← Back to branch summary

~ubuntu-branches/ubuntu/karmic/python-scipy/karmic

« back to all changes in this revision

Viewing changes to Lib/ga/tree_opt.py

  • Committer: Bazaar Package Importer
  • Author(s): Daniel T. Chen (new)
  • Date: 2005-03-16 02:15:29 UTC
  • Revision ID: james.westby@ubuntu.com-20050316021529-xrjlowsejs0cijig
Tags: upstream-0.3.2
Import upstream version 0.3.2

Show diffs side-by-side

added added

removed removed

Lines of Context:
Lib/ga/tree_opt.py
 
1
"""
 
2
    some basic language objects
 
3
    if you don't specify a node_type for these, it is usually a good idea to 
 
4
    not allow crossovers at these nodes.  That would allow two (maybe different)
 
5
    list_range nodes to exchange places.
 
6
"""
 
7
 
 
8
import tree
 
9
import gene
 
10
 
 
11
 
 
12
"""
 
13
import weakdict
 
14
class opt_object(tree.tree_node,weakdict.WeakValue):
 
15
        opt_dict = weakdict.WeakDict()
 
16
        def __init__(self,node_type, sub_nodes):
 
17
                tree.tree_node.__init__(self,sub_nodes,node_type=node_type)
 
18
                weakdict.WeakValue.__init__(self)
 
19
                opt_object.opt_dict[id(self)] = self
 
20
        def _create(self,parent = None):        
 
21
                new = tree.tree_node._create(self,parent)
 
22
                weakdict.WeakValue.__init__(new)
 
23
                self._WeakValue__cid = None #force a reset of the __cid value
 
24
                weakdict.WeakValue.__init__(new) #now get a new value for it
 
25
                opt_object.opt_dict[id(new)] = new
 
26
                return new              
 
27
        def __del__(self): 
 
28
                tree.tree_node.__del__(self)
 
29
                weakdict.WeakValue.__del__(self)
 
30
"""             
 
31
 
 
32
class opt_object(tree.tree_node):
 
33
        def __init__(self,node_type, sub_nodes):
 
34
                tree.tree_node.__init__(self,sub_nodes,node_type=node_type)
 
35
 
 
36
class float_range(gene.float_gene,opt_object):
 
37
        optimize = 1
 
38
        def __init__(self,bounds,node_type='float_range', sub_nodes = 0):
 
39
                opt_object.__init__(self,node_type,sub_nodes)
 
40
                gene.float_gene.__init__(self,bounds[:2])
 
41
                tree.tree_node.__init__(self,sub_nodes,node_type=node_type)
 
42
                if(len(bounds) == 3): self.default = bounds[2]
 
43
                else: self.default = (bounds[0] + bounds[1])/2.
 
44
                self._value = self.default
 
45
                print self._value
 
46
        def clone(self,parent = None):return tree.tree_node.clone(self,parent)                  
 
47
        """
 
48
        def mutate(self):
 
49
                m = gene.float_gene.mutate(self)
 
50
                if(m and self.parent): self.parent.recalc(force_parent=1)
 
51
                return m
 
52
        """
 
53
        def scale(self,sc):
 
54
                self.bounds = (self.bounds[0]*sc,self.bounds[1]*sc)
 
55
                self.default = self.default*sc
 
56
                self._value = self._value*sc
 
57
        def defaultize(self):
 
58
                self._value = self.default
 
59
                for child in self._children: child.defaultize()         
 
60
        def create(self,parent):
 
61
                new = tree.tree_node.create(self,parent)
 
62
                new.initialize() 
 
63
                gene.float_gene.initialize(new)                 
 
64
                return new
 
65
        def __del__(self):
 
66
#               gene.float_gene.__del__(self)
 
67
                opt_object.__del__(self)        
 
68
        def __repr__(self):
 
69
                try: 
 
70
                        val = self.value()
 
71
                        if( val < .01 or val > 1000): v = "%4.3e" % self.value()
 
72
                        else: v = "%4.3f" % self.value()                                
 
73
                except gene.GAError: v = 'not initialized'
 
74
                self.label = '%s = %s' % (self.node_type, v)            
 
75
                return tree.tree_node.__repr__(self)
 
76
import math
 
77
class log_float_range(gene.log_float_gene,opt_object):
 
78
        optimize = 1
 
79
        def __init__(self,bounds,node_type='log_float_range', sub_nodes = 0):
 
80
                gene.log_float_gene.__init__(self,bounds[:2])
 
81
                opt_object.__init__(self,node_type,sub_nodes)
 
82
                if(len(bounds) == 3): self.default = bounds[2]
 
83
                else: self.default = (bounds[0] + bounds[1])/2.
 
84
                self._value = math.log10(self.default)
 
85
        def clone(self,parent = None):return tree.tree_node.clone(self,parent)          
 
86
        """
 
87
        def mutate(self):
 
88
                m=gene.log_float_gene.mutate(self)
 
89
                if(m and self.parent): self.parent.recalc(force_parent=1)
 
90
                return m
 
91
        """
 
92
        def scale(self,sc):
 
93
                self.default = self.default*sc
 
94
                sc = log10(sc)
 
95
                self.bounds = (self.bounds[0]*sc,self.bounds[1]*sc)
 
96
                self._value = self._value*sc
 
97
        def defaultize(self):
 
98
                self._value = self.default
 
99
                for child in self._children: child.defaultize()         
 
100
        def create(self,parent):
 
101
                new = tree.tree_node.create(self,parent)
 
102
                new.initialize() 
 
103
                gene.log_float_gene.initialize(new)             
 
104
                return new
 
105
        def __del__(self):
 
106
#               gene.log_float_gene.__del__(self)
 
107
                opt_object.__del__(self)
 
108
        def __repr__(self):
 
109
                try: 
 
110
                        val = self.value()
 
111
                        if( val < .01 or val > 1000): v = "%4.3e" % self.value()
 
112
                        else: v = "%4.3f" % self.value()                                
 
113
                except gene.GAError: v = 'not initialized'
 
114
                self.label = '%s = %s' % (self.node_type, v)            
 
115
                return tree.tree_node.__repr__(self)
 
116
                
 
117
class list_range(gene.list_gene,opt_object):
 
118
        optimize = 1
 
119
        def __init__(self,allele_set,node_type='list_range', default=None, sub_nodes = 0):
 
120
                gene.list_gene.__init__(self,allele_set)
 
121
                opt_object.__init__(self,node_type,sub_nodes)
 
122
                gene.list_gene.initialize(self) # prevents trouble in tree generation
 
123
                if(default): self.default = default
 
124
                else: self.default = allele_set[int(len(allele_set)/2.)] #the center item
 
125
                self._value = self.default
 
126
        """
 
127
        def mutate(self):
 
128
                m=gene.list_gene.mutate(self)
 
129
                if(m and self.parent): self.parent.recalc(force_parent=1)
 
130
                return m
 
131
        """
 
132
        def clone(self,parent = None):return tree.tree_node.clone(self,parent)          
 
133
        def scale(self,sc):
 
134
                for i in range(len(self.allele_set)): 
 
135
                        self.allele_set[i] = self.allele_set[i] *sc
 
136
                self.default = self.default*sc
 
137
                self._value = self._value*sc
 
138
        def defaultize(self):
 
139
                self._value = self.default
 
140
                for child in self._children: child.defaultize()         
 
141
        def create(self,parent):
 
142
                new = tree.tree_node.create(self,parent)
 
143
                new.initialize()
 
144
                gene.list_gene.initialize(new) 
 
145
                return new
 
146
        def __del__(self):
 
147
#               gene.list_gene.__del__(self)
 
148
                opt_object.__del__(self)
 
149
        def __repr__(self):
 
150
                self.label = '%s = %s' % (self.node_type, self.value())         
 
151
                return tree.tree_node.__repr__(self)
 
152
 
 
153
class val(gene.frozen,opt_object):
 
154
        optimize = 0
 
155
        def __init__(self,val,node_type='val',sub_nodes=0):
 
156
                gene.frozen.__init__(self,val)
 
157
                opt_object.__init__(self,node_type,sub_nodes)
 
158
        def clone(self,parent = None):return tree.tree_node.clone(self,parent)          
 
159
        def scale(self,sc): self._value = self._value*sc
 
160
        def defaultize(self): pass
 
161
        def create(self,parent):
 
162
                new = tree.tree_node.create(self,parent)
 
163
                new.initialize() 
 
164
                return new
 
165
        def __del__(self):
 
166
#               gene.frozen.__del__(self)
 
167
                opt_object.__del__(self)        
 
168
        def __repr__(self):
 
169
                self.label = '%s = %s' % (self.node_type, self.value())         
 
170
                return tree.tree_node.__repr__(self)
 
171
 
 
172
 
 
173
"""
 
174
These two routines are useful for picking off or replacing the nodes in a 
 
175
tree that should be that should be numerically optimized.  They are helpful if
 
176
your interested in using a gradient method to optimize some of the paramters
 
177
of the array
 
178
"""
 
179
 
 
180
def pick_numbers(node):
 
181
        start = []; lower = []; upper =[];
 
182
        for child in node.children():   
 
183
                s, l, u = pick_numbers(child)
 
184
                start = start + s
 
185
                lower = lower + l
 
186
                upper = upper + u
 
187
        #for now only works with float_genes    
 
188
        if hasattr(node,'optimize') and node.optimize == 1:     
 
189
                s = node._value
 
190
                l,u = node.bounds
 
191
                start = start + [s]
 
192
                lower = lower + [l]
 
193
                upper = upper + [u]
 
194
        else:
 
195
                print 'no opt:', node.__class__         
 
196
        return start, lower, upper
 
197
 
 
198
def put_numbers(node,vals, index = 0):
 
199
        for child in node.children():   
 
200
                index = put_numbers(child,vals,index)
 
201
        if hasattr(node,'optimize') and node.optimize == 1:     
 
202
                s = node._value = vals[index]
 
203
                index = index + 1
 
204
        return index
 
205
 
 
206
"""
 
207
Grab the numerical nodes that need to be optimized so that you can directly
 
208
manipulate them
 
209
"""
 
210
def pick_optimize_nodes(node):
 
211
        nodes = [];
 
212
        for child in node.children():   
 
213
                nodes = nodes + pick_optimize_nodes(child)
 
214
        #for now only works with float_genes    
 
215
        if hasattr(node,'optimize') and node.optimize == 1:     
 
216
                nodes = nodes + [node]
 
217
        return nodes
 
 
b'\\ No newline at end of file'