conjugate_gradient.py 5.47 KB
Newer Older
1
2
3
# -*- coding: utf-8 -*-


4
5
6
7
from __future__ import division
import numpy as np


8
class ConjugateGradient(object):
9
10
    def __init__(self, convergence_tolerance=1E-4, convergence_level=3,
                 iteration_limit=None, reset_count=None,
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
                 preconditioner=None, callback=None):
        """
            Initializes the conjugate_gradient and sets the attributes (except
            for `x`).

            Parameters
            ----------
            A : {operator, function}
                Operator `A` applicable to a field.
            b : field
                Resulting field of the operation `A(x)`.
            W : {operator, function}, *optional*
                Operator `W` that is a preconditioner on `A` and is applicable to a
                field (default: None).
            spam : function, *optional*
                Callback function which is given the current `x` and iteration
                counter each iteration (default: None).
            reset : integer, *optional*
                Number of iterations after which to restart; i.e., forget previous
                conjugated directions (default: sqrt(b.dim)).
            note : bool, *optional*
                Indicates whether notes are printed or not (default: False).

        """
        self.convergence_tolerance = np.float(convergence_tolerance)
        self.convergence_level = np.float(convergence_level)
37
38
39
40
41
42
43
44

        if iteration_limit is not None:
            iteration_limit = int(iteration_limit)
        self.iteration_limit = iteration_limit

        if reset_count is not None:
            reset_count = int(reset_count)
        self.reset_count = reset_count
45
46
47
48
49
50
51

        if preconditioner is None:
            preconditioner = lambda z: z

        self.preconditioner = preconditioner
        self.callback = callback

52
    def __call__(self, A, b, x0):
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
        """
            Runs the conjugate gradient minimization.

            Parameters
            ----------
            x0 : field, *optional*
                Starting guess for the minimization.
            tol : scalar, *optional*
                Tolerance specifying convergence; measured by current relative
                residual (default: 1E-4).
            clevel : integer, *optional*
                Number of times the tolerance should be undershot before
                exiting (default: 1).
            limii : integer, *optional*
                Maximum number of iterations performed (default: 10 * b.dim).

            Returns
            -------
            x : field
                Latest `x` of the minimization.
            convergence : integer
                Latest convergence level indicating whether the minimization
                has converged or not.

            """
        r = b - A(x0)
        d = self.preconditioner(r)
        previous_gamma = r.dot(d)
        if previous_gamma == 0:
82
83
            self.logger.info("The starting guess is already perfect solution "
                             "for the inverse problem.")
84
            return x0, self.convergence_level+1
85
        norm_b = np.sqrt(b.dot(b))
86
87
88
        x = x0
        convergence = 0
        iteration_number = 1
89
        self.logger.info("Starting conjugate gradient.")
90

91
        while True:
92
93
94
95
96
97
98
            if self.callback is not None:
                self.callback(x, iteration_number)

            q = A(d)
            alpha = previous_gamma/d.dot(q)

            if not np.isfinite(alpha):
99
                self.logger.error("Alpha became infinite! Stopping.")
100
101
102
103
104
105
                return x0, 0

            x += d * alpha

            reset = False
            if alpha.real < 0:
106
                self.logger.warn("Positive definiteness of A violated!")
107
                reset = True
108
109
110
            if self.reset_count is not None:
                reset += (iteration_number % self.reset_count == 0)
            if reset:
111
                self.logger.info("Resetting conjugate directions.")
112
113
114
115
116
117
118
119
                r = b - A(x)
            else:
                r -= q * alpha

            s = self.preconditioner(r)
            gamma = r.dot(s)

            if gamma.real < 0:
120
121
                self.logger.warn("Positive definitness of preconditioner "
                                 "violated!")
122
123
124

            beta = max(0, gamma/previous_gamma)

125
            delta = np.sqrt(gamma)/norm_b
126

127
128
129
130
131
132
            self.logger.debug("Iteration : %08u   alpha = %3.1E   "
                              "beta = %3.1E   delta = %3.1E" %
                              (iteration_number,
                               np.real(alpha),
                               np.real(beta),
                               np.real(delta)))
133
134
135

            if gamma == 0:
                convergence = self.convergence_level+1
136
                self.logger.info("Reached infinite convergence.")
137
138
139
                break
            elif abs(delta) < self.convergence_tolerance:
                convergence += 1
140
141
                self.logger.info("Updated convergence level to: %u" %
                                 convergence)
142
                if convergence == self.convergence_level:
143
                    self.logger.info("Reached target convergence level.")
144
145
146
147
                    break
            else:
                convergence = max(0, convergence-1)

148
149
            if self.iteration_limit is not None:
                if iteration_number == self.iteration_limit:
150
                    self.logger.warn("Reached iteration limit. Stopping.")
151
152
153
                    break

            d = s + d * beta
154
155
156
157
158

            iteration_number += 1
            previous_gamma = gamma

        return x, convergence