295
|
|
|
|
|
|
|
# |
296
|
44
|
|
|
|
|
185
|
@czp = grep(($#coefficients % $_) == 0, |
297
|
|
|
|
|
|
|
(2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37) |
298
|
|
|
|
|
|
|
); |
299
|
|
|
|
|
|
|
|
300
|
|
|
|
|
|
|
# |
301
|
|
|
|
|
|
|
# Any coefficients zero at the non-N degrees? (1==T,0==F). |
302
|
|
|
|
|
|
|
# |
303
|
|
|
|
|
|
|
#### @czp |
304
|
|
|
|
|
|
|
# |
305
|
44
|
100
|
|
|
|
84
|
if (@czp) |
306
|
|
|
|
|
|
|
{ |
307
|
30
|
|
|
|
|
58
|
for my $j (1..$#coefficients - 1) |
308
|
|
|
|
|
|
|
{ |
309
|
106
|
100
|
|
|
|
187
|
if (abs($coefficients[$j]) > $epsilon) |
310
|
|
|
|
|
|
|
{ |
311
|
20
|
|
|
|
|
44
|
@czp = grep(($j % $_) == 0, @czp); |
312
|
|
|
|
|
|
|
} |
313
|
|
|
|
|
|
|
} |
314
|
|
|
|
|
|
|
|
315
|
|
|
|
|
|
|
# |
316
|
|
|
|
|
|
|
# The remaining list of primes represent the gap size |
317
|
|
|
|
|
|
|
# between non-zero coefficients. |
318
|
|
|
|
|
|
|
# |
319
|
30
|
|
|
|
|
65
|
map(($m *= $_), @czp); |
320
|
|
|
|
|
|
|
|
321
|
|
|
|
|
|
|
#### Substitution degree: $m |
322
|
|
|
|
|
|
|
} |
323
|
|
|
|
|
|
|
|
324
|
|
|
|
|
|
|
# |
325
|
|
|
|
|
|
|
# If there's a sequence of zero-filled gaps in the coefficients, |
326
|
|
|
|
|
|
|
# reduce the polynomial by degree $m and check again for the |
327
|
|
|
|
|
|
|
# next round of factors (e.g., X**8 + X**4 + 1 needs two rounds |
328
|
|
|
|
|
|
|
# to get to a factor of 4). |
329
|
|
|
|
|
|
|
# |
330
|
44
|
100
|
|
|
|
85
|
if ($m > 1) |
331
|
|
|
|
|
|
|
{ |
332
|
22
|
|
|
|
|
38
|
my @alt_coefs; |
333
|
22
|
|
|
|
|
85
|
push @alt_coefs, $coefficients[$_*$m] for (0..$#coefficients/$m); |
334
|
22
|
|
|
|
|
55
|
my($cf, $m1) = poly_analysis(@alt_coefs); |
335
|
22
|
|
|
|
|
46
|
@coefficients = @$cf; |
336
|
22
|
|
|
|
|
42
|
$m *= $m1; |
337
|
|
|
|
|
|
|
} |
338
|
|
|
|
|
|
|
|
339
|
44
|
|
|
|
|
100
|
return \@coefficients, $m; |
340
|
|
|
|
|
|
|
} |
341
|
|
|
|
|
|
|
|
342
|
|
|
|
|
|
|
=head1 EXPORT |
343
|
|
|
|
|
|
|
|
344
|
|
|
|
|
|
|
There is an B tag that exports everything. |
345
|
|
|
|
|
|
|
|
346
|
|
|
|
|
|
|
Currently there is one default export, L. |
347
|
|
|
|
|
|
|
|
348
|
|
|
|
|
|
|
If you want to have more fine-grained control you may |
349
|
|
|
|
|
|
|
individually name the functions in an export list, or |
350
|
|
|
|
|
|
|
use one four export tags: |
351
|
|
|
|
|
|
|
|
352
|
|
|
|
|
|
|
L, |
353
|
|
|
|
|
|
|
L, |
354
|
|
|
|
|
|
|
L, |
355
|
|
|
|
|
|
|
L, |
356
|
|
|
|
|
|
|
|
357
|
|
|
|
|
|
|
=head2 EXPORTED BY DEFAULT |
358
|
|
|
|
|
|
|
|
359
|
|
|
|
|
|
|
=head3 ascending_order() |
360
|
|
|
|
|
|
|
|
361
|
|
|
|
|
|
|
Changes the default order of the coefficents to the functions. |
362
|
|
|
|
|
|
|
|
363
|
|
|
|
|
|
|
When Math::Polynomial::Solve was originally written, it followed the |
364
|
|
|
|
|
|
|
calling convention of L, using the highest degree |
365
|
|
|
|
|
|
|
coefficient, followed by the next highest degree coefficient, and so |
366
|
|
|
|
|
|
|
on in descending order. |
367
|
|
|
|
|
|
|
|
368
|
|
|
|
|
|
|
Later Math::Polynomial was re-written, and the order of the coefficients were |
369
|
|
|
|
|
|
|
put in ascending order, e.g.: |
370
|
|
|
|
|
|
|
|
371
|
|
|
|
|
|
|
use Math::Polynomial; |
372
|
|
|
|
|
|
|
|
373
|
|
|
|
|
|
|
# |
374
|
|
|
|
|
|
|
# Create the polynomial 8*x**3 - 6*x - 1. |
375
|
|
|
|
|
|
|
# |
376
|
|
|
|
|
|
|
$fpx = Math::Polynomial->new(-1, -6, 0, 8); |
377
|
|
|
|
|
|
|
|
378
|
|
|
|
|
|
|
If you use Math::Polynomial with this module, it will probably be |
379
|
|
|
|
|
|
|
more convenient to change the default parameter list of |
380
|
|
|
|
|
|
|
Math::Polynomial::Solve's functions, using the ascending_order() function: |
381
|
|
|
|
|
|
|
|
382
|
|
|
|
|
|
|
use Math::Polynomial; |
383
|
|
|
|
|
|
|
use Math::Polynomial::Solve qw(:all); |
384
|
|
|
|
|
|
|
|
385
|
|
|
|
|
|
|
ascending_order(1); |
386
|
|
|
|
|
|
|
|
387
|
|
|
|
|
|
|
my $fp4 = Math::Polynomial->interpolate([1 .. 4], [14, 19, 25, 32]); |
388
|
|
|
|
|
|
|
|
389
|
|
|
|
|
|
|
# |
390
|
|
|
|
|
|
|
# Find roots of $fp4. |
391
|
|
|
|
|
|
|
# |
392
|
|
|
|
|
|
|
my @fp4_roots = quartic_roots($fp4->coefficients); |
393
|
|
|
|
|
|
|
|
394
|
|
|
|
|
|
|
or |
395
|
|
|
|
|
|
|
|
396
|
|
|
|
|
|
|
my @fp4_roots = poly_roots($fp4->coefficients); |
397
|
|
|
|
|
|
|
|
398
|
|
|
|
|
|
|
If C had not been called, the previous line of code |
399
|
|
|
|
|
|
|
would have been written instead as |
400
|
|
|
|
|
|
|
|
401
|
|
|
|
|
|
|
my @fp4_roots = poly_roots(reverse $fp4->coefficients); |
402
|
|
|
|
|
|
|
|
403
|
|
|
|
|
|
|
The function is a way to help with the change in the API when version 3.00 of |
404
|
|
|
|
|
|
|
this module is released. At that point coefficients will be in ascending |
405
|
|
|
|
|
|
|
order by default, and you will need to use C to use the |
406
|
|
|
|
|
|
|
old (current) style. |
407
|
|
|
|
|
|
|
|
408
|
|
|
|
|
|
|
=head2 Numeric Functions |
409
|
|
|
|
|
|
|
|
410
|
|
|
|
|
|
|
These are the functions that calculate the polynomial's roots through numeric |
411
|
|
|
|
|
|
|
algorithms. They are all exported under the tag "numeric". |
412
|
|
|
|
|
|
|
|
413
|
|
|
|
|
|
|
=head3 poly_roots() |
414
|
|
|
|
|
|
|
|
415
|
|
|
|
|
|
|
Returns the roots of a polynomial equation, regardless of degree. |
416
|
|
|
|
|
|
|
Unlike the other root-finding functions, it will check for coefficients |
417
|
|
|
|
|
|
|
of zero for the highest power, and 'step down' the degree of the |
418
|
|
|
|
|
|
|
polynomial to the appropriate case. Additionally, it will check for |
419
|
|
|
|
|
|
|
coefficients of zero for the lowest power terms, and add zeros to its |
420
|
|
|
|
|
|
|
root list before calling one of the root-finding functions. |
421
|
|
|
|
|
|
|
|
422
|
|
|
|
|
|
|
By default, C will use the Hessenberg matrix method for solving |
423
|
|
|
|
|
|
|
polynomials. This can be changed by calling L. |
424
|
|
|
|
|
|
|
|
425
|
|
|
|
|
|
|
The method of poly_roots() is almost equivalent to |
426
|
|
|
|
|
|
|
|
427
|
|
|
|
|
|
|
@x = hqr_eigen_hessenberg( |
428
|
|
|
|
|
|
|
balance_matrix(build_companion(@coefficients)) |
429
|
|
|
|
|
|
|
); |
430
|
|
|
|
|
|
|
|
431
|
|
|
|
|
|
|
except this wouldn't check for leading and trailing zero coefficients, and it |
432
|
|
|
|
|
|
|
ignores the settings of C. |
433
|
|
|
|
|
|
|
|
434
|
|
|
|
|
|
|
=cut |
435
|
|
|
|
|
|
|
|
436
|
|
|
|
|
|
|
sub poly_roots |
437
|
|
|
|
|
|
|
{ |
438
|
135
|
50
|
|
135
|
1
|
355655
|
my(@coefficients) = ($ascending_flag == 0)? reverse @_: @_; |
439
|
135
|
|
|
|
|
249
|
my(@x, @zero_x); |
440
|
135
|
|
|
|
|
189
|
my $subst_degree = 1; |
441
|
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
# |
443
|
|
|
|
|
|
|
#### @coefficients |
444
|
|
|
|
|
|
|
# |
445
|
|
|
|
|
|
|
# Check for zero coefficients in the higher-powered terms. |
446
|
|
|
|
|
|
|
# |
447
|
135
|
|
33
|
|
|
753
|
pop @coefficients while (scalar @coefficients and |
448
|
|
|
|
|
|
|
abs($coefficients[$#coefficients]) < $epsilon); |
449
|
|
|
|
|
|
|
|
450
|
135
|
50
|
|
|
|
339
|
if (@coefficients == 0) |
451
|
|
|
|
|
|
|
{ |
452
|
0
|
|
|
|
|
0
|
carp "All coefficients are zero\n"; |
453
|
0
|
|
|
|
|
0
|
return (0); |
454
|
|
|
|
|
|
|
} |
455
|
|
|
|
|
|
|
|
456
|
|
|
|
|
|
|
# |
457
|
|
|
|
|
|
|
# How about zero coefficients in the low terms? |
458
|
|
|
|
|
|
|
# |
459
|
135
|
|
66
|
|
|
552
|
while (scalar @coefficients and |
460
|
|
|
|
|
|
|
abs($coefficients[0]) < $epsilon) |
461
|
|
|
|
|
|
|
{ |
462
|
12
|
|
|
|
|
16
|
push @zero_x, 0; |
463
|
|
|
|
|
|
|
shift @coefficients |
464
|
12
|
|
|
|
|
33
|
} |
465
|
|
|
|
|
|
|
|
466
|
|
|
|
|
|
|
# |
467
|
|
|
|
|
|
|
# If the polynomial is of the form c + ax**n, and if the |
468
|
|
|
|
|
|
|
# root_function option is set, use the Math::Complex::root() |
469
|
|
|
|
|
|
|
# function to return the roots. |
470
|
|
|
|
|
|
|
# |
471
|
|
|
|
|
|
|
### %option |
472
|
|
|
|
|
|
|
# |
473
|
135
|
100
|
100
|
|
|
415
|
if ($option{root_function} and |
474
|
|
|
|
|
|
|
poly_nonzero_term_count(@coefficients) == 2) |
475
|
|
|
|
|
|
|
{ |
476
|
15
|
|
|
|
|
89
|
return @zero_x, |
477
|
|
|
|
|
|
|
root(-$coefficients[0]/$coefficients[$#coefficients], |
478
|
|
|
|
|
|
|
$#coefficients); |
479
|
|
|
|
|
|
|
} |
480
|
|
|
|
|
|
|
|
481
|
|
|
|
|
|
|
# |
482
|
|
|
|
|
|
|
# Next do some analysis of the coefficients. |
483
|
|
|
|
|
|
|
# See if we can reduce the size of the polynomial by |
484
|
|
|
|
|
|
|
# doing some variable substitution. |
485
|
|
|
|
|
|
|
# |
486
|
120
|
100
|
66
|
|
|
319
|
if ($option{varsubst} and $#coefficients > 1) |
487
|
|
|
|
|
|
|
{ |
488
|
22
|
|
|
|
|
34
|
my $cf; |
489
|
22
|
|
|
|
|
59
|
($cf, $subst_degree) = poly_analysis(@coefficients); |
490
|
22
|
50
|
|
|
|
70
|
@coefficients = @$cf if ($subst_degree > 1); |
491
|
|
|
|
|
|
|
} |
492
|
|
|
|
|
|
|
|
493
|
|
|
|
|
|
|
# |
494
|
|
|
|
|
|
|
# If the remaining polynomial is a quintic or higher, or |
495
|
|
|
|
|
|
|
# if $option{hessenberg} is set, continue with the matrix |
496
|
|
|
|
|
|
|
# calculation. |
497
|
|
|
|
|
|
|
# |
498
|
|
|
|
|
|
|
#### @coefficients |
499
|
|
|
|
|
|
|
#### $subst_degree |
500
|
|
|
|
|
|
|
# |
501
|
|
|
|
|
|
|
# |
502
|
|
|
|
|
|
|
# With the coefficents in ascending order, |
503
|
|
|
|
|
|
|
# pretend it was always that way for the next |
504
|
|
|
|
|
|
|
# function calls. |
505
|
|
|
|
|
|
|
# |
506
|
120
|
|
|
|
|
174
|
my $temp_ascending_flag = $ascending_flag; |
507
|
120
|
|
|
|
|
209
|
$ascending_flag = 1; |
508
|
|
|
|
|
|
|
|
509
|
120
|
100
|
66
|
|
|
431
|
if ($option{hessenberg} or $#coefficients > 4) |
|
|
100
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
510
|
|
|
|
|
|
|
{ |
511
|
|
|
|
|
|
|
# |
512
|
|
|
|
|
|
|
# QR iterations from the matrix. |
513
|
|
|
|
|
|
|
# |
514
|
87
|
|
|
|
|
218
|
@x = hqr_eigen_hessenberg( |
515
|
|
|
|
|
|
|
balance_matrix(build_companion(@coefficients)) |
516
|
|
|
|
|
|
|
); |
517
|
|
|
|
|
|
|
} |
518
|
|
|
|
|
|
|
elsif ($#coefficients == 4) |
519
|
|
|
|
|
|
|
{ |
520
|
15
|
|
|
|
|
35
|
@x = quartic_roots(@coefficients); |
521
|
|
|
|
|
|
|
} |
522
|
|
|
|
|
|
|
elsif ($#coefficients == 3) |
523
|
|
|
|
|
|
|
{ |
524
|
4
|
|
|
|
|
10
|
@x = cubic_roots(@coefficients); |
525
|
|
|
|
|
|
|
} |
526
|
|
|
|
|
|
|
elsif ($#coefficients == 2) |
527
|
|
|
|
|
|
|
{ |
528
|
5
|
|
|
|
|
18
|
@x = quadratic_roots(@coefficients); |
529
|
|
|
|
|
|
|
} |
530
|
|
|
|
|
|
|
elsif ($#coefficients == 1) |
531
|
|
|
|
|
|
|
{ |
532
|
9
|
|
|
|
|
37
|
@x = linear_roots(@coefficients); |
533
|
|
|
|
|
|
|
} |
534
|
|
|
|
|
|
|
|
535
|
120
|
|
|
|
|
5071
|
$ascending_flag = $temp_ascending_flag; |
536
|
|
|
|
|
|
|
|
537
|
120
|
100
|
|
|
|
304
|
@x = map(root($_, $subst_degree), @x) if ($subst_degree > 1); |
538
|
|
|
|
|
|
|
|
539
|
120
|
|
|
|
|
7612
|
return @zero_x, @x; |
540
|
|
|
|
|
|
|
} |
541
|
|
|
|
|
|
|
|
542
|
|
|
|
|
|
|
|
543
|
|
|
|
|
|
|
=head3 poly_option() |
544
|
|
|
|
|
|
|
|
545
|
|
|
|
|
|
|
Set options that affect the behavior of the C function. All |
546
|
|
|
|
|
|
|
options are set to either 1 ("on") or 0 ("off"). See also L |
547
|
|
|
|
|
|
|
and L. |
548
|
|
|
|
|
|
|
|
549
|
|
|
|
|
|
|
Options may be set and saved: |
550
|
|
|
|
|
|
|
|
551
|
|
|
|
|
|
|
# |
552
|
|
|
|
|
|
|
# Set a few options... |
553
|
|
|
|
|
|
|
# |
554
|
|
|
|
|
|
|
poly_option(hessenberg => 0, root_function => 1); |
555
|
|
|
|
|
|
|
|
556
|
|
|
|
|
|
|
# |
557
|
|
|
|
|
|
|
# Get all of the current options and their values. |
558
|
|
|
|
|
|
|
# |
559
|
|
|
|
|
|
|
my %all_options = poly_option(); |
560
|
|
|
|
|
|
|
|
561
|
|
|
|
|
|
|
# |
562
|
|
|
|
|
|
|
# Set some options but save the former option values |
563
|
|
|
|
|
|
|
# for later. |
564
|
|
|
|
|
|
|
# |
565
|
|
|
|
|
|
|
my %changed_options = poly_option(hessenberg => 1, varsubst => 1); |
566
|
|
|
|
|
|
|
|
567
|
|
|
|
|
|
|
The current options available are: |
568
|
|
|
|
|
|
|
|
569
|
|
|
|
|
|
|
=over 4 |
570
|
|
|
|
|
|
|
|
571
|
|
|
|
|
|
|
=item 'hessenberg' |
572
|
|
|
|
|
|
|
|
573
|
|
|
|
|
|
|
Use the QR Hessenberg matrix method to solve the polynomial. By default, this |
574
|
|
|
|
|
|
|
is set to 1. If set to 0, C uses one of the L |
575
|
|
|
|
|
|
|
root-finding functions listed below, I the degree of the polynomial is four |
576
|
|
|
|
|
|
|
or less. |
577
|
|
|
|
|
|
|
|
578
|
|
|
|
|
|
|
=item 'root_function' |
579
|
|
|
|
|
|
|
|
580
|
|
|
|
|
|
|
Use the L function from Math::Complex if the |
581
|
|
|
|
|
|
|
polynomial is of the form C. This will take precedence over the other |
582
|
|
|
|
|
|
|
solving methods. |
583
|
|
|
|
|
|
|
|
584
|
|
|
|
|
|
|
=item 'varsubst' |
585
|
|
|
|
|
|
|
|
586
|
|
|
|
|
|
|
Reduce polynomials to a lower degree through variable substitution, if possible. |
587
|
|
|
|
|
|
|
|
588
|
|
|
|
|
|
|
For example, with C set to one and the polynomial to solve being |
589
|
|
|
|
|
|
|
C<9x**6 + 128x**3 + 21>, C will reduce the polynomial to |
590
|
|
|
|
|
|
|
C<9y**2 + 128y + 21> (where C), |
591
|
|
|
|
|
|
|
and solve the quadratic (either classically or numerically, depending |
592
|
|
|
|
|
|
|
on the hessenberg option). Taking the cube root of each quadratic root |
593
|
|
|
|
|
|
|
completes the operation. |
594
|
|
|
|
|
|
|
|
595
|
|
|
|
|
|
|
This has the benefit of having a simpler matrix to solve, or if the |
596
|
|
|
|
|
|
|
C option is set to zero, has the effect of being able to use one of |
597
|
|
|
|
|
|
|
the classical methods on a polynomial of high degree. In the above example, the |
598
|
|
|
|
|
|
|
order-six polynomial gets solved with the quadratic_roots() function if the |
599
|
|
|
|
|
|
|
hessenberg option is zero. |
600
|
|
|
|
|
|
|
|
601
|
|
|
|
|
|
|
Currently the variable substitution is fairly simple and will only look |
602
|
|
|
|
|
|
|
for gaps of zeros in the coefficients that are multiples of the prime numbers |
603
|
|
|
|
|
|
|
less than or equal to 37 (2, 3, 5, et cetera). |
604
|
|
|
|
|
|
|
|
605
|
|
|
|
|
|
|
=back |
606
|
|
|
|
|
|
|
|
607
|
|
|
|
|
|
|
=cut |
608
|
|
|
|
|
|
|
|
609
|
|
|
|
|
|
|
sub poly_option |
610
|
|
|
|
|
|
|
{ |
611
|
22
|
|
|
22
|
1
|
35273
|
my %opts = @_; |
612
|
22
|
|
|
|
|
34
|
my %old_opts; |
613
|
|
|
|
|
|
|
|
614
|
22
|
100
|
|
|
|
80
|
return %option if (scalar @_ == 0); |
615
|
|
|
|
|
|
|
|
616
|
15
|
|
|
|
|
46
|
for my $okey (keys %opts) |
617
|
|
|
|
|
|
|
{ |
618
|
|
|
|
|
|
|
# |
619
|
|
|
|
|
|
|
# If this is a real option, save its old value, then set it. |
620
|
|
|
|
|
|
|
# |
621
|
15
|
50
|
|
|
|
39
|
if (exists $option{$okey}) |
622
|
|
|
|
|
|
|
{ |
623
|
15
|
|
|
|
|
27
|
$old_opts{$okey} = $option{$okey}; |
624
|
15
|
100
|
|
|
|
46
|
$option{$okey} = ($opts{$okey})? 1: 0; |
625
|
|
|
|
|
|
|
} |
626
|
|
|
|
|
|
|
else |
627
|
|
|
|
|
|
|
{ |
628
|
0
|
|
|
|
|
0
|
carp "poly_option(): unknown key $okey."; |
629
|
|
|
|
|
|
|
} |
630
|
|
|
|
|
|
|
} |
631
|
|
|
|
|
|
|
|
632
|
15
|
|
|
|
|
42
|
return %old_opts; |
633
|
|
|
|
|
|
|
} |
634
|
|
|
|
|
|
|
|
635
|
|
|
|
|
|
|
=head3 build_companion |
636
|
|
|
|
|
|
|
|
637
|
|
|
|
|
|
|
Creates the initial companion matrix of the polynomial. Returns an array |
638
|
|
|
|
|
|
|
of arrays (the internal representation of a matrix). This may be used as |
639
|
|
|
|
|
|
|
an argument to the L contructor: |
640
|
|
|
|
|
|
|
|
641
|
|
|
|
|
|
|
my @cm = build_companion(@coef); |
642
|
|
|
|
|
|
|
|
643
|
|
|
|
|
|
|
my $m = Math::Matrix->new(@cm); |
644
|
|
|
|
|
|
|
$m->print(); |
645
|
|
|
|
|
|
|
|
646
|
|
|
|
|
|
|
The Wikipedia article at L has |
647
|
|
|
|
|
|
|
more information on the subject. |
648
|
|
|
|
|
|
|
|
649
|
|
|
|
|
|
|
=cut |
650
|
|
|
|
|
|
|
|
651
|
|
|
|
|
|
|
# |
652
|
|
|
|
|
|
|
# Perl code to find roots of a polynomial translated by Nick Ing-Simmons |
653
|
|
|
|
|
|
|
# from FORTRAN code by Hiroshi Murakami. |
654
|
|
|
|
|
|
|
# |
655
|
|
|
|
|
|
|
# From the netlib archive: http://netlib.bell-labs.com/netlib/search.html |
656
|
|
|
|
|
|
|
# In particular http://netlib.bell-labs.com/netlib/opt/companion.tgz |
657
|
|
|
|
|
|
|
# |
658
|
|
|
|
|
|
|
sub build_companion |
659
|
|
|
|
|
|
|
{ |
660
|
87
|
50
|
|
87
|
1
|
239
|
my @coefficients = ($ascending_flag == 0)? reverse @_: @_; |
661
|
87
|
|
|
|
|
162
|
my $n = $#coefficients - 1; |
662
|
87
|
|
|
|
|
125
|
my @h; |
663
|
|
|
|
|
|
|
|
664
|
|
|
|
|
|
|
# |
665
|
|
|
|
|
|
|
### build_companion called with: @coefficients |
666
|
|
|
|
|
|
|
# |
667
|
|
|
|
|
|
|
# First step: Divide by the leading coefficient and negate. |
668
|
|
|
|
|
|
|
# |
669
|
87
|
|
|
|
|
145
|
my $cn = - (pop @coefficients); |
670
|
87
|
|
|
|
|
311
|
map($_ /= $cn, @coefficients); |
671
|
|
|
|
|
|
|
|
672
|
|
|
|
|
|
|
# |
673
|
|
|
|
|
|
|
# Next: set up the diagonal matrix. |
674
|
|
|
|
|
|
|
# |
675
|
87
|
|
|
|
|
209
|
for my $i (0 .. $n) |
676
|
|
|
|
|
|
|
{ |
677
|
351
|
|
|
|
|
628
|
$h[$i][$n] = shift @coefficients; |
678
|
351
|
|
|
|
|
897
|
map($h[$i][$_] = 0.0, 0 .. $n - 1); |
679
|
|
|
|
|
|
|
} |
680
|
|
|
|
|
|
|
|
681
|
87
|
|
|
|
|
249
|
map($h[$_][$_ - 1] = 1.0, 1 .. $n); |
682
|
|
|
|
|
|
|
|
683
|
87
|
|
|
|
|
242
|
return @h; |
684
|
|
|
|
|
|
|
} |
685
|
|
|
|
|
|
|
|
686
|
|
|
|
|
|
|
=head3 balance_matrix |
687
|
|
|
|
|
|
|
|
688
|
|
|
|
|
|
|
Balances the matrix (makes the rows and columns have similar norms) created |
689
|
|
|
|
|
|
|
by build_companion() by applying a matrix transformation with a diagonal |
690
|
|
|
|
|
|
|
matrix of powers of two. |
691
|
|
|
|
|
|
|
|
692
|
|
|
|
|
|
|
This is used to help prevent any rounding errors that occur if the elements |
693
|
|
|
|
|
|
|
of the matrix differ greatly in magnitude. |
694
|
|
|
|
|
|
|
|
695
|
|
|
|
|
|
|
=cut |
696
|
|
|
|
|
|
|
|
697
|
|
|
|
|
|
|
# BASE is the base of the floating point representation on the machine. |
698
|
|
|
|
|
|
|
# It is 16 for base 16 float : for example, IBM system 360/370. |
699
|
|
|
|
|
|
|
# It is 2 for base 2 float : for example, IEEE float. |
700
|
|
|
|
|
|
|
sub BASE () { 2 } |
701
|
|
|
|
|
|
|
sub BASESQR () { BASE * BASE } |
702
|
|
|
|
|
|
|
|
703
|
|
|
|
|
|
|
# |
704
|
|
|
|
|
|
|
# @matrix = balance_matrix(@cm); |
705
|
|
|
|
|
|
|
# |
706
|
|
|
|
|
|
|
# Balance the companion matrix created by build_companion(). |
707
|
|
|
|
|
|
|
# |
708
|
|
|
|
|
|
|
# Return an array of arrays representing the N by N matrix. |
709
|
|
|
|
|
|
|
# |
710
|
|
|
|
|
|
|
# In the :numeric export set. |
711
|
|
|
|
|
|
|
# |
712
|
|
|
|
|
|
|
sub balance_matrix |
713
|
|
|
|
|
|
|
{ |
714
|
87
|
|
|
87
|
1
|
184
|
my @h = @_; |
715
|
87
|
|
|
|
|
141
|
my $n = $#h; |
716
|
|
|
|
|
|
|
|
717
|
|
|
|
|
|
|
# |
718
|
|
|
|
|
|
|
### Balancing the unsymmetric matrix A. |
719
|
|
|
|
|
|
|
# |
720
|
|
|
|
|
|
|
##### @h |
721
|
|
|
|
|
|
|
# |
722
|
|
|
|
|
|
|
# Perl code translated by Nick Ing-Simmons from FORTRAN code |
723
|
|
|
|
|
|
|
# by Hiroshi Murakami. |
724
|
|
|
|
|
|
|
# |
725
|
|
|
|
|
|
|
# The Fortran code is based on the Algol code "balance" from paper: |
726
|
|
|
|
|
|
|
# "Balancing a Matrix for Calculation of Eigenvalues and Eigenvectors" |
727
|
|
|
|
|
|
|
# by B. N. Parlett and C. Reinsch, Numer. Math. 13, 293-304(1969). |
728
|
|
|
|
|
|
|
# |
729
|
|
|
|
|
|
|
# Note: The only non-zero elements of the companion matrix are touched. |
730
|
|
|
|
|
|
|
# |
731
|
87
|
|
|
|
|
128
|
my $noconv = 1; |
732
|
87
|
|
|
|
|
179
|
while ($noconv) |
733
|
|
|
|
|
|
|
{ |
734
|
156
|
|
|
|
|
207
|
$noconv = 0; |
735
|
156
|
|
|
|
|
274
|
for my $i (0 .. $n) |
736
|
|
|
|
|
|
|
{ |
737
|
|
|
|
|
|
|
# |
738
|
|
|
|
|
|
|
# Touch only non-zero elements of companion. |
739
|
|
|
|
|
|
|
# |
740
|
681
|
|
|
|
|
784
|
my $c; |
741
|
681
|
100
|
|
|
|
983
|
if ($i != $n) |
742
|
|
|
|
|
|
|
{ |
743
|
525
|
|
|
|
|
783
|
$c = abs($h[$i + 1][$i]); |
744
|
|
|
|
|
|
|
} |
745
|
|
|
|
|
|
|
else |
746
|
|
|
|
|
|
|
{ |
747
|
156
|
|
|
|
|
195
|
$c = 0.0; |
748
|
156
|
|
|
|
|
290
|
for my $j (0 .. $n - 1) |
749
|
|
|
|
|
|
|
{ |
750
|
525
|
|
|
|
|
782
|
$c += abs($h[$j][$n]); |
751
|
|
|
|
|
|
|
} |
752
|
|
|
|
|
|
|
} |
753
|
|
|
|
|
|
|
|
754
|
681
|
|
|
|
|
828
|
my $r; |
755
|
681
|
100
|
|
|
|
1121
|
if ($i == 0) |
|
|
100
|
|
|
|
|
|
756
|
|
|
|
|
|
|
{ |
757
|
156
|
|
|
|
|
530
|
$r = abs($h[0][$n]); |
758
|
|
|
|
|
|
|
} |
759
|
|
|
|
|
|
|
elsif ($i != $n) |
760
|
|
|
|
|
|
|
{ |
761
|
378
|
|
|
|
|
621
|
$r = abs($h[$i][$i - 1]) + abs($h[$i][$n]); |
762
|
|
|
|
|
|
|
} |
763
|
|
|
|
|
|
|
else |
764
|
|
|
|
|
|
|
{ |
765
|
147
|
|
|
|
|
220
|
$r = abs($h[$i][$i - 1]); |
766
|
|
|
|
|
|
|
} |
767
|
|
|
|
|
|
|
|
768
|
681
|
100
|
66
|
|
|
1728
|
next if ($c == 0.0 || $r == 0.0); |
769
|
|
|
|
|
|
|
|
770
|
672
|
|
|
|
|
905
|
my $g = $r / BASE; |
771
|
672
|
|
|
|
|
771
|
my $f = 1.0; |
772
|
672
|
|
|
|
|
863
|
my $s = $c + $r; |
773
|
672
|
|
|
|
|
1133
|
while ( $c < $g ) |
774
|
|
|
|
|
|
|
{ |
775
|
205
|
|
|
|
|
257
|
$f = $f * BASE; |
776
|
205
|
|
|
|
|
325
|
$c = $c * BASESQR; |
777
|
|
|
|
|
|
|
} |
778
|
|
|
|
|
|
|
|
779
|
672
|
|
|
|
|
794
|
$g = $r * BASE; |
780
|
672
|
|
|
|
|
1070
|
while ($c >= $g) |
781
|
|
|
|
|
|
|
{ |
782
|
113
|
|
|
|
|
152
|
$f = $f / BASE; |
783
|
113
|
|
|
|
|
191
|
$c = $c / BASESQR; |
784
|
|
|
|
|
|
|
} |
785
|
|
|
|
|
|
|
|
786
|
672
|
100
|
|
|
|
1377
|
if (($c + $r) < 0.95 * $s * $f) |
787
|
|
|
|
|
|
|
{ |
788
|
173
|
|
|
|
|
226
|
$g = 1.0 / $f; |
789
|
173
|
|
|
|
|
213
|
$noconv = 1; |
790
|
|
|
|
|
|
|
|
791
|
|
|
|
|
|
|
#C Generic code. |
792
|
|
|
|
|
|
|
#C do $j=1,$n |
793
|
|
|
|
|
|
|
#C $h($i,$j)=$h($i,$j)*$g |
794
|
|
|
|
|
|
|
#C enddo |
795
|
|
|
|
|
|
|
#C do $j=1,$n |
796
|
|
|
|
|
|
|
#C $h($j,$i)=$h($j,$i)*$f |
797
|
|
|
|
|
|
|
#C enddo |
798
|
|
|
|
|
|
|
#C begin specific code. Touch only non-zero elements of companion. |
799
|
173
|
100
|
|
|
|
299
|
if ($i == 0) |
800
|
|
|
|
|
|
|
{ |
801
|
41
|
|
|
|
|
61
|
$h[0][$n] *= $g; |
802
|
|
|
|
|
|
|
} |
803
|
|
|
|
|
|
|
else |
804
|
|
|
|
|
|
|
{ |
805
|
132
|
|
|
|
|
202
|
$h[$i][$i - 1] *= $g; |
806
|
132
|
|
|
|
|
181
|
$h[$i][$n] *= $g; |
807
|
|
|
|
|
|
|
} |
808
|
173
|
100
|
|
|
|
262
|
if ($i != $n) |
809
|
|
|
|
|
|
|
{ |
810
|
145
|
|
|
|
|
268
|
$h[$i + 1][$i] *= $f; |
811
|
|
|
|
|
|
|
} |
812
|
|
|
|
|
|
|
else |
813
|
|
|
|
|
|
|
{ |
814
|
28
|
|
|
|
|
56
|
for my $j (0 .. $n) |
815
|
|
|
|
|
|
|
{ |
816
|
142
|
|
|
|
|
210
|
$h[$j][$i] *= $f; |
817
|
|
|
|
|
|
|
} |
818
|
|
|
|
|
|
|
} |
819
|
|
|
|
|
|
|
} |
820
|
|
|
|
|
|
|
} # for $i |
821
|
|
|
|
|
|
|
} # while $noconv |
822
|
|
|
|
|
|
|
|
823
|
|
|
|
|
|
|
# |
824
|
|
|
|
|
|
|
### Returning balanced matrix. |
825
|
|
|
|
|
|
|
##### @h |
826
|
|
|
|
|
|
|
# |
827
|
87
|
|
|
|
|
242
|
return @h; |
828
|
|
|
|
|
|
|
} |
829
|
|
|
|
|
|
|
|
830
|
|
|
|
|
|
|
|
831
|
|
|
|
|
|
|
=head3 hqr_eigen_hessenberg |
832
|
|
|
|
|
|
|
|
833
|
|
|
|
|
|
|
Returns the roots of the polynomial equation by solving the matrix created by |
834
|
|
|
|
|
|
|
C and C. See L. |
835
|
|
|
|
|
|
|
|
836
|
|
|
|
|
|
|
=cut |
837
|
|
|
|
|
|
|
|
838
|
|
|
|
|
|
|
sub hqr_eigen_hessenberg |
839
|
|
|
|
|
|
|
{ |
840
|
87
|
|
|
87
|
1
|
178
|
my @h = @_; |
841
|
87
|
|
|
|
|
125
|
my $n = $#h; |
842
|
|
|
|
|
|
|
|
843
|
|
|
|
|
|
|
# |
844
|
|
|
|
|
|
|
### hqr_eigen_hessenberg() |
845
|
|
|
|
|
|
|
# |
846
|
|
|
|
|
|
|
# Eigenvalue Computation by the Householder QR method for the |
847
|
|
|
|
|
|
|
# Real Hessenberg matrix. |
848
|
|
|
|
|
|
|
# |
849
|
|
|
|
|
|
|
# Perl code translated by Nick Ing-Simmons from FORTRAN code |
850
|
|
|
|
|
|
|
# by Hiroshi Murakami. |
851
|
|
|
|
|
|
|
# |
852
|
|
|
|
|
|
|
# The Fortran code is based on the Algol code "hqr" from the paper: |
853
|
|
|
|
|
|
|
# "The QR Algorithm for Real Hessenberg Matrices" |
854
|
|
|
|
|
|
|
# by R. S. Martin, G. Peters and J. H. Wilkinson, |
855
|
|
|
|
|
|
|
# Numer. Math. 14, 219-231(1970). |
856
|
|
|
|
|
|
|
# |
857
|
87
|
|
|
|
|
131
|
my($p, $q, $r); |
858
|
87
|
|
|
|
|
118
|
my $t = 0.0; |
859
|
|
|
|
|
|
|
|
860
|
87
|
|
|
|
|
109
|
my @roots; |
861
|
|
|
|
|
|
|
|
862
|
|
|
|
|
|
|
ROOT: |
863
|
87
|
|
|
|
|
166
|
while ($n >= 0) |
864
|
|
|
|
|
|
|
{ |
865
|
207
|
|
|
|
|
289
|
my $its = 0; |
866
|
207
|
|
|
|
|
277
|
my $na = $n - 1; |
867
|
|
|
|
|
|
|
|
868
|
207
|
|
|
|
|
430
|
while ($its < $iteration{hessenberg}) |
869
|
|
|
|
|
|
|
{ |
870
|
1298
|
|
|
|
|
1705
|
my($w, $x, $y); |
871
|
|
|
|
|
|
|
|
872
|
|
|
|
|
|
|
# |
873
|
|
|
|
|
|
|
# Look for single small sub-diagonal element; |
874
|
|
|
|
|
|
|
# |
875
|
1298
|
|
|
|
|
1574
|
my $l = 0; |
876
|
1298
|
|
|
|
|
2160
|
for my $d (reverse 1 .. $n) |
877
|
|
|
|
|
|
|
{ |
878
|
4215
|
100
|
|
|
|
9031
|
if (abs( $h[$d][ $d - 1 ] ) <= $epsilon * |
879
|
|
|
|
|
|
|
(abs( $h[ $d - 1 ][ $d - 1 ] ) + |
880
|
|
|
|
|
|
|
abs( $h[$d][$d] ) ) ) |
881
|
|
|
|
|
|
|
{ |
882
|
157
|
|
|
|
|
220
|
$l = $d; |
883
|
157
|
|
|
|
|
222
|
last; |
884
|
|
|
|
|
|
|
} |
885
|
|
|
|
|
|
|
} |
886
|
|
|
|
|
|
|
|
887
|
1298
|
|
|
|
|
1927
|
$x = $h[$n][$n]; |
888
|
|
|
|
|
|
|
|
889
|
1298
|
100
|
|
|
|
2048
|
if ($l == $n) |
890
|
|
|
|
|
|
|
{ |
891
|
|
|
|
|
|
|
# |
892
|
|
|
|
|
|
|
# One (real) root found. |
893
|
|
|
|
|
|
|
# |
894
|
63
|
|
|
|
|
87
|
$n--; |
895
|
63
|
|
|
|
|
113
|
push @roots, $x + $t; |
896
|
63
|
|
|
|
|
165
|
next ROOT; |
897
|
|
|
|
|
|
|
} |
898
|
|
|
|
|
|
|
|
899
|
1235
|
|
|
|
|
1492
|
$y = $h[$na][$na]; |
900
|
1235
|
|
|
|
|
1590
|
$w = $h[$n][$na] * $h[$na][$n]; |
901
|
|
|
|
|
|
|
|
902
|
1235
|
100
|
|
|
|
1860
|
if ($l == $na) |
903
|
|
|
|
|
|
|
{ |
904
|
144
|
|
|
|
|
211
|
$p = ( $y - $x ) / 2; |
905
|
144
|
|
|
|
|
191
|
$q = $p * $p + $w; |
906
|
144
|
|
|
|
|
358
|
$y = sqrt( abs($q) ); |
907
|
144
|
|
|
|
|
909
|
$x += $t; |
908
|
|
|
|
|
|
|
|
909
|
144
|
100
|
|
|
|
264
|
if ($q > 0.0) |
910
|
|
|
|
|
|
|
{ |
911
|
|
|
|
|
|
|
# |
912
|
|
|
|
|
|
|
# Real pair. |
913
|
|
|
|
|
|
|
# |
914
|
17
|
100
|
|
|
|
50
|
$y = -$y if ( $p < 0.0 ); |
915
|
17
|
|
|
|
|
26
|
$y += $p; |
916
|
17
|
|
|
|
|
33
|
push @roots, $x - $w / $y; |
917
|
17
|
|
|
|
|
29
|
push @roots, $x + $y; |
918
|
|
|
|
|
|
|
} |
919
|
|
|
|
|
|
|
else |
920
|
|
|
|
|
|
|
{ |
921
|
|
|
|
|
|
|
# |
922
|
|
|
|
|
|
|
# Complex or twin pair. |
923
|
|
|
|
|
|
|
# |
924
|
127
|
|
|
|
|
307
|
push @roots, $x + $p - $y * i; |
925
|
127
|
|
|
|
|
32947
|
push @roots, $x + $p + $y * i; |
926
|
|
|
|
|
|
|
} |
927
|
|
|
|
|
|
|
|
928
|
144
|
|
|
|
|
27470
|
$n -= 2; |
929
|
144
|
|
|
|
|
464
|
next ROOT; |
930
|
|
|
|
|
|
|
} |
931
|
|
|
|
|
|
|
|
932
|
1091
|
50
|
|
|
|
1758
|
croak "Too many iterations ($its) at n=$n\n" if ($its >= $iteration{hessenberg}); |
933
|
|
|
|
|
|
|
|
934
|
1091
|
100
|
100
|
|
|
2954
|
if ($its && $its % 10 == 0) |
935
|
|
|
|
|
|
|
{ |
936
|
|
|
|
|
|
|
# |
937
|
|
|
|
|
|
|
# Form exceptional shift. |
938
|
|
|
|
|
|
|
# |
939
|
|
|
|
|
|
|
### Exceptional shift at: $its |
940
|
|
|
|
|
|
|
# |
941
|
|
|
|
|
|
|
|
942
|
53
|
|
|
|
|
83
|
$t += $x; |
943
|
53
|
|
|
|
|
89
|
for my $i (0 .. $n) |
944
|
|
|
|
|
|
|
{ |
945
|
249
|
|
|
|
|
355
|
$h[$i][$i] -= $x; |
946
|
|
|
|
|
|
|
} |
947
|
|
|
|
|
|
|
|
948
|
53
|
|
|
|
|
131
|
my $s = abs($h[$n][$na]) + abs($h[$na][$n - 2]); |
949
|
53
|
|
|
|
|
69
|
$y = 0.75 * $s; |
950
|
53
|
|
|
|
|
70
|
$x = $y; |
951
|
53
|
|
|
|
|
80
|
$w = -0.4375 * $s * $s; |
952
|
|
|
|
|
|
|
} |
953
|
|
|
|
|
|
|
|
954
|
1091
|
|
|
|
|
1331
|
$its++; |
955
|
|
|
|
|
|
|
|
956
|
|
|
|
|
|
|
# |
957
|
|
|
|
|
|
|
### Look for two consecutive small |
958
|
|
|
|
|
|
|
### sub-diagonal elements. |
959
|
|
|
|
|
|
|
# |
960
|
1091
|
|
|
|
|
1385
|
my $m = $l; # Set in case we fall through the loop. |
961
|
1091
|
|
|
|
|
1757
|
for my $d (reverse $l .. $n - 2) |
962
|
|
|
|
|
|
|
{ |
963
|
2809
|
|
|
|
|
3588
|
my $z = $h[$d][$d]; |
964
|
2809
|
|
|
|
|
3636
|
my $s = $y - $z; |
965
|
2809
|
|
|
|
|
3246
|
$r = $x - $z; |
966
|
2809
|
|
|
|
|
4436
|
$p = ($r * $s - $w) / $h[$d + 1][$d] + $h[$d][$d + 1]; |
967
|
2809
|
|
|
|
|
3849
|
$q = $h[$d + 1][$d + 1] - $z - $r - $s; |
968
|
2809
|
|
|
|
|
3494
|
$r = $h[$d + 2][$d + 1]; |
969
|
|
|
|
|
|
|
|
970
|
2809
|
|
|
|
|
3829
|
$s = abs($p) + abs($q) + abs($r); |
971
|
2809
|
|
|
|
|
3280
|
$p /= $s; |
972
|
2809
|
|
|
|
|
3087
|
$q /= $s; |
973
|
2809
|
|
|
|
|
3096
|
$r /= $s; |
974
|
|
|
|
|
|
|
|
975
|
|
|
|
|
|
|
# |
976
|
|
|
|
|
|
|
# The sub-diagonal check doesn't get made for |
977
|
|
|
|
|
|
|
# the last iteration of the loop, and the only |
978
|
|
|
|
|
|
|
# reason we have the loop continue up to this |
979
|
|
|
|
|
|
|
# point is to set $p, $q, and $r. |
980
|
|
|
|
|
|
|
# |
981
|
2809
|
100
|
|
|
|
4327
|
last if ($d == $l); |
982
|
|
|
|
|
|
|
|
983
|
1726
|
100
|
|
|
|
4620
|
if (abs($h[$d][$d - 1]) * (abs($q) + abs($r)) <= |
984
|
|
|
|
|
|
|
$epsilon * abs($p) * ( |
985
|
|
|
|
|
|
|
abs($h[$d - 1][$d - 1]) + |
986
|
|
|
|
|
|
|
abs($z) + |
987
|
|
|
|
|
|
|
abs($h[$d + 1][$d + 1]) |
988
|
|
|
|
|
|
|
)) |
989
|
|
|
|
|
|
|
{ |
990
|
8
|
|
|
|
|
23
|
$m = $d; |
991
|
8
|
|
|
|
|
14
|
last; |
992
|
|
|
|
|
|
|
} |
993
|
|
|
|
|
|
|
} |
994
|
|
|
|
|
|
|
|
995
|
|
|
|
|
|
|
# |
996
|
|
|
|
|
|
|
#### $n |
997
|
|
|
|
|
|
|
#### $l |
998
|
|
|
|
|
|
|
#### $m |
999
|
|
|
|
|
|
|
# |
1000
|
1091
|
|
|
|
|
1839
|
for my $i (($m + 2) .. $n) |
1001
|
|
|
|
|
|
|
{ |
1002
|
2809
|
|
|
|
|
3924
|
$h[$i][$i - 2] = 0.0; |
1003
|
|
|
|
|
|
|
} |
1004
|
1091
|
|
|
|
|
1526
|
for my $i (($m + 3) .. $n) |
1005
|
|
|
|
|
|
|
{ |
1006
|
1718
|
|
|
|
|
2372
|
$h[$i][$i - 3] = 0.0; |
1007
|
|
|
|
|
|
|
} |
1008
|
|
|
|
|
|
|
|
1009
|
|
|
|
|
|
|
# |
1010
|
|
|
|
|
|
|
# Double QR step involving rows $l to $n and |
1011
|
|
|
|
|
|
|
# columns $m to $n. |
1012
|
|
|
|
|
|
|
# |
1013
|
1091
|
|
|
|
|
1460
|
for my $k ($m .. $na) |
1014
|
|
|
|
|
|
|
{ |
1015
|
3900
|
|
|
|
|
4478
|
my $z; |
1016
|
3900
|
|
|
|
|
5107
|
my $notlast = ($k != $na); |
1017
|
3900
|
100
|
|
|
|
5908
|
if ($k != $m) |
1018
|
|
|
|
|
|
|
{ |
1019
|
2809
|
|
|
|
|
3758
|
$p = $h[$k][$k - 1]; |
1020
|
2809
|
|
|
|
|
3854
|
$q = $h[$k + 1][$k - 1]; |
1021
|
2809
|
100
|
|
|
|
4371
|
$r = ($notlast)? $h[$k + 2][$k - 1]: 0.0; |
1022
|
|
|
|
|
|
|
|
1023
|
2809
|
|
|
|
|
4247
|
$x = abs($p) + abs($q) + abs($r); |
1024
|
2809
|
50
|
|
|
|
4293
|
next if ( $x == 0.0 ); |
1025
|
|
|
|
|
|
|
|
1026
|
2809
|
|
|
|
|
3359
|
$p /= $x; |
1027
|
2809
|
|
|
|
|
3099
|
$q /= $x; |
1028
|
2809
|
|
|
|
|
3243
|
$r /= $x; |
1029
|
|
|
|
|
|
|
} |
1030
|
|
|
|
|
|
|
|
1031
|
3900
|
|
|
|
|
8419
|
my $s = sqrt($p * $p + $q * $q + $r * $r); |
1032
|
3900
|
100
|
|
|
|
23642
|
$s = -$s if ($p < 0.0); |
1033
|
|
|
|
|
|
|
|
1034
|
3900
|
100
|
|
|
|
6326
|
if ($k != $m) |
|
|
100
|
|
|
|
|
|
1035
|
|
|
|
|
|
|
{ |
1036
|
2809
|
|
|
|
|
4157
|
$h[$k][$k - 1] = -$s * $x; |
1037
|
|
|
|
|
|
|
} |
1038
|
|
|
|
|
|
|
elsif ($l != $m) |
1039
|
|
|
|
|
|
|
{ |
1040
|
8
|
|
|
|
|
17
|
$h[$k][$k - 1] *= -1; |
1041
|
|
|
|
|
|
|
} |
1042
|
|
|
|
|
|
|
|
1043
|
3900
|
|
|
|
|
4596
|
$p += $s; |
1044
|
3900
|
|
|
|
|
4713
|
$x = $p / $s; |
1045
|
3900
|
|
|
|
|
4540
|
$y = $q / $s; |
1046
|
3900
|
|
|
|
|
4526
|
$z = $r / $s; |
1047
|
3900
|
|
|
|
|
4436
|
$q /= $p; |
1048
|
3900
|
|
|
|
|
4297
|
$r /= $p; |
1049
|
|
|
|
|
|
|
|
1050
|
|
|
|
|
|
|
# |
1051
|
|
|
|
|
|
|
# Row modification. |
1052
|
|
|
|
|
|
|
# |
1053
|
3900
|
|
|
|
|
5719
|
for my $j ($k .. $n) |
1054
|
|
|
|
|
|
|
{ |
1055
|
14074
|
|
|
|
|
18695
|
$p = $h[$k][$j] + $q * $h[$k + 1][$j]; |
1056
|
|
|
|
|
|
|
|
1057
|
14074
|
100
|
|
|
|
19825
|
if ($notlast) |
1058
|
|
|
|
|
|
|
{ |
1059
|
11892
|
|
|
|
|
14917
|
$p += $r * $h[ $k + 2 ][$j]; |
1060
|
11892
|
|
|
|
|
15148
|
$h[ $k + 2 ][$j] -= $p * $z; |
1061
|
|
|
|
|
|
|
} |
1062
|
|
|
|
|
|
|
|
1063
|
14074
|
|
|
|
|
17445
|
$h[ $k + 1 ][$j] -= $p * $y; |
1064
|
14074
|
|
|
|
|
18856
|
$h[$k][$j] -= $p * $x; |
1065
|
|
|
|
|
|
|
} |
1066
|
|
|
|
|
|
|
|
1067
|
3900
|
|
|
|
|
4913
|
my $j = $k + 3; |
1068
|
3900
|
100
|
|
|
|
6018
|
$j = $n if ($j > $n); |
1069
|
|
|
|
|
|
|
|
1070
|
|
|
|
|
|
|
# |
1071
|
|
|
|
|
|
|
# Column modification. |
1072
|
|
|
|
|
|
|
# |
1073
|
3900
|
|
|
|
|
5625
|
for my $i ($l .. $j) |
1074
|
|
|
|
|
|
|
{ |
1075
|
18629
|
|
|
|
|
24660
|
$p = $x * $h[$i][$k] + |
1076
|
|
|
|
|
|
|
$y * $h[$i][$k + 1]; |
1077
|
|
|
|
|
|
|
|
1078
|
18629
|
100
|
|
|
|
25987
|
if ($notlast) |
1079
|
|
|
|
|
|
|
{ |
1080
|
13624
|
|
|
|
|
16583
|
$p += $z * $h[$i][$k + 2]; |
1081
|
13624
|
|
|
|
|
17001
|
$h[$i][$k + 2] -= $p * $r; |
1082
|
|
|
|
|
|
|
} |
1083
|
|
|
|
|
|
|
|
1084
|
18629
|
|
|
|
|
22603
|
$h[$i][$k + 1] -= $p * $q; |
1085
|
18629
|
|
|
|
|
25631
|
$h[$i][$k] -= $p; |
1086
|
|
|
|
|
|
|
} |
1087
|
|
|
|
|
|
|
} # for $k |
1088
|
|
|
|
|
|
|
} # while $its |
1089
|
|
|
|
|
|
|
} # while $n |
1090
|
87
|
|
|
|
|
292
|
return @roots; |
1091
|
|
|
|
|
|
|
} |
1092
|
|
|
|
|
|
|
|
1093
|
|
|
|
|
|
|
|
1094
|
|
|
|
|
|
|
=head2 Classical Functions |
1095
|
|
|
|
|
|
|
|
1096
|
|
|
|
|
|
|
These are the functions that solve polynomials via the classical methods. |
1097
|
|
|
|
|
|
|
Quartic, cubic, quadratic, and even linear equations may be solved with |
1098
|
|
|
|
|
|
|
these functions. They are all exported under the tag "classical". |
1099
|
|
|
|
|
|
|
|
1100
|
|
|
|
|
|
|
L will use these functions I the hessenberg option |
1101
|
|
|
|
|
|
|
is set to 0, I the degree of the polynomial is four or less. |
1102
|
|
|
|
|
|
|
|
1103
|
|
|
|
|
|
|
The leading coefficient C<$a> must always be non-zero for the classical |
1104
|
|
|
|
|
|
|
functions. |
1105
|
|
|
|
|
|
|
|
1106
|
|
|
|
|
|
|
=head3 linear_roots() |
1107
|
|
|
|
|
|
|
|
1108
|
|
|
|
|
|
|
Here for completeness's sake more than anything else. Returns the |
1109
|
|
|
|
|
|
|
solution for |
1110
|
|
|
|
|
|
|
|
1111
|
|
|
|
|
|
|
ax + b = 0 |
1112
|
|
|
|
|
|
|
|
1113
|
|
|
|
|
|
|
by returning C<-b/a>. This may be in either a scalar or an array context. |
1114
|
|
|
|
|
|
|
|
1115
|
|
|
|
|
|
|
=cut |
1116
|
|
|
|
|
|
|
|
1117
|
|
|
|
|
|
|
sub linear_roots |
1118
|
|
|
|
|
|
|
{ |
1119
|
9
|
50
|
|
9
|
1
|
29
|
my($b, $a) = ($ascending_flag == 0)? reverse @_: @_; |
1120
|
|
|
|
|
|
|
|
1121
|
9
|
50
|
|
|
|
26
|
if (abs($a) < $epsilon) |
1122
|
|
|
|
|
|
|
{ |
1123
|
0
|
|
|
|
|
0
|
carp "The coefficient of the highest power must not be zero!\n"; |
1124
|
0
|
|
|
|
|
0
|
return (); |
1125
|
|
|
|
|
|
|
} |
1126
|
|
|
|
|
|
|
|
1127
|
9
|
50
|
|
|
|
33
|
return wantarray? (-$b/$a): -$b/$a; |
1128
|
|
|
|
|
|
|
} |
1129
|
|
|
|
|
|
|
|
1130
|
|
|
|
|
|
|
|
1131
|
|
|
|
|
|
|
=head3 quadratic_roots() |
1132
|
|
|
|
|
|
|
|
1133
|
|
|
|
|
|
|
Gives the roots of the quadratic equation |
1134
|
|
|
|
|
|
|
|
1135
|
|
|
|
|
|
|
ax**2 + bx + c = 0 |
1136
|
|
|
|
|
|
|
|
1137
|
|
|
|
|
|
|
using the well-known quadratic formula. Returns a two-element list. |
1138
|
|
|
|
|
|
|
|
1139
|
|
|
|
|
|
|
=cut |
1140
|
|
|
|
|
|
|
|
1141
|
|
|
|
|
|
|
sub quadratic_roots |
1142
|
|
|
|
|
|
|
{ |
1143
|
40
|
50
|
|
40
|
1
|
4158
|
my($c, $b, $a) = ($ascending_flag == 0)? reverse @_: @_; |
1144
|
|
|
|
|
|
|
|
1145
|
40
|
50
|
|
|
|
103
|
if (abs($a) < $epsilon) |
1146
|
|
|
|
|
|
|
{ |
1147
|
0
|
|
|
|
|
0
|
carp "The coefficient of the highest power must not be zero!\n"; |
1148
|
0
|
|
|
|
|
0
|
return (); |
1149
|
|
|
|
|
|
|
} |
1150
|
|
|
|
|
|
|
|
1151
|
40
|
100
|
|
|
|
97
|
return (0, -$b/$a) if (abs($c) < $epsilon); |
1152
|
|
|
|
|
|
|
|
1153
|
38
|
|
|
|
|
163
|
my $dis_sqrt = sqrt($b*$b - $a * 4 * $c); |
1154
|
|
|
|
|
|
|
|
1155
|
38
|
100
|
|
|
|
1686
|
$dis_sqrt = -$dis_sqrt if ($b < $epsilon); # Avoid catastrophic cancellation. |
1156
|
|
|
|
|
|
|
|
1157
|
38
|
|
|
|
|
802
|
my $xt = ($b + $dis_sqrt)/-2; |
1158
|
|
|
|
|
|
|
|
1159
|
38
|
|
|
|
|
2686
|
return ($xt/$a, $c/$xt); |
1160
|
|
|
|
|
|
|
} |
1161
|
|
|
|
|
|
|
|
1162
|
|
|
|
|
|
|
|
1163
|
|
|
|
|
|
|
=head3 cubic_roots() |
1164
|
|
|
|
|
|
|
|
1165
|
|
|
|
|
|
|
Gives the roots of the cubic equation |
1166
|
|
|
|
|
|
|
|
1167
|
|
|
|
|
|
|
ax**3 + bx**2 + cx + d = 0 |
1168
|
|
|
|
|
|
|
|
1169
|
|
|
|
|
|
|
by the method described by R. W. D. Nickalls (see the L |
1170
|
|
|
|
|
|
|
section below). Returns a three-element list. The first element will |
1171
|
|
|
|
|
|
|
always be real. The next two values will either be both real or both |
1172
|
|
|
|
|
|
|
complex numbers. |
1173
|
|
|
|
|
|
|
|
1174
|
|
|
|
|
|
|
=cut |
1175
|
|
|
|
|
|
|
|
1176
|
|
|
|
|
|
|
sub cubic_roots |
1177
|
|
|
|
|
|
|
{ |
1178
|
23
|
50
|
|
23
|
1
|
11201
|
my($d, $c, $b, $a) = ($ascending_flag == 0)? reverse @_: @_; |
1179
|
23
|
|
|
|
|
45
|
my @x; |
1180
|
|
|
|
|
|
|
|
1181
|
23
|
50
|
|
|
|
68
|
if (abs($a) < $epsilon) |
1182
|
|
|
|
|
|
|
{ |
1183
|
0
|
|
|
|
|
0
|
carp "The coefficient of the highest power must not be zero!\n"; |
1184
|
0
|
|
|
|
|
0
|
return @x; |
1185
|
|
|
|
|
|
|
} |
1186
|
|
|
|
|
|
|
|
1187
|
|
|
|
|
|
|
# |
1188
|
|
|
|
|
|
|
# We're calling exported functions that also check |
1189
|
|
|
|
|
|
|
# the $ascending_flag. To avoid reversing the reversed, |
1190
|
|
|
|
|
|
|
# temporarily set the flag to zero and reset before returning. |
1191
|
|
|
|
|
|
|
# |
1192
|
23
|
|
|
|
|
42
|
my $temp_ascending_flag = $ascending_flag; |
1193
|
23
|
|
|
|
|
38
|
$ascending_flag = 1; |
1194
|
|
|
|
|
|
|
|
1195
|
23
|
100
|
|
|
|
61
|
if (abs($d) < $epsilon) |
1196
|
|
|
|
|
|
|
{ |
1197
|
2
|
|
|
|
|
9
|
@x = quadratic_roots($c, $b, $a); |
1198
|
2
|
|
|
|
|
3
|
$ascending_flag = $temp_ascending_flag; |
1199
|
2
|
|
|
|
|
6
|
return (0, @x); |
1200
|
|
|
|
|
|
|
} |
1201
|
|
|
|
|
|
|
|
1202
|
21
|
|
|
|
|
56
|
my $xN = -$b/3/$a; |
1203
|
21
|
|
|
|
|
53
|
my $yN = $d + $xN * ($c + $xN * ($b + $a * $xN)); |
1204
|
|
|
|
|
|
|
|
1205
|
21
|
|
|
|
|
42
|
my $two_a = 2 * $a; |
1206
|
21
|
|
|
|
|
57
|
my $delta_sq = ($b * $b - 3 * $a * $c)/(9 * $a * $a); |
1207
|
21
|
|
|
|
|
68
|
my $h_sq = 4/9 * ($b * $b - 3 * $a * $c) * $delta_sq**2; |
1208
|
21
|
|
|
|
|
44
|
my $dis = $yN * $yN - $h_sq; |
1209
|
21
|
|
|
|
|
35
|
my $twothirds_pi = (2 * pi)/3; |
1210
|
|
|
|
|
|
|
|
1211
|
|
|
|
|
|
|
# |
1212
|
|
|
|
|
|
|
### cubic_roots() calculations... |
1213
|
|
|
|
|
|
|
#### $two_a |
1214
|
|
|
|
|
|
|
#### $delta_sq |
1215
|
|
|
|
|
|
|
#### $h_sq |
1216
|
|
|
|
|
|
|
#### $dis |
1217
|
|
|
|
|
|
|
# |
1218
|
21
|
100
|
|
|
|
71
|
if ($dis > $epsilon) |
|
|
100
|
|
|
|
|
|
1219
|
|
|
|
|
|
|
{ |
1220
|
|
|
|
|
|
|
# |
1221
|
|
|
|
|
|
|
### Cubic branch 1, $dis is greater than 0... |
1222
|
|
|
|
|
|
|
# |
1223
|
|
|
|
|
|
|
# One real root, two complex roots. |
1224
|
|
|
|
|
|
|
# |
1225
|
10
|
|
|
|
|
38
|
my $dis_sqrt = sqrt($dis); |
1226
|
10
|
|
|
|
|
96
|
my $r_p = $yN - $dis_sqrt; |
1227
|
10
|
|
|
|
|
22
|
my $r_q = $yN + $dis_sqrt; |
1228
|
10
|
|
|
|
|
35
|
my $p = cbrt( abs($r_p)/$two_a ); |
1229
|
10
|
|
|
|
|
100
|
my $q = cbrt( abs($r_q)/$two_a ); |
1230
|
|
|
|
|
|
|
|
1231
|
10
|
100
|
|
|
|
78
|
$p = -$p if ($r_p > 0); |
1232
|
10
|
100
|
|
|
|
34
|
$q = -$q if ($r_q > 0); |
1233
|
|
|
|
|
|
|
|
1234
|
10
|
|
|
|
|
25
|
$x[0] = $xN + $p + $q; |
1235
|
10
|
|
|
|
|
30
|
$x[1] = $xN + $p * exp($twothirds_pi * i) |
1236
|
|
|
|
|
|
|
+ $q * exp(-$twothirds_pi * i); |
1237
|
10
|
|
|
|
|
6579
|
$x[2] = ~$x[1]; |
1238
|
|
|
|
|
|
|
} |
1239
|
|
|
|
|
|
|
elsif ($dis < -$epsilon) |
1240
|
|
|
|
|
|
|
{ |
1241
|
|
|
|
|
|
|
# |
1242
|
|
|
|
|
|
|
### Cubic branch 2, $dis is less than 0... |
1243
|
|
|
|
|
|
|
# |
1244
|
|
|
|
|
|
|
# Three distinct real roots. |
1245
|
|
|
|
|
|
|
# |
1246
|
7
|
|
|
|
|
31
|
my $theta = acos(-$yN/sqrt($h_sq))/3; |
1247
|
7
|
|
|
|
|
139
|
my $delta = sqrt($b * $b - 3 * $a * $c)/(3 * $a); |
1248
|
7
|
|
|
|
|
70
|
my $two_d = 2 * $delta; |
1249
|
|
|
|
|
|
|
|
1250
|
7
|
|
|
|
|
39
|
@x = ($xN + $two_d * cos($theta), |
1251
|
|
|
|
|
|
|
$xN + $two_d * cos($twothirds_pi - $theta), |
1252
|
|
|
|
|
|
|
$xN + $two_d * cos($twothirds_pi + $theta)); |
1253
|
|
|
|
|
|
|
} |
1254
|
|
|
|
|
|
|
else |
1255
|
|
|
|
|
|
|
{ |
1256
|
|
|
|
|
|
|
# |
1257
|
|
|
|
|
|
|
### Cubic branch 3, $dis equals 0, within epsilon... |
1258
|
|
|
|
|
|
|
# |
1259
|
|
|
|
|
|
|
# abs($dis) <= $epsilon (effectively zero). |
1260
|
|
|
|
|
|
|
# |
1261
|
|
|
|
|
|
|
# Three real roots (two or three equal). |
1262
|
|
|
|
|
|
|
# |
1263
|
4
|
|
|
|
|
17
|
my $delta = cbrt($yN/$two_a); |
1264
|
|
|
|
|
|
|
|
1265
|
4
|
|
|
|
|
48
|
@x = ($xN + $delta, $xN + $delta, $xN - 2 * $delta); |
1266
|
|
|
|
|
|
|
} |
1267
|
|
|
|
|
|
|
|
1268
|
21
|
|
|
|
|
454
|
$ascending_flag = $temp_ascending_flag; |
1269
|
21
|
|
|
|
|
87
|
return @x; |
1270
|
|
|
|
|
|
|
} |
1271
|
|
|
|
|
|
|
|
1272
|
|
|
|
|
|
|
=head3 quartic_roots() |
1273
|
|
|
|
|
|
|
|
1274
|
|
|
|
|
|
|
Gives the roots of the quartic equation |
1275
|
|
|
|
|
|
|
|
1276
|
|
|
|
|
|
|
ax**4 + bx**3 + cx**2 + dx + e = 0 |
1277
|
|
|
|
|
|
|
|
1278
|
|
|
|
|
|
|
using Ferrari's method (see the L section below). Returns |
1279
|
|
|
|
|
|
|
a four-element list. The first two elements will be either |
1280
|
|
|
|
|
|
|
both real or both complex. The next two elements will also be alike in |
1281
|
|
|
|
|
|
|
type. |
1282
|
|
|
|
|
|
|
|
1283
|
|
|
|
|
|
|
=cut |
1284
|
|
|
|
|
|
|
|
1285
|
|
|
|
|
|
|
sub quartic_roots |
1286
|
|
|
|
|
|
|
{ |
1287
|
22
|
50
|
|
22
|
1
|
8841
|
my($e, $d, $c, $b, $a) = ($ascending_flag == 0)? reverse @_: @_; |
1288
|
22
|
|
|
|
|
35
|
my @x = (); |
1289
|
|
|
|
|
|
|
|
1290
|
22
|
50
|
|
|
|
50
|
if (abs($a) < $epsilon) |
1291
|
|
|
|
|
|
|
{ |
1292
|
0
|
|
|
|
|
0
|
carp "Coefficient of highest power must not be zero!\n"; |
1293
|
0
|
|
|
|
|
0
|
return @x; |
1294
|
|
|
|
|
|
|
} |
1295
|
|
|
|
|
|
|
|
1296
|
|
|
|
|
|
|
# |
1297
|
|
|
|
|
|
|
# We're calling exported functions that also check |
1298
|
|
|
|
|
|
|
# the $ascending_flag. To avoid reversing the reversed, |
1299
|
|
|
|
|
|
|
# temporarily set the flag to one and reset before returning. |
1300
|
|
|
|
|
|
|
# |
1301
|
22
|
|
|
|
|
31
|
my $temp_ascending_flag = $ascending_flag; |
1302
|
22
|
|
|
|
|
29
|
$ascending_flag = 1; |
1303
|
|
|
|
|
|
|
|
1304
|
22
|
50
|
|
|
|
46
|
if (abs($e) < $epsilon) |
1305
|
|
|
|
|
|
|
{ |
1306
|
0
|
|
|
|
|
0
|
@x = cubic_roots($d, $c, $b, $a); |
1307
|
0
|
|
|
|
|
0
|
$ascending_flag = $temp_ascending_flag; |
1308
|
0
|
|
|
|
|
0
|
return (0, @x); |
1309
|
|
|
|
|
|
|
} |
1310
|
|
|
|
|
|
|
|
1311
|
|
|
|
|
|
|
# |
1312
|
|
|
|
|
|
|
# First step: Divide by the leading coefficient. |
1313
|
|
|
|
|
|
|
# |
1314
|
22
|
|
|
|
|
35
|
$b /= $a; |
1315
|
22
|
|
|
|
|
27
|
$c /= $a; |
1316
|
22
|
|
|
|
|
28
|
$d /= $a; |
1317
|
22
|
|
|
|
|
31
|
$e /= $a; |
1318
|
|
|
|
|
|
|
|
1319
|
|
|
|
|
|
|
# |
1320
|
|
|
|
|
|
|
# Second step: simplify the equation to the |
1321
|
|
|
|
|
|
|
# "resolvent cubic" y**4 + fy**2 + gy + h. |
1322
|
|
|
|
|
|
|
# |
1323
|
|
|
|
|
|
|
# (This is done by setting x = y - b/4). |
1324
|
|
|
|
|
|
|
# |
1325
|
22
|
|
|
|
|
33
|
my $b4 = $b/4; |
1326
|
|
|
|
|
|
|
|
1327
|
|
|
|
|
|
|
# |
1328
|
|
|
|
|
|
|
# The f, g, and h values are: |
1329
|
|
|
|
|
|
|
# |
1330
|
22
|
|
|
|
|
48
|
my $f = $c - |
1331
|
|
|
|
|
|
|
6 * $b4 * $b4; |
1332
|
22
|
|
|
|
|
47
|
my $g = $d + |
1333
|
|
|
|
|
|
|
2 * $b4 * (-$c + 4 * $b4 * $b4); |
1334
|
22
|
|
|
|
|
44
|
my $h = $e + |
1335
|
|
|
|
|
|
|
$b4 * (-$d + $b4 * ($c - 3 * $b4 * $b4)); |
1336
|
|
|
|
|
|
|
|
1337
|
|
|
|
|
|
|
# |
1338
|
|
|
|
|
|
|
### quartic_roots calculations |
1339
|
|
|
|
|
|
|
#### $b4 |
1340
|
|
|
|
|
|
|
#### $f |
1341
|
|
|
|
|
|
|
#### $g |
1342
|
|
|
|
|
|
|
#### $h |
1343
|
|
|
|
|
|
|
# |
1344
|
22
|
100
|
|
|
|
76
|
if (abs($h) < $epsilon) |
|
|
100
|
|
|
|
|
|
1345
|
|
|
|
|
|
|
{ |
1346
|
|
|
|
|
|
|
# |
1347
|
|
|
|
|
|
|
### Quartic branch 1, $h equals 0, within epsilon... |
1348
|
|
|
|
|
|
|
# |
1349
|
|
|
|
|
|
|
# Special case: h == 0. We have a cubic times y. |
1350
|
|
|
|
|
|
|
# |
1351
|
2
|
|
|
|
|
6
|
@x = (0, cubic_roots($g, $f, 0, 1)); |
1352
|
|
|
|
|
|
|
} |
1353
|
|
|
|
|
|
|
elsif (abs($g * $g) < $epsilon) |
1354
|
|
|
|
|
|
|
{ |
1355
|
|
|
|
|
|
|
# |
1356
|
|
|
|
|
|
|
### Quartic branch 2, $g equals 0, within epsilon... |
1357
|
|
|
|
|
|
|
# |
1358
|
|
|
|
|
|
|
# Another special case: g == 0. We have a quadratic |
1359
|
|
|
|
|
|
|
# with y-squared. |
1360
|
|
|
|
|
|
|
# |
1361
|
|
|
|
|
|
|
# (We check $g**2 because that's what the constant |
1362
|
|
|
|
|
|
|
# value actually is in Ferrari's method, and it is |
1363
|
|
|
|
|
|
|
# possible for $g to be outside of epsilon while |
1364
|
|
|
|
|
|
|
# $g**2 is inside, i.e., "zero"). |
1365
|
|
|
|
|
|
|
# |
1366
|
16
|
|
|
|
|
28
|
my($p, $q) = quadratic_roots($h, $f, 1); |
1367
|
16
|
|
|
|
|
1118
|
$p = sqrt($p); |
1368
|
16
|
|
|
|
|
813
|
$q = sqrt($q); |
1369
|
16
|
|
|
|
|
864
|
@x = ($p, -$p, $q, -$q); |
1370
|
|
|
|
|
|
|
} |
1371
|
|
|
|
|
|
|
else |
1372
|
|
|
|
|
|
|
{ |
1373
|
|
|
|
|
|
|
# |
1374
|
|
|
|
|
|
|
### Quartic branch 3, Ferrari's method... |
1375
|
|
|
|
|
|
|
# |
1376
|
|
|
|
|
|
|
# Special cases don't apply, so continue on with Ferrari's |
1377
|
|
|
|
|
|
|
# method. This involves setting up the resolvent cubic |
1378
|
|
|
|
|
|
|
# as the product of two quadratics. |
1379
|
|
|
|
|
|
|
# |
1380
|
|
|
|
|
|
|
# After setting up conditions that guarantee that the |
1381
|
|
|
|
|
|
|
# coefficients come out right (including the zero value |
1382
|
|
|
|
|
|
|
# for the third-power term), we wind up with a 6th |
1383
|
|
|
|
|
|
|
# degree polynomial with, fortunately, only even-powered |
1384
|
|
|
|
|
|
|
# terms. In other words, a cubic with z = y**2. |
1385
|
|
|
|
|
|
|
# |
1386
|
|
|
|
|
|
|
# Take a root of that equation, and get the |
1387
|
|
|
|
|
|
|
# quadratics from it. |
1388
|
|
|
|
|
|
|
# |
1389
|
4
|
|
|
|
|
5
|
my $z; |
1390
|
4
|
|
|
|
|
14
|
($z, undef, undef) = cubic_roots(-$g*$g, $f*$f - 4*$h, 2*$f, 1); |
1391
|
|
|
|
|
|
|
|
1392
|
|
|
|
|
|
|
#### $z |
1393
|
|
|
|
|
|
|
|
1394
|
4
|
|
|
|
|
14
|
my $alpha = sqrt($z); |
1395
|
4
|
|
|
|
|
89
|
my $rho = $g/$alpha; |
1396
|
4
|
|
|
|
|
68
|
my $beta = ($f + $z - $rho)/2; |
1397
|
4
|
|
|
|
|
169
|
my $gamma = ($f + $z + $rho)/2; |
1398
|
|
|
|
|
|
|
|
1399
|
4
|
|
|
|
|
162
|
@x = quadratic_roots($beta, $alpha, 1); |
1400
|
4
|
|
|
|
|
238
|
push @x, quadratic_roots($gamma, -$alpha, 1); |
1401
|
|
|
|
|
|
|
} |
1402
|
|
|
|
|
|
|
|
1403
|
22
|
|
|
|
|
1298
|
$ascending_flag = $temp_ascending_flag; |
1404
|
22
|
|
|
|
|
57
|
return ($x[0] - $b4, $x[1] - $b4, $x[2] - $b4, $x[3] - $b4); |
1405
|
|
|
|
|
|
|
} |
1406
|
|
|
|
|
|
|
|
1407
|
|
|
|
|
|
|
=head2 Sturm Functions |
1408
|
|
|
|
|
|
|
|
1409
|
|
|
|
|
|
|
These are the functions that create and make use of the Sturm sequence. |
1410
|
|
|
|
|
|
|
They are all exported under the tag "sturm". |
1411
|
|
|
|
|
|
|
|
1412
|
|
|
|
|
|
|
=head3 poly_real_root_count() |
1413
|
|
|
|
|
|
|
|
1414
|
|
|
|
|
|
|
Return the number of I, I roots of the polynomial. |
1415
|
|
|
|
|
|
|
|
1416
|
|
|
|
|
|
|
$unique_roots = poly_real_root_count(@coefficients); |
1417
|
|
|
|
|
|
|
|
1418
|
|
|
|
|
|
|
For example, the equation C<(x + 3)**3> forms the polynomial |
1419
|
|
|
|
|
|
|
C, but since all three of its roots are identical, |
1420
|
|
|
|
|
|
|
C will return 1. |
1421
|
|
|
|
|
|
|
|
1422
|
|
|
|
|
|
|
Likewise, C will return 0 because the two roots |
1423
|
|
|
|
|
|
|
of C are both complex. |
1424
|
|
|
|
|
|
|
|
1425
|
|
|
|
|
|
|
This function is the all-in-one function to use instead of |
1426
|
|
|
|
|
|
|
|
1427
|
|
|
|
|
|
|
my @chain = poly_sturm_chain(@coefficients); |
1428
|
|
|
|
|
|
|
|
1429
|
|
|
|
|
|
|
return sturm_sign_count(sturm_sign_minus_inf(\@chain)) - |
1430
|
|
|
|
|
|
|
sturm_sign_count(sturm_sign_plus_inf(\@chain)); |
1431
|
|
|
|
|
|
|
|
1432
|
|
|
|
|
|
|
if you don't intend to use the Sturm chain for anything else. |
1433
|
|
|
|
|
|
|
|
1434
|
|
|
|
|
|
|
=cut |
1435
|
|
|
|
|
|
|
|
1436
|
|
|
|
|
|
|
sub poly_real_root_count |
1437
|
|
|
|
|
|
|
{ |
1438
|
17
|
|
|
17
|
1
|
4048
|
my @coefficients = @_; |
1439
|
|
|
|
|
|
|
|
1440
|
17
|
|
|
|
|
32
|
my @chain = poly_sturm_chain(@coefficients); |
1441
|
|
|
|
|
|
|
|
1442
|
17
|
|
|
|
|
36
|
return sturm_sign_count(sturm_sign_minus_inf(\@chain)) - |
1443
|
|
|
|
|
|
|
sturm_sign_count(sturm_sign_plus_inf(\@chain)); |
1444
|
|
|
|
|
|
|
} |
1445
|
|
|
|
|
|
|
|
1446
|
|
|
|
|
|
|
=head3 sturm_real_root_range_count() |
1447
|
|
|
|
|
|
|
|
1448
|
|
|
|
|
|
|
Return the number of I, I roots of the polynomial between two X values. |
1449
|
|
|
|
|
|
|
|
1450
|
|
|
|
|
|
|
my($x0, $x1) = (0, 1000); |
1451
|
|
|
|
|
|
|
|
1452
|
|
|
|
|
|
|
my @chain = poly_sturm_chain(@coefficients); |
1453
|
|
|
|
|
|
|
$root_count = sturm_real_root_range_count(\@chain, $x0, $x1); |
1454
|
|
|
|
|
|
|
|
1455
|
|
|
|
|
|
|
This is equivalent to: |
1456
|
|
|
|
|
|
|
|
1457
|
|
|
|
|
|
|
my($x0, $x1) = (0, 1000); |
1458
|
|
|
|
|
|
|
|
1459
|
|
|
|
|
|
|
my @chain = poly_sturm_chain(@coefficients); |
1460
|
|
|
|
|
|
|
my @signs = sturm_sign_chain(\@chain, [$x0, $x1]); |
1461
|
|
|
|
|
|
|
$root_count = sturm_sign_count(@{$signs[0]}) - sturm_sign_count(@{$signs[1]}); |
1462
|
|
|
|
|
|
|
|
1463
|
|
|
|
|
|
|
=cut |
1464
|
|
|
|
|
|
|
|
1465
|
|
|
|
|
|
|
sub sturm_real_root_range_count |
1466
|
|
|
|
|
|
|
{ |
1467
|
156
|
|
|
156
|
1
|
249
|
my($chain_ref, $x0, $x1) = @_; |
1468
|
|
|
|
|
|
|
|
1469
|
156
|
|
|
|
|
285
|
my @signs = sturm_sign_chain($chain_ref, [$x0, $x1]); |
1470
|
|
|
|
|
|
|
|
1471
|
156
|
|
|
|
|
219
|
my $count0 = sturm_sign_count(@{$signs[0]}); |
|
156
|
|
|
|
|
241
|
|
1472
|
156
|
|
|
|
|
178
|
my $count1 = sturm_sign_count(@{$signs[1]}); |
|
156
|
|
|
|
|
218
|
|
1473
|
|
|
|
|
|
|
|
1474
|
|
|
|
|
|
|
# |
1475
|
|
|
|
|
|
|
###### (from, to): join(", ", ($x0, $x1)) |
1476
|
|
|
|
|
|
|
###### sign count from: $count0 |
1477
|
|
|
|
|
|
|
###### sign count to: $count1 |
1478
|
|
|
|
|
|
|
# |
1479
|
156
|
|
|
|
|
302
|
return $count0 - $count1; |
1480
|
|
|
|
|
|
|
} |
1481
|
|
|
|
|
|
|
|
1482
|
|
|
|
|
|
|
|
1483
|
|
|
|
|
|
|
=head3 sturm_bisection() |
1484
|
|
|
|
|
|
|
|
1485
|
|
|
|
|
|
|
Finds the boundaries around the roots of a polynomial function, |
1486
|
|
|
|
|
|
|
using the root count method of Sturm. |
1487
|
|
|
|
|
|
|
|
1488
|
|
|
|
|
|
|
@boundaries = sturm_bisection(\@chain, $from, $to); |
1489
|
|
|
|
|
|
|
|
1490
|
|
|
|
|
|
|
The elements of @boundaries will be a list of two-element arrays, each |
1491
|
|
|
|
|
|
|
one bracketing a root. |
1492
|
|
|
|
|
|
|
|
1493
|
|
|
|
|
|
|
It will not bracket complex roots. |
1494
|
|
|
|
|
|
|
|
1495
|
|
|
|
|
|
|
This allows you to use a different root-finding function than laguerre(), |
1496
|
|
|
|
|
|
|
which is the default function used by sturm_bisection_roots(). |
1497
|
|
|
|
|
|
|
|
1498
|
|
|
|
|
|
|
=cut |
1499
|
|
|
|
|
|
|
|
1500
|
|
|
|
|
|
|
sub sturm_bisection |
1501
|
|
|
|
|
|
|
{ |
1502
|
21
|
|
|
21
|
1
|
31
|
my($chain_ref, $from, $to) = @_; |
1503
|
21
|
|
|
|
|
24
|
my(@coefficients) = @{${$chain_ref}[0]}; |
|
21
|
|
|
|
|
54
|
|
|
21
|
|
|
|
|
42
|
|
1504
|
21
|
|
|
|
|
26
|
my @boundaries; |
1505
|
|
|
|
|
|
|
|
1506
|
|
|
|
|
|
|
# |
1507
|
|
|
|
|
|
|
#### @coefficients |
1508
|
|
|
|
|
|
|
# |
1509
|
|
|
|
|
|
|
# |
1510
|
|
|
|
|
|
|
# If we have a linear equation, just solve the thing. We're not |
1511
|
|
|
|
|
|
|
# going to find a useful second derivative, after all. (Which |
1512
|
|
|
|
|
|
|
# would raise the question of why we're here without a useful |
1513
|
|
|
|
|
|
|
# Sturm chain, but never mind...) |
1514
|
|
|
|
|
|
|
# |
1515
|
21
|
50
|
|
|
|
35
|
if ($#coefficients == 1) |
1516
|
|
|
|
|
|
|
{ |
1517
|
0
|
|
|
|
|
0
|
my $root = linear_roots(@coefficients); |
1518
|
|
|
|
|
|
|
|
1519
|
|
|
|
|
|
|
# |
1520
|
|
|
|
|
|
|
# But make sure the root is within the |
1521
|
|
|
|
|
|
|
# asked-for range. |
1522
|
|
|
|
|
|
|
# |
1523
|
0
|
0
|
0
|
|
|
0
|
return () if ($root < $from or $root > $to); |
1524
|
0
|
|
|
|
|
0
|
return ([$root, $root]); |
1525
|
|
|
|
|
|
|
} |
1526
|
|
|
|
|
|
|
|
1527
|
|
|
|
|
|
|
# |
1528
|
|
|
|
|
|
|
# Do Sturm bisection here. |
1529
|
|
|
|
|
|
|
# |
1530
|
21
|
|
|
|
|
34
|
my $range_count = sturm_real_root_range_count($chain_ref, $from, $to); |
1531
|
|
|
|
|
|
|
|
1532
|
|
|
|
|
|
|
# |
1533
|
|
|
|
|
|
|
# If we're down to one root in this range, use Laguerre's method |
1534
|
|
|
|
|
|
|
# to hunt it down. |
1535
|
|
|
|
|
|
|
# |
1536
|
21
|
50
|
|
|
|
35
|
return () if ($range_count == 0); |
1537
|
21
|
100
|
|
|
|
44
|
return ([$from, $to]) if ($range_count == 1); |
1538
|
|
|
|
|
|
|
|
1539
|
|
|
|
|
|
|
# |
1540
|
|
|
|
|
|
|
# More than one root in this range, so subdivide |
1541
|
|
|
|
|
|
|
# until each root has its own range. |
1542
|
|
|
|
|
|
|
# |
1543
|
8
|
|
|
|
|
10
|
my $its = 0; |
1544
|
|
|
|
|
|
|
|
1545
|
|
|
|
|
|
|
ROOT: |
1546
|
8
|
|
|
|
|
12
|
for (;;) |
1547
|
|
|
|
|
|
|
{ |
1548
|
59
|
|
|
|
|
89
|
my $mid = ($to + $from)/2.0; |
1549
|
59
|
|
|
|
|
81
|
my $frommid_count = sturm_real_root_range_count($chain_ref, $from, $mid); |
1550
|
59
|
|
|
|
|
87
|
my $midto_count = sturm_real_root_range_count($chain_ref, $mid, $to); |
1551
|
|
|
|
|
|
|
|
1552
|
|
|
|
|
|
|
# |
1553
|
|
|
|
|
|
|
#### $its |
1554
|
|
|
|
|
|
|
#### $from |
1555
|
|
|
|
|
|
|
#### $mid |
1556
|
|
|
|
|
|
|
#### $to |
1557
|
|
|
|
|
|
|
#### $frommid_count |
1558
|
|
|
|
|
|
|
#### $midto_count |
1559
|
|
|
|
|
|
|
# |
1560
|
|
|
|
|
|
|
|
1561
|
|
|
|
|
|
|
# |
1562
|
|
|
|
|
|
|
# Bisect again if we only narrowed down to a range |
1563
|
|
|
|
|
|
|
# containing all the roots. |
1564
|
|
|
|
|
|
|
# |
1565
|
59
|
100
|
|
|
|
98
|
if ($frommid_count == 0) |
|
|
100
|
|
|
|
|
|
1566
|
|
|
|
|
|
|
{ |
1567
|
39
|
|
|
|
|
47
|
$from = $mid; |
1568
|
|
|
|
|
|
|
} |
1569
|
|
|
|
|
|
|
elsif ($midto_count == 0) |
1570
|
|
|
|
|
|
|
{ |
1571
|
12
|
|
|
|
|
15
|
$to = $mid; |
1572
|
|
|
|
|
|
|
} |
1573
|
|
|
|
|
|
|
else |
1574
|
|
|
|
|
|
|
{ |
1575
|
|
|
|
|
|
|
# |
1576
|
|
|
|
|
|
|
# We've divided the roots between two ranges. Do it |
1577
|
|
|
|
|
|
|
# again until each range has a single root in it. |
1578
|
|
|
|
|
|
|
# |
1579
|
8
|
|
|
|
|
19
|
push @boundaries, sturm_bisection($chain_ref, $from, $mid); |
1580
|
8
|
|
|
|
|
14
|
push @boundaries, sturm_bisection($chain_ref, $mid, $to); |
1581
|
8
|
|
|
|
|
12
|
last ROOT; |
1582
|
|
|
|
|
|
|
} |
1583
|
51
|
50
|
|
|
|
85
|
croak "Too many iterations ($its) at mid=$mid\n" if ($its >= $iteration{sturm_bisection}); |
1584
|
51
|
|
|
|
|
60
|
$its++; |
1585
|
|
|
|
|
|
|
} |
1586
|
8
|
|
|
|
|
14
|
return @boundaries; |
1587
|
|
|
|
|
|
|
} |
1588
|
|
|
|
|
|
|
|
1589
|
|
|
|
|
|
|
|
1590
|
|
|
|
|
|
|
=head3 sturm_bisection_roots() |
1591
|
|
|
|
|
|
|
|
1592
|
|
|
|
|
|
|
Return the I roots counted by L. |
1593
|
|
|
|
|
|
|
Uses L to bracket the roots of the polynomial, |
1594
|
|
|
|
|
|
|
then uses L to close in on each root. |
1595
|
|
|
|
|
|
|
|
1596
|
|
|
|
|
|
|
my($from, $to) = (-1000, 0); |
1597
|
|
|
|
|
|
|
my @chain = poly_sturm_chain(@coefficients); |
1598
|
|
|
|
|
|
|
my @roots = sturm_bisection_roots(\@chain, $from, $to); |
1599
|
|
|
|
|
|
|
|
1600
|
|
|
|
|
|
|
As it is using the Sturm functions, it will find only the real roots. |
1601
|
|
|
|
|
|
|
|
1602
|
|
|
|
|
|
|
=cut |
1603
|
|
|
|
|
|
|
|
1604
|
|
|
|
|
|
|
sub sturm_bisection_roots |
1605
|
|
|
|
|
|
|
{ |
1606
|
5
|
|
|
5
|
1
|
17
|
my($chain_ref, $from, $to) = @_; |
1607
|
5
|
|
|
|
|
6
|
my $cref0 = ${$chain_ref}[0]; |
|
5
|
|
|
|
|
6
|
|
1608
|
5
|
|
|
|
|
11
|
my @boundaries = sturm_bisection($chain_ref, $from, $to); |
1609
|
5
|
|
|
|
|
8
|
my @roots; |
1610
|
|
|
|
|
|
|
|
1611
|
5
|
|
|
|
|
6
|
my $temp_ascending_flag = $ascending_flag; |
1612
|
5
|
|
|
|
|
6
|
$ascending_flag = 1; |
1613
|
|
|
|
|
|
|
|
1614
|
|
|
|
|
|
|
# |
1615
|
|
|
|
|
|
|
#### sturm_bisection() returns: @boundaries |
1616
|
|
|
|
|
|
|
# |
1617
|
5
|
|
|
|
|
6
|
for my $bracket (@boundaries) |
1618
|
|
|
|
|
|
|
{ |
1619
|
13
|
|
|
|
|
22
|
my ($left, $right) = @$bracket; |
1620
|
13
|
|
|
|
|
23
|
push @roots, laguerre($cref0, ($left + $right)/2.0); |
1621
|
|
|
|
|
|
|
} |
1622
|
|
|
|
|
|
|
|
1623
|
5
|
|
|
|
|
8
|
$ascending_flag = $temp_ascending_flag; |
1624
|
|
|
|
|
|
|
|
1625
|
5
|
|
|
|
|
16
|
return @roots; |
1626
|
|
|
|
|
|
|
} |
1627
|
|
|
|
|
|
|
|
1628
|
|
|
|
|
|
|
|
1629
|
|
|
|
|
|
|
=head3 poly_sturm_chain() |
1630
|
|
|
|
|
|
|
|
1631
|
|
|
|
|
|
|
Returns the chain of Sturm functions used to evaluate the number of roots of a |
1632
|
|
|
|
|
|
|
polynomial in a range of X values. The chain is a list of coefficient |
1633
|
|
|
|
|
|
|
references, the coefficients being stored in ascending order. |
1634
|
|
|
|
|
|
|
|
1635
|
|
|
|
|
|
|
If you feed in a sequence of X values to the Sturm functions, you can tell where |
1636
|
|
|
|
|
|
|
the (real, not complex) roots of the polynomial are by counting the number of |
1637
|
|
|
|
|
|
|
times the Y values change sign. |
1638
|
|
|
|
|
|
|
|
1639
|
|
|
|
|
|
|
See L above for an example of its use. |
1640
|
|
|
|
|
|
|
|
1641
|
|
|
|
|
|
|
=cut |
1642
|
|
|
|
|
|
|
|
1643
|
|
|
|
|
|
|
sub poly_sturm_chain |
1644
|
|
|
|
|
|
|
{ |
1645
|
55
|
|
|
55
|
1
|
7546
|
my @coefficients = @_; |
1646
|
55
|
|
|
|
|
82
|
my $degree = $#coefficients; |
1647
|
55
|
|
|
|
|
115
|
my(@chain, @remd); |
1648
|
55
|
|
|
|
|
0
|
my($f1, $f2); |
1649
|
|
|
|
|
|
|
|
1650
|
55
|
100
|
|
|
|
106
|
@coefficients = reverse @coefficients unless ($ascending_flag); |
1651
|
|
|
|
|
|
|
|
1652
|
55
|
|
|
|
|
91
|
$f1 = [@coefficients]; |
1653
|
55
|
|
|
|
|
132
|
$f2 = pl_derivative(\@coefficients); |
1654
|
|
|
|
|
|
|
|
1655
|
55
|
|
|
|
|
580
|
push @chain, $f1; |
1656
|
|
|
|
|
|
|
|
1657
|
|
|
|
|
|
|
# |
1658
|
|
|
|
|
|
|
# Start the first links of the chain. |
1659
|
|
|
|
|
|
|
# |
1660
|
|
|
|
|
|
|
SKIPIT: { |
1661
|
55
|
100
|
|
|
|
66
|
last SKIPIT if ($degree < 1); #return @chain if ($degree < 1); |
|
55
|
|
|
|
|
105
|
|
1662
|
|
|
|
|
|
|
|
1663
|
52
|
|
|
|
|
64
|
push @chain, $f2; |
1664
|
|
|
|
|
|
|
|
1665
|
52
|
100
|
|
|
|
87
|
last SKIPIT if ($degree < 2); #return @chain if ($degree < 2); |
1666
|
|
|
|
|
|
|
|
1667
|
|
|
|
|
|
|
# |
1668
|
|
|
|
|
|
|
###### poly_sturm_chain chain before do loop: |
1669
|
|
|
|
|
|
|
###### @chain |
1670
|
|
|
|
|
|
|
# |
1671
|
|
|
|
|
|
|
do |
1672
|
49
|
|
|
|
|
61
|
{ |
1673
|
77
|
|
|
|
|
152
|
my ($q, $r) = pl_div($f1, $f2); |
1674
|
|
|
|
|
|
|
|
1675
|
|
|
|
|
|
|
# |
1676
|
|
|
|
|
|
|
# Remove any leading zeros in the remainder. |
1677
|
|
|
|
|
|
|
# |
1678
|
77
|
|
|
|
|
2081
|
@remd = @{$r}; |
|
77
|
|
|
|
|
130
|
|
1679
|
77
|
|
100
|
|
|
336
|
pop @remd while (@remd and abs($remd[$#remd]) < $epsilon); |
1680
|
|
|
|
|
|
|
|
1681
|
77
|
|
|
|
|
104
|
$f1 = $f2; |
1682
|
77
|
100
|
|
|
|
147
|
$f2 = (@remd)? [map {$_ * -1} @remd]: [0]; |
|
103
|
|
|
|
|
188
|
|
1683
|
77
|
|
|
|
|
217
|
push @chain, $f2; |
1684
|
|
|
|
|
|
|
} |
1685
|
|
|
|
|
|
|
while ($#remd > 0); |
1686
|
|
|
|
|
|
|
} |
1687
|
|
|
|
|
|
|
|
1688
|
|
|
|
|
|
|
# |
1689
|
|
|
|
|
|
|
###### poly_sturm_chain: |
1690
|
|
|
|
|
|
|
###### @chain |
1691
|
|
|
|
|
|
|
# |
1692
|
55
|
|
|
|
|
131
|
return @chain; |
1693
|
|
|
|
|
|
|
} |
1694
|
|
|
|
|
|
|
|
1695
|
|
|
|
|
|
|
=head3 sturm_sign_count() |
1696
|
|
|
|
|
|
|
|
1697
|
|
|
|
|
|
|
Counts and returns the number of sign changes in a sequence of signs, |
1698
|
|
|
|
|
|
|
such as those returned by the L |
1699
|
|
|
|
|
|
|
|
1700
|
|
|
|
|
|
|
See L and L for |
1701
|
|
|
|
|
|
|
examples of its use. |
1702
|
|
|
|
|
|
|
|
1703
|
|
|
|
|
|
|
=cut |
1704
|
|
|
|
|
|
|
|
1705
|
|
|
|
|
|
|
sub sturm_sign_count |
1706
|
|
|
|
|
|
|
{ |
1707
|
346
|
|
|
346
|
1
|
464
|
my @sign_seq = @_; |
1708
|
346
|
|
|
|
|
381
|
my $scnt = 0; |
1709
|
|
|
|
|
|
|
|
1710
|
346
|
|
|
|
|
394
|
my $s1 = shift @sign_seq; |
1711
|
346
|
|
|
|
|
450
|
for my $s2 (@sign_seq) |
1712
|
|
|
|
|
|
|
{ |
1713
|
1060
|
100
|
|
|
|
1494
|
$scnt++ if ($s1 != $s2); |
1714
|
1060
|
|
|
|
|
1234
|
$s1 = $s2; |
1715
|
|
|
|
|
|
|
} |
1716
|
|
|
|
|
|
|
|
1717
|
346
|
|
|
|
|
820
|
return $scnt; |
1718
|
|
|
|
|
|
|
} |
1719
|
|
|
|
|
|
|
|
1720
|
|
|
|
|
|
|
|
1721
|
|
|
|
|
|
|
=head3 Sturm Sign Sequence Functions |
1722
|
|
|
|
|
|
|
|
1723
|
|
|
|
|
|
|
=head4 sturm_sign_chain() |
1724
|
|
|
|
|
|
|
|
1725
|
|
|
|
|
|
|
=head4 sturm_sign_minus_inf() |
1726
|
|
|
|
|
|
|
|
1727
|
|
|
|
|
|
|
=head4 sturm_sign_plus_inf() |
1728
|
|
|
|
|
|
|
|
1729
|
|
|
|
|
|
|
These functions return the array of signs that are used by the functions |
1730
|
|
|
|
|
|
|
L and L to find |
1731
|
|
|
|
|
|
|
the number of real roots in a polynomial. |
1732
|
|
|
|
|
|
|
|
1733
|
|
|
|
|
|
|
In normal use you will probably never need to use them, unless you want |
1734
|
|
|
|
|
|
|
to examine the internals of the Sturm functions: |
1735
|
|
|
|
|
|
|
|
1736
|
|
|
|
|
|
|
# |
1737
|
|
|
|
|
|
|
# Examine the sign changes that occur at each endpoint of |
1738
|
|
|
|
|
|
|
# the x range. |
1739
|
|
|
|
|
|
|
# |
1740
|
|
|
|
|
|
|
my(@coefficients) = (1, 4, 7, 23); |
1741
|
|
|
|
|
|
|
my(@xvals) = (-12, 12); |
1742
|
|
|
|
|
|
|
|
1743
|
|
|
|
|
|
|
my @chain = poly_sturm_chain( @coefficients); |
1744
|
|
|
|
|
|
|
my @signs = sturm_sign_chain(\@chain, \@xvals); # An array of arrays. |
1745
|
|
|
|
|
|
|
|
1746
|
|
|
|
|
|
|
print "\nPolynomial: [", join(", ", @coefficients), "]\n"; |
1747
|
|
|
|
|
|
|
|
1748
|
|
|
|
|
|
|
for my $j (0..$#signs) |
1749
|
|
|
|
|
|
|
{ |
1750
|
|
|
|
|
|
|
my @s = @{$signs[$j]}; |
1751
|
|
|
|
|
|
|
print $xval[$j], "\n", |
1752
|
|
|
|
|
|
|
"\t", join(", ", @s), "], sign count = ", |
1753
|
|
|
|
|
|
|
sturm_sign_count(@s), "\n\n"; |
1754
|
|
|
|
|
|
|
} |
1755
|
|
|
|
|
|
|
|
1756
|
|
|
|
|
|
|
Similar examinations can be made at plus and minus infinity: |
1757
|
|
|
|
|
|
|
|
1758
|
|
|
|
|
|
|
# |
1759
|
|
|
|
|
|
|
# Examine the sign changes that occur between plus and minus |
1760
|
|
|
|
|
|
|
# infinity. |
1761
|
|
|
|
|
|
|
# |
1762
|
|
|
|
|
|
|
my @coefficients = (1, 4, 7, 23); |
1763
|
|
|
|
|
|
|
|
1764
|
|
|
|
|
|
|
my @chain = poly_sturm_chain( @coefficients); |
1765
|
|
|
|
|
|
|
my @smi = sturm_sign_minus_inf(\@chain); |
1766
|
|
|
|
|
|
|
my @spi = sturm_sign_plus_inf(\@chain); |
1767
|
|
|
|
|
|
|
|
1768
|
|
|
|
|
|
|
print "\nPolynomial: [", join(", ", @coefficients), "]\n"; |
1769
|
|
|
|
|
|
|
|
1770
|
|
|
|
|
|
|
print "Minus Inf:\n", |
1771
|
|
|
|
|
|
|
"\t", join(", ", @smi), "], sign count = ", |
1772
|
|
|
|
|
|
|
sturm_sign_count(@smi), "\n\n"; |
1773
|
|
|
|
|
|
|
|
1774
|
|
|
|
|
|
|
print "Plus Inf:\n", |
1775
|
|
|
|
|
|
|
"\t", join(", ", @spi), "], sign count = ", |
1776
|
|
|
|
|
|
|
sturm_sign_count(@spi), "\n\n"; |
1777
|
|
|
|
|
|
|
|
1778
|
|
|
|
|
|
|
=cut |
1779
|
|
|
|
|
|
|
|
1780
|
|
|
|
|
|
|
# |
1781
|
|
|
|
|
|
|
# @signs = sturm_minus_inf(\@chain); |
1782
|
|
|
|
|
|
|
# |
1783
|
|
|
|
|
|
|
# Return an array of signs from the chain at minus infinity. |
1784
|
|
|
|
|
|
|
# |
1785
|
|
|
|
|
|
|
# In the :sturm export set. |
1786
|
|
|
|
|
|
|
# |
1787
|
|
|
|
|
|
|
sub sturm_sign_minus_inf |
1788
|
|
|
|
|
|
|
{ |
1789
|
17
|
|
|
17
|
1
|
26
|
my($chain_ref) = @_; |
1790
|
17
|
|
|
|
|
22
|
my @signs; |
1791
|
|
|
|
|
|
|
|
1792
|
17
|
|
|
|
|
27
|
for my $c (@$chain_ref) |
1793
|
|
|
|
|
|
|
{ |
1794
|
56
|
|
|
|
|
178
|
my @coefficients = @$c; |
1795
|
56
|
100
|
|
|
|
102
|
push @signs, sign($coefficients[$#coefficients]) * |
1796
|
|
|
|
|
|
|
((($#coefficients & 1) == 1)? -1: 1); |
1797
|
|
|
|
|
|
|
} |
1798
|
|
|
|
|
|
|
|
1799
|
17
|
|
|
|
|
73
|
return @signs; |
1800
|
|
|
|
|
|
|
} |
1801
|
|
|
|
|
|
|
|
1802
|
|
|
|
|
|
|
# |
1803
|
|
|
|
|
|
|
# @signs = sturm_plus_inf(\@chain); |
1804
|
|
|
|
|
|
|
# |
1805
|
|
|
|
|
|
|
# Return an array of signs from the chain at infinity. |
1806
|
|
|
|
|
|
|
# |
1807
|
|
|
|
|
|
|
# In the :sturm export set. |
1808
|
|
|
|
|
|
|
# |
1809
|
|
|
|
|
|
|
sub sturm_sign_plus_inf |
1810
|
|
|
|
|
|
|
{ |
1811
|
17
|
|
|
17
|
1
|
23
|
my($chain_ref) = @_; |
1812
|
17
|
|
|
|
|
21
|
my @signs; |
1813
|
|
|
|
|
|
|
|
1814
|
17
|
|
|
|
|
23
|
for my $c (@$chain_ref) |
1815
|
|
|
|
|
|
|
{ |
1816
|
56
|
|
|
|
|
231
|
my @coefficients = @$c; |
1817
|
56
|
|
|
|
|
93
|
push @signs, sign($coefficients[$#coefficients]); |
1818
|
|
|
|
|
|
|
} |
1819
|
|
|
|
|
|
|
|
1820
|
17
|
|
|
|
|
82
|
return @signs; |
1821
|
|
|
|
|
|
|
} |
1822
|
|
|
|
|
|
|
|
1823
|
|
|
|
|
|
|
# |
1824
|
|
|
|
|
|
|
# @sign_chains = sturm_sign_chain(\@chain, \@xvals); |
1825
|
|
|
|
|
|
|
# |
1826
|
|
|
|
|
|
|
# Return an array of signs for each x-value passed in each function in |
1827
|
|
|
|
|
|
|
# the Sturm chain. |
1828
|
|
|
|
|
|
|
# |
1829
|
|
|
|
|
|
|
# In the :sturm export set. |
1830
|
|
|
|
|
|
|
# |
1831
|
|
|
|
|
|
|
sub sturm_sign_chain |
1832
|
|
|
|
|
|
|
{ |
1833
|
156
|
|
|
156
|
1
|
197
|
my($chain_ref, $xvals_ref) = @_; |
1834
|
156
|
|
|
|
|
197
|
my $fn_count = $#$chain_ref; |
1835
|
156
|
|
|
|
|
192
|
my $x_count = $#$xvals_ref; |
1836
|
156
|
|
|
|
|
207
|
my @sign_chain; |
1837
|
|
|
|
|
|
|
|
1838
|
156
|
|
|
|
|
341
|
push @sign_chain, [] for (0..$x_count); |
1839
|
|
|
|
|
|
|
|
1840
|
156
|
|
|
|
|
214
|
for my $p_ref (@$chain_ref) |
1841
|
|
|
|
|
|
|
{ |
1842
|
647
|
|
|
|
|
1050
|
my @ysigns = sign(pl_evaluate($p_ref, $xvals_ref)); |
1843
|
|
|
|
|
|
|
|
1844
|
|
|
|
|
|
|
# |
1845
|
|
|
|
|
|
|
# We just retrieved the signs of a single function across |
1846
|
|
|
|
|
|
|
# our x-vals. We want it the other way around; signs listed |
1847
|
|
|
|
|
|
|
# by x-val across functions. |
1848
|
|
|
|
|
|
|
# |
1849
|
|
|
|
|
|
|
# (list of lists) |
1850
|
|
|
|
|
|
|
# | |
1851
|
|
|
|
|
|
|
# v |
1852
|
|
|
|
|
|
|
# f0 f1 f2 f3 f4 ... |
1853
|
|
|
|
|
|
|
# x0 - - + - + (list 0) |
1854
|
|
|
|
|
|
|
# |
1855
|
|
|
|
|
|
|
# x1 + - - + + (list 1) |
1856
|
|
|
|
|
|
|
# |
1857
|
|
|
|
|
|
|
# x2 + - + + + (list 2) |
1858
|
|
|
|
|
|
|
# |
1859
|
|
|
|
|
|
|
# ... |
1860
|
|
|
|
|
|
|
# |
1861
|
647
|
|
|
|
|
12900
|
for my $j (0..$x_count) |
1862
|
|
|
|
|
|
|
{ |
1863
|
1294
|
|
|
|
|
1412
|
push @{$sign_chain[$j]}, shift @ysigns; |
|
1294
|
|
|
|
|
1978
|
|
1864
|
|
|
|
|
|
|
} |
1865
|
|
|
|
|
|
|
} |
1866
|
|
|
|
|
|
|
|
1867
|
|
|
|
|
|
|
# |
1868
|
|
|
|
|
|
|
###### sturm_sign_chain() returns |
1869
|
|
|
|
|
|
|
###### @sign_chain: @sign_chain |
1870
|
|
|
|
|
|
|
# |
1871
|
156
|
|
|
|
|
240
|
return @sign_chain; |
1872
|
|
|
|
|
|
|
} |
1873
|
|
|
|
|
|
|
|
1874
|
|
|
|
|
|
|
|
1875
|
|
|
|
|
|
|
=head2 Utility Functions |
1876
|
|
|
|
|
|
|
|
1877
|
|
|
|
|
|
|
These are internal functions used by the other functions listed above |
1878
|
|
|
|
|
|
|
that may also be useful to the user, or which affect the behavior of |
1879
|
|
|
|
|
|
|
other functions. They are all exported under the tag "utility". |
1880
|
|
|
|
|
|
|
|
1881
|
|
|
|
|
|
|
=head3 epsilon() |
1882
|
|
|
|
|
|
|
|
1883
|
|
|
|
|
|
|
Returns the machine epsilon value that was calculated when this module was |
1884
|
|
|
|
|
|
|
loaded. |
1885
|
|
|
|
|
|
|
|
1886
|
|
|
|
|
|
|
The value may be changed, although this in general is not recommended. |
1887
|
|
|
|
|
|
|
|
1888
|
|
|
|
|
|
|
my $old_epsilon = epsilon($new_epsilon); |
1889
|
|
|
|
|
|
|
|
1890
|
|
|
|
|
|
|
The previous value of epsilon may be saved to be restored later. |
1891
|
|
|
|
|
|
|
|
1892
|
|
|
|
|
|
|
The Wikipedia article at L has |
1893
|
|
|
|
|
|
|
more information on the subject. |
1894
|
|
|
|
|
|
|
|
1895
|
|
|
|
|
|
|
=cut |
1896
|
|
|
|
|
|
|
|
1897
|
|
|
|
|
|
|
sub epsilon |
1898
|
|
|
|
|
|
|
{ |
1899
|
0
|
|
|
0
|
1
|
0
|
my $eps = $epsilon; |
1900
|
0
|
0
|
|
|
|
0
|
$epsilon = $_[0] if (scalar @_ > 0); |
1901
|
0
|
|
|
|
|
0
|
return $eps; |
1902
|
|
|
|
|
|
|
} |
1903
|
|
|
|
|
|
|
|
1904
|
|
|
|
|
|
|
=head3 laguerre() |
1905
|
|
|
|
|
|
|
|
1906
|
|
|
|
|
|
|
A numerical method for finding a root of an equation, especially made |
1907
|
|
|
|
|
|
|
for polynomials. |
1908
|
|
|
|
|
|
|
|
1909
|
|
|
|
|
|
|
@roots = laguerre(\@coefficients, \@xvalues); |
1910
|
|
|
|
|
|
|
push @roots, laguerre(\@coefficients, $another_xvalue); |
1911
|
|
|
|
|
|
|
|
1912
|
|
|
|
|
|
|
For each x value the function will attempt to find a root closest to it. |
1913
|
|
|
|
|
|
|
The function will return real roots only. |
1914
|
|
|
|
|
|
|
|
1915
|
|
|
|
|
|
|
This is the function used by L after using |
1916
|
|
|
|
|
|
|
L to narrow its search to a range containing a single |
1917
|
|
|
|
|
|
|
root. |
1918
|
|
|
|
|
|
|
|
1919
|
|
|
|
|
|
|
=cut |
1920
|
|
|
|
|
|
|
|
1921
|
|
|
|
|
|
|
sub laguerre |
1922
|
|
|
|
|
|
|
{ |
1923
|
17
|
|
|
17
|
|
144
|
no Math::Complex; |
|
17
|
|
|
|
|
34
|
|
|
17
|
|
|
|
|
4296
|
|
1924
|
16
|
|
|
16
|
1
|
1668
|
my($p_ref, $xval_ref) = @_; |
1925
|
16
|
|
|
|
|
22
|
my $n = $#$p_ref; |
1926
|
16
|
|
|
|
|
22
|
my @xvalues; |
1927
|
|
|
|
|
|
|
my @roots; |
1928
|
|
|
|
|
|
|
|
1929
|
16
|
50
|
|
|
|
28
|
$p_ref = [reverse @$p_ref] unless ($ascending_flag); |
1930
|
|
|
|
|
|
|
|
1931
|
|
|
|
|
|
|
# |
1932
|
|
|
|
|
|
|
# Allow some flexibility in sending the x-values. |
1933
|
|
|
|
|
|
|
# |
1934
|
16
|
100
|
|
|
|
35
|
if (ref $xval_ref eq "ARRAY") |
1935
|
|
|
|
|
|
|
{ |
1936
|
3
|
|
|
|
|
4
|
@xvalues = @$xval_ref; |
1937
|
|
|
|
|
|
|
} |
1938
|
|
|
|
|
|
|
else |
1939
|
|
|
|
|
|
|
{ |
1940
|
|
|
|
|
|
|
# |
1941
|
|
|
|
|
|
|
# It could happen. Someone might type \$x instead of $x. |
1942
|
|
|
|
|
|
|
# |
1943
|
13
|
50
|
|
|
|
22
|
@xvalues = ((ref $xval_ref eq "SCALAR")? $$xval_ref: $xval_ref); |
1944
|
|
|
|
|
|
|
} |
1945
|
|
|
|
|
|
|
|
1946
|
16
|
|
|
|
|
25
|
for my $x (@xvalues) |
1947
|
|
|
|
|
|
|
{ |
1948
|
|
|
|
|
|
|
# |
1949
|
|
|
|
|
|
|
#### laguerre looking near: $x |
1950
|
|
|
|
|
|
|
#### Coefficient: @$p_ref |
1951
|
|
|
|
|
|
|
#### Degree: $n |
1952
|
|
|
|
|
|
|
# |
1953
|
21
|
|
|
|
|
25
|
my $its = 0; |
1954
|
|
|
|
|
|
|
|
1955
|
|
|
|
|
|
|
ROOT: |
1956
|
21
|
|
|
|
|
26
|
for (;;) |
1957
|
|
|
|
|
|
|
{ |
1958
|
|
|
|
|
|
|
# |
1959
|
|
|
|
|
|
|
# Get the values of the function and its first and |
1960
|
|
|
|
|
|
|
# second derivatives at X. |
1961
|
|
|
|
|
|
|
# |
1962
|
95
|
|
|
|
|
151
|
my($y, $dy, $d2y) = pl_dxevaluate($p_ref, $x); |
1963
|
|
|
|
|
|
|
|
1964
|
95
|
100
|
|
|
|
11420
|
if (abs($y) <= $tolerance{laguerre}) |
1965
|
|
|
|
|
|
|
{ |
1966
|
20
|
|
|
|
|
62
|
push @roots, $x; |
1967
|
20
|
|
|
|
|
41
|
last ROOT; |
1968
|
|
|
|
|
|
|
} |
1969
|
|
|
|
|
|
|
|
1970
|
|
|
|
|
|
|
# |
1971
|
|
|
|
|
|
|
#### At Iteration: $its |
1972
|
|
|
|
|
|
|
#### x: $x |
1973
|
|
|
|
|
|
|
#### f(x): $y |
1974
|
|
|
|
|
|
|
#### f'(x): $dy |
1975
|
|
|
|
|
|
|
#### f''(x): $d2y |
1976
|
|
|
|
|
|
|
# |
1977
|
75
|
|
|
|
|
211
|
my $g = $dy/$y; |
1978
|
75
|
|
|
|
|
514
|
my $h = $g * $g - $d2y/$y; |
1979
|
75
|
|
|
|
|
1207
|
my $f = sqrt(($n - 1) * ($n * $h - $g*$g)); |
1980
|
75
|
100
|
|
|
|
2441
|
$f = - $f if (abs($g - $f) > abs($g + $f)); |
1981
|
|
|
|
|
|
|
|
1982
|
|
|
|
|
|
|
# |
1983
|
|
|
|
|
|
|
#### g: $g |
1984
|
|
|
|
|
|
|
#### h: $h |
1985
|
|
|
|
|
|
|
#### f: $f |
1986
|
|
|
|
|
|
|
# |
1987
|
|
|
|
|
|
|
# Divide by the largest value of $g plus |
1988
|
|
|
|
|
|
|
# $f, bearing in mind that $f is the result |
1989
|
|
|
|
|
|
|
# of a square root function and may be positive |
1990
|
|
|
|
|
|
|
# or negative. |
1991
|
|
|
|
|
|
|
# |
1992
|
|
|
|
|
|
|
# Use the abs() function to determine size |
1993
|
|
|
|
|
|
|
# since $g or $f may be complex numbers. |
1994
|
|
|
|
|
|
|
# |
1995
|
75
|
|
|
|
|
1818
|
my $dx = $n/($g + $f); |
1996
|
|
|
|
|
|
|
|
1997
|
75
|
|
|
|
|
1012
|
$x -= $dx; |
1998
|
75
|
100
|
|
|
|
416
|
if (abs($dx) <= $tolerance{laguerre}) |
1999
|
|
|
|
|
|
|
{ |
2000
|
1
|
|
|
|
|
2
|
push @roots, $x; |
2001
|
1
|
|
|
|
|
3
|
last ROOT; |
2002
|
|
|
|
|
|
|
} |
2003
|
|
|
|
|
|
|
|
2004
|
74
|
50
|
|
|
|
249
|
croak "Too many iterations ($its) at dx=$dx\n" if ($its >= $iteration{laguerre}); |
2005
|
74
|
|
|
|
|
117
|
$its++; |
2006
|
|
|
|
|
|
|
} |
2007
|
|
|
|
|
|
|
|
2008
|
|
|
|
|
|
|
### root found at iteration $its |
2009
|
|
|
|
|
|
|
#### $x |
2010
|
|
|
|
|
|
|
} |
2011
|
|
|
|
|
|
|
|
2012
|
16
|
|
|
|
|
34
|
return @roots; |
2013
|
|
|
|
|
|
|
} |
2014
|
|
|
|
|
|
|
|
2015
|
|
|
|
|
|
|
|
2016
|
|
|
|
|
|
|
=head3 newtonraphson() |
2017
|
|
|
|
|
|
|
|
2018
|
|
|
|
|
|
|
Like L, a numerical method for finding a root of an equation. |
2019
|
|
|
|
|
|
|
|
2020
|
|
|
|
|
|
|
@roots = newtonraphson(\@coefficients, \@xvalues); |
2021
|
|
|
|
|
|
|
push @roots, newtonraphson(\@coefficients, $another_xvalue); |
2022
|
|
|
|
|
|
|
|
2023
|
|
|
|
|
|
|
For each x value the function will attempt to find a root closest to it. |
2024
|
|
|
|
|
|
|
The function will return real roots only. |
2025
|
|
|
|
|
|
|
|
2026
|
|
|
|
|
|
|
This function is provided as an alternative to laguerre(). It is not |
2027
|
|
|
|
|
|
|
used internally by any other functions. |
2028
|
|
|
|
|
|
|
|
2029
|
|
|
|
|
|
|
=cut |
2030
|
|
|
|
|
|
|
|
2031
|
|
|
|
|
|
|
sub newtonraphson |
2032
|
|
|
|
|
|
|
{ |
2033
|
17
|
|
|
17
|
|
111
|
no Math::Complex; |
|
17
|
|
|
|
|
33
|
|
|
17
|
|
|
|
|
8845
|
|
2034
|
2
|
|
|
2
|
1
|
874
|
my($p_ref, $xval_ref) = @_; |
2035
|
2
|
|
|
|
|
4
|
my $n = $#$p_ref; |
2036
|
2
|
|
|
|
|
3
|
my @xvalues; |
2037
|
|
|
|
|
|
|
my @roots; |
2038
|
|
|
|
|
|
|
|
2039
|
2
|
50
|
|
|
|
5
|
$p_ref = [reverse @$p_ref] unless ($ascending_flag); |
2040
|
|
|
|
|
|
|
|
2041
|
|
|
|
|
|
|
# |
2042
|
|
|
|
|
|
|
# Allow some flexibility in sending the x-values. |
2043
|
|
|
|
|
|
|
# |
2044
|
2
|
50
|
|
|
|
5
|
if (ref $xval_ref eq "ARRAY") |
2045
|
|
|
|
|
|
|
{ |
2046
|
2
|
|
|
|
|
4
|
@xvalues = @$xval_ref; |
2047
|
|
|
|
|
|
|
} |
2048
|
|
|
|
|
|
|
else |
2049
|
|
|
|
|
|
|
{ |
2050
|
|
|
|
|
|
|
# |
2051
|
|
|
|
|
|
|
# It could happen. Someone might type \$x instead of $x. |
2052
|
|
|
|
|
|
|
# |
2053
|
0
|
0
|
|
|
|
0
|
@xvalues = ((ref $xval_ref eq "SCALAR")? $$xval_ref: $xval_ref); |
2054
|
|
|
|
|
|
|
} |
2055
|
|
|
|
|
|
|
|
2056
|
|
|
|
|
|
|
# |
2057
|
|
|
|
|
|
|
### newtonraphson() |
2058
|
|
|
|
|
|
|
#### @xvalues |
2059
|
|
|
|
|
|
|
# |
2060
|
2
|
|
|
|
|
5
|
for my $x (@xvalues) |
2061
|
|
|
|
|
|
|
{ |
2062
|
6
|
|
|
|
|
7
|
my $its = 0; |
2063
|
|
|
|
|
|
|
|
2064
|
|
|
|
|
|
|
ROOT: |
2065
|
6
|
|
|
|
|
7
|
for (;;) |
2066
|
|
|
|
|
|
|
{ |
2067
|
|
|
|
|
|
|
# |
2068
|
|
|
|
|
|
|
# Get the values of the function and its |
2069
|
|
|
|
|
|
|
# first derivative at X. |
2070
|
|
|
|
|
|
|
# |
2071
|
40
|
|
|
|
|
64
|
my($y, $dy, undef) = pl_dxevaluate($p_ref, $x); |
2072
|
40
|
|
|
|
|
719
|
my $dx = $y/$dy; |
2073
|
40
|
|
|
|
|
49
|
$x -= $dx; |
2074
|
|
|
|
|
|
|
|
2075
|
40
|
100
|
|
|
|
69
|
if (abs($dx) <= $tolerance{newtonraphson}) |
2076
|
|
|
|
|
|
|
{ |
2077
|
6
|
|
|
|
|
8
|
push @roots, $x; |
2078
|
6
|
|
|
|
|
11
|
last ROOT; |
2079
|
|
|
|
|
|
|
} |
2080
|
|
|
|
|
|
|
|
2081
|
|
|
|
|
|
|
# |
2082
|
|
|
|
|
|
|
#### At Iteration: $its |
2083
|
|
|
|
|
|
|
#### x: $x |
2084
|
|
|
|
|
|
|
#### f(x): $y |
2085
|
|
|
|
|
|
|
#### f'(x): $dy |
2086
|
|
|
|
|
|
|
# |
2087
|
34
|
50
|
|
|
|
47
|
croak "Too many iterations ($its) at dx=$dx\n" if ($its >= $iteration{newtonraphson}); |
2088
|
34
|
|
|
|
|
40
|
$its++; |
2089
|
|
|
|
|
|
|
} |
2090
|
|
|
|
|
|
|
|
2091
|
|
|
|
|
|
|
### root found at iteration $its |
2092
|
|
|
|
|
|
|
#### $x |
2093
|
|
|
|
|
|
|
} |
2094
|
|
|
|
|
|
|
|
2095
|
2
|
|
|
|
|
8
|
return @roots; |
2096
|
|
|
|
|
|
|
} |
2097
|
|
|
|
|
|
|
|
2098
|
|
|
|
|
|
|
=head3 poly_iteration() |
2099
|
|
|
|
|
|
|
|
2100
|
|
|
|
|
|
|
Sets the limit to the number of iterations that a solving method may go |
2101
|
|
|
|
|
|
|
through before giving up trying to find a root. Each method of root-finding |
2102
|
|
|
|
|
|
|
used by L, L, and L |
2103
|
|
|
|
|
|
|
has its own iteration limit, which may be found, like L, |
2104
|
|
|
|
|
|
|
simply by looking at the return value of poly_iteration(). |
2105
|
|
|
|
|
|
|
|
2106
|
|
|
|
|
|
|
# |
2107
|
|
|
|
|
|
|
# Get all of the current iteration limits. |
2108
|
|
|
|
|
|
|
# |
2109
|
|
|
|
|
|
|
my %its_limits = poly_iteration(); |
2110
|
|
|
|
|
|
|
|
2111
|
|
|
|
|
|
|
# |
2112
|
|
|
|
|
|
|
# Double the limit for the hessenberg method, but set the limit |
2113
|
|
|
|
|
|
|
# for Laguerre's method to 20. |
2114
|
|
|
|
|
|
|
# |
2115
|
|
|
|
|
|
|
my %old_limits = poly_iteration(hessenberg => $its_limits{hessenberg} * 2, |
2116
|
|
|
|
|
|
|
laguerre => 20); |
2117
|
|
|
|
|
|
|
|
2118
|
|
|
|
|
|
|
# |
2119
|
|
|
|
|
|
|
# Reset the limits with the former values, but save the values we had |
2120
|
|
|
|
|
|
|
# for later. |
2121
|
|
|
|
|
|
|
# |
2122
|
|
|
|
|
|
|
my %hl_limits = poly_iteration(%old_limits); |
2123
|
|
|
|
|
|
|
|
2124
|
|
|
|
|
|
|
There are iteration limit values for: |
2125
|
|
|
|
|
|
|
|
2126
|
|
|
|
|
|
|
=over 4 |
2127
|
|
|
|
|
|
|
|
2128
|
|
|
|
|
|
|
=item 'hessenberg' |
2129
|
|
|
|
|
|
|
|
2130
|
|
|
|
|
|
|
The numeric method used by poly_roots(), if the hessenberg option is set. |
2131
|
|
|
|
|
|
|
Its default value is 60. |
2132
|
|
|
|
|
|
|
|
2133
|
|
|
|
|
|
|
=item 'laguerre' |
2134
|
|
|
|
|
|
|
|
2135
|
|
|
|
|
|
|
The numeric method used by L. Laguerre's method is used within |
2136
|
|
|
|
|
|
|
sturm_bisection_roots() once it has narrowed its search in on an individual |
2137
|
|
|
|
|
|
|
root, and of course laguerre() may be called independently. Its default value |
2138
|
|
|
|
|
|
|
is 60. |
2139
|
|
|
|
|
|
|
|
2140
|
|
|
|
|
|
|
=item 'newtonraphson' |
2141
|
|
|
|
|
|
|
|
2142
|
|
|
|
|
|
|
The numeric method used by newtonraphson(). The Newton-Raphson method is offered |
2143
|
|
|
|
|
|
|
as an alternative to Laguerre's method. Its default value is 60. |
2144
|
|
|
|
|
|
|
|
2145
|
|
|
|
|
|
|
=item 'sturm_bisection' |
2146
|
|
|
|
|
|
|
|
2147
|
|
|
|
|
|
|
The bisection method used to find roots within a range. Its default value |
2148
|
|
|
|
|
|
|
is 100. |
2149
|
|
|
|
|
|
|
|
2150
|
|
|
|
|
|
|
=back |
2151
|
|
|
|
|
|
|
|
2152
|
|
|
|
|
|
|
=cut |
2153
|
|
|
|
|
|
|
|
2154
|
|
|
|
|
|
|
sub poly_iteration |
2155
|
|
|
|
|
|
|
{ |
2156
|
13
|
|
|
13
|
1
|
1512
|
my %limits = @_; |
2157
|
13
|
|
|
|
|
17
|
my %old_limits; |
2158
|
|
|
|
|
|
|
|
2159
|
13
|
100
|
|
|
|
43
|
return %iteration if (scalar @_ == 0); |
2160
|
|
|
|
|
|
|
|
2161
|
6
|
|
|
|
|
15
|
for my $k (keys %limits) |
2162
|
|
|
|
|
|
|
{ |
2163
|
|
|
|
|
|
|
# |
2164
|
|
|
|
|
|
|
# If this is a real iteration limit, save its old |
2165
|
|
|
|
|
|
|
# value, then set it. |
2166
|
|
|
|
|
|
|
# |
2167
|
6
|
50
|
|
|
|
12
|
if (exists $iteration{$k}) |
2168
|
|
|
|
|
|
|
{ |
2169
|
6
|
|
|
|
|
9
|
my $val = abs(int($limits{$k})); |
2170
|
|
|
|
|
|
|
|
2171
|
6
|
50
|
|
|
|
11
|
carp "poly_iteration(): Unreasonably small value for $k => $val\n" if ($val < 10); |
2172
|
|
|
|
|
|
|
|
2173
|
6
|
|
|
|
|
10
|
$old_limits{$k} = $iteration{$k}; |
2174
|
6
|
|
|
|
|
9
|
$iteration{$k} = $val; |
2175
|
|
|
|
|
|
|
} |
2176
|
|
|
|
|
|
|
else |
2177
|
|
|
|
|
|
|
{ |
2178
|
0
|
|
|
|
|
0
|
croak "poly_iteration(): unknown key $k."; |
2179
|
|
|
|
|
|
|
} |
2180
|
|
|
|
|
|
|
} |
2181
|
|
|
|
|
|
|
|
2182
|
6
|
|
|
|
|
15
|
return %old_limits; |
2183
|
|
|
|
|
|
|
} |
2184
|
|
|
|
|
|
|
|
2185
|
|
|
|
|
|
|
=head3 poly_tolerance() |
2186
|
|
|
|
|
|
|
|
2187
|
|
|
|
|
|
|
Set the degree of accuracy needed for comparisons to be equal or roots |
2188
|
|
|
|
|
|
|
to be found. Amongst the root finding functions this currently only |
2189
|
|
|
|
|
|
|
affects laguerre() and newtonraphson(), as the Hessenberg matrix method |
2190
|
|
|
|
|
|
|
determines how close it needs to get using a complicated formula based |
2191
|
|
|
|
|
|
|
on L. |
2192
|
|
|
|
|
|
|
|
2193
|
|
|
|
|
|
|
# |
2194
|
|
|
|
|
|
|
# Print the tolerances. |
2195
|
|
|
|
|
|
|
# |
2196
|
|
|
|
|
|
|
my %tolerances = poly_tolerance(); |
2197
|
|
|
|
|
|
|
print "Default tolerances:\n"; |
2198
|
|
|
|
|
|
|
for my $k (keys %tolerances) |
2199
|
|
|
|
|
|
|
{ |
2200
|
|
|
|
|
|
|
print "$k => ", $tolerances{$k}, "\n"; |
2201
|
|
|
|
|
|
|
} |
2202
|
|
|
|
|
|
|
|
2203
|
|
|
|
|
|
|
# |
2204
|
|
|
|
|
|
|
# Quadruple the tolerance for Laguerre's method. |
2205
|
|
|
|
|
|
|
# |
2206
|
|
|
|
|
|
|
poly_tolerance(laguerre => 4 * $tolerances{laguerre}); |
2207
|
|
|
|
|
|
|
|
2208
|
|
|
|
|
|
|
Tolerances may be set for: |
2209
|
|
|
|
|
|
|
|
2210
|
|
|
|
|
|
|
=over 4 |
2211
|
|
|
|
|
|
|
|
2212
|
|
|
|
|
|
|
=item 'laguerre' |
2213
|
|
|
|
|
|
|
|
2214
|
|
|
|
|
|
|
The numeric method used by laguerre(). Laguerre's method is used within |
2215
|
|
|
|
|
|
|
sturm_bisection_roots() once an individual root has been found within a |
2216
|
|
|
|
|
|
|
range, and of course it may be called independently. |
2217
|
|
|
|
|
|
|
|
2218
|
|
|
|
|
|
|
=item 'newtonraphson' |
2219
|
|
|
|
|
|
|
|
2220
|
|
|
|
|
|
|
The numeric method used by newtonraphson(). Newton-Raphson is, like |
2221
|
|
|
|
|
|
|
Laguerre's method, a method for finding a root near the starting X value. |
2222
|
|
|
|
|
|
|
|
2223
|
|
|
|
|
|
|
=back |
2224
|
|
|
|
|
|
|
|
2225
|
|
|
|
|
|
|
=cut |
2226
|
|
|
|
|
|
|
|
2227
|
|
|
|
|
|
|
sub poly_tolerance |
2228
|
|
|
|
|
|
|
{ |
2229
|
5
|
|
|
5
|
1
|
662
|
my %tols = @_; |
2230
|
5
|
|
|
|
|
6
|
my %old_tols; |
2231
|
|
|
|
|
|
|
|
2232
|
5
|
100
|
|
|
|
17
|
return %tolerance if (scalar @_ == 0); |
2233
|
|
|
|
|
|
|
|
2234
|
2
|
|
|
|
|
5
|
for my $k (keys %tols) |
2235
|
|
|
|
|
|
|
{ |
2236
|
|
|
|
|
|
|
# |
2237
|
|
|
|
|
|
|
# If this is a real tolerance limit, save its old |
2238
|
|
|
|
|
|
|
# value, then set it. |
2239
|
|
|
|
|
|
|
# |
2240
|
2
|
50
|
|
|
|
5
|
if (exists $tolerance{$k}) |
2241
|
|
|
|
|
|
|
{ |
2242
|
2
|
|
|
|
|
5
|
my $val = abs($tols{$k}); |
2243
|
|
|
|
|
|
|
|
2244
|
2
|
|
|
|
|
2
|
$old_tols{$k} = $tolerance{$k}; |
2245
|
2
|
|
|
|
|
5
|
$tolerance{$k} = $val; |
2246
|
|
|
|
|
|
|
} |
2247
|
|
|
|
|
|
|
else |
2248
|
|
|
|
|
|
|
{ |
2249
|
0
|
|
|
|
|
0
|
croak "poly_tolerance(): unknown key $k."; |
2250
|
|
|
|
|
|
|
} |
2251
|
|
|
|
|
|
|
} |
2252
|
|
|
|
|
|
|
|
2253
|
2
|
|
|
|
|
4
|
return %old_tols; |
2254
|
|
|
|
|
|
|
} |
2255
|
|
|
|
|
|
|
|
2256
|
|
|
|
|
|
|
=head3 poly_nonzero_term_count() |
2257
|
|
|
|
|
|
|
|
2258
|
|
|
|
|
|
|
Returns a simple count of the number of coefficients that aren't zero. |
2259
|
|
|
|
|
|
|
|
2260
|
|
|
|
|
|
|
=cut |
2261
|
|
|
|
|
|
|
|
2262
|
|
|
|
|
|
|
sub poly_nonzero_term_count |
2263
|
|
|
|
|
|
|
{ |
2264
|
39
|
|
|
39
|
1
|
115068
|
my(@coefficients) = @_; |
2265
|
39
|
|
|
|
|
70
|
my $nzc = 0; |
2266
|
|
|
|
|
|
|
|
2267
|
39
|
|
|
|
|
108
|
for my $j (0..$#coefficients) |
2268
|
|
|
|
|
|
|
{ |
2269
|
227
|
100
|
|
|
|
459
|
$nzc++ if (abs($coefficients[$j]) > $epsilon); |
2270
|
|
|
|
|
|
|
} |
2271
|
39
|
|
|
|
|
98
|
return $nzc; |
2272
|
|
|
|
|
|
|
} |
2273
|
|
|
|
|
|
|
|
2274
|
|
|
|
|
|
|
END { |
2275
|
17
|
|
|
17
|
|
34772
|
unless (1) #unless ($ascending_order_called) |
2276
|
|
|
|
|
|
|
{ |
2277
|
|
|
|
|
|
|
my $end_list = join("\n", keys %called_by); |
2278
|
|
|
|
|
|
|
warn "Coefficient order is in a default state, which will change by version 3.00.\n\n", |
2279
|
|
|
|
|
|
|
"Please use ascending_order(0) at the beginning of your code to make\n", |
2280
|
|
|
|
|
|
|
"sure your function parameters will be in the correct order when the\n", |
2281
|
|
|
|
|
|
|
"default order changes.\n\n", |
2282
|
|
|
|
|
|
|
"See the README file and the Math::Polynomial::Solve documentation for\n", |
2283
|
|
|
|
|
|
|
"more information.\n", |
2284
|
|
|
|
|
|
|
"File(s): $end_list\n"; |
2285
|
|
|
|
|
|
|
} |
2286
|
|
|
|
|
|
|
} |
2287
|
|
|
|
|
|
|
|
2288
|
|
|
|
|
|
|
1; |
2289
|
|
|
|
|
|
|
__END__ |