1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
|
#!/usr/bin/env python
#
# Author: Mike McKerns (mmckerns @caltech and @uqfoundation)
# Copyright (c) 2010-2016 California Institute of Technology.
# Copyright (c) 2016-2024 The Uncertainty Quantification Foundation.
# License: 3-clause BSD. The full license text is available at:
# - https://github.com/uqfoundation/mystic/blob/master/LICENSE
debug = False
verbose = True
MINMAX = -1 ## NOTE: sup = maximize = -1; inf = minimize = 1
#######################################################################
# scaling and mpi info; also optimizer configuration parameters
# hard-wired: use DE solver, don't use mpi, F-F' calculation
#######################################################################
npop = 40
maxiter = 6000
maxfun = 1e+6
convergence_tol = 1e-6; ngen = 200
crossover = 0.9
percent_change = 0.9
#XXX: tolW = 0.05; tolP = 0.05; ngcol = 200
#######################################################################
# the model function
#######################################################################
from surrogate import marc_surr as model
#######################################################################
# the differential evolution optimizer
#######################################################################
def optimize(cost,_bounds,_constraints):
from mystic.solvers import DifferentialEvolutionSolver2
from mystic.termination import ChangeOverGeneration as COG
from mystic.strategy import Best1Exp
from mystic.monitors import VerboseMonitor, Monitor
from mystic.tools import random_seed
from mystic.termination import Or, CollapseWeight, CollapsePosition, state
if debug:
random_seed(123) # or 666 to force impose_unweighted reweighting
stepmon = VerboseMonitor(1,1)
else:
stepmon = VerboseMonitor(10) if verbose else Monitor()
stepmon._npts = npts
evalmon = Monitor()
lb,ub = _bounds
ndim = len(lb)
solver = DifferentialEvolutionSolver2(ndim,npop)
solver.SetRandomInitialPoints(min=lb,max=ub)
solver.SetStrictRanges(min=lb,max=ub)
solver.SetEvaluationLimits(maxiter,maxfun)
solver.SetEvaluationMonitor(evalmon)
solver.SetGenerationMonitor(stepmon)
solver.SetConstraints(_constraints)
tol = convergence_tol
term = Or(COG(tol,ngen), CollapseWeight(), CollapsePosition())
solver.Solve(cost,termination=term,strategy=Best1Exp, disp=verbose, \
CrossProbability=crossover,ScalingFactor=percent_change)
#while collapse and solver.Collapse(verbose): #XXX: total_evaluations?
# if debug: print(state(solver._termination).keys())
# solver.Solve() #XXX: cost, term, strategy, cross, scale ?
# if debug: solver.SaveSolver('debug.pkl')
solved = solver.bestSolution
#print("solved: %s" % solver.Solution())
func_max = MINMAX * solver.bestEnergy #NOTE: -solution assumes -Max
#func_max = 1.0 + MINMAX*solver.bestEnergy #NOTE: 1-sol => 1-success = fail
func_evals = solver.evaluations
from mystic.munge import write_support_file
write_support_file(stepmon, npts=npts)
return solved, func_max, func_evals
#######################################################################
# maximize the function
#######################################################################
def maximize(params,npts,bounds):
from mystic.math.measures import split_param
from mystic.math.discrete import product_measure
from mystic.math import almostEqual
from numpy import inf
atol = 1e-18 # default is 1e-18
rtol = 1e-7 # default is 1e-7
target,error = params
lb,ub = bounds
# split lower & upper bounds into weight-only & sample-only
w_lb, x_lb = split_param(lb, npts)
w_ub, x_ub = split_param(ub, npts)
# NOTE: rv, lb, ub are of the form:
# rv = [wxi]*nx + [xi]*nx + [wyi]*ny + [yi]*ny + [wzi]*nz + [zi]*nz
# generate primary constraints function
from mystic import suppressed
@suppressed(1e-2)
def constraints(rv):
c = product_measure().load(rv, npts)
# NOTE: bounds wi in [0,1] enforced by filtering
# impose norm on each discrete measure
for measure in c:
if not almostEqual(float(measure.mass), 1.0, tol=atol, rel=rtol):
measure.normalize()
# impose expectation on product measure
##################### begin function-specific #####################
E = float(c.expect(model))
if not (E <= float(target[0] + error[0])) \
or not (float(target[0] - error[0]) <= E):
# if debug: print(c)
c.set_expect(target[0], model, (x_lb,x_ub), tol=error[0])
###################### end function-specific ######################
# extract weights and positions
return c.flatten()
# generate maximizing function
def cost(rv):
c = product_measure().load(rv, npts)
E = float(c.expect(model))
if E > (target[0] + error[0]) or E < (target[0] - error[0]):
if debug: print("skipping expect: %s" % E)
return inf #XXX: FORCE TO SATISFY E CONSTRAINTS
return MINMAX * c.pof(model)
# maximize
solved, func_max, func_evals = optimize(cost,(lb,ub),constraints)
if MINMAX == 1:
print("func_minimum: %s" % func_max) # inf
else:
print("func_maximum: %s" % func_max) # sup
print("func_evals: %s" % func_evals)
return solved, func_max
#######################################################################
# rank, bounds, and restart information
#######################################################################
if __name__ == '__main__':
function_name = model.__name__
H_mean = 6.5 #NOTE: SET THE 'mean' HERE!
H_range = 1.0 #NOTE: SET THE 'range' HERE!
nx = 3 #NOTE: SET THE NUMBER OF 'h' POINTS HERE!
ny = 3 #NOTE: SET THE NUMBER OF 'a' POINTS HERE!
nz = 3 #NOTE: SET THE NUMBER OF 'v' POINTS HERE!
target = (H_mean,)
error = (H_range,)
w_lower = [0.0]
w_upper = [1.0]
h_lower = [60.0]; a_lower = [0.0]; v_lower = [2.1]
h_upper = [105.0]; a_upper = [30.0]; v_upper = [2.8]
lower_bounds = (nx * w_lower) + (nx * h_lower) \
+ (ny * w_lower) + (ny * a_lower) \
+ (nz * w_lower) + (nz * v_lower)
upper_bounds = (nx * w_upper) + (nx * h_upper) \
+ (ny * w_upper) + (ny * a_upper) \
+ (nz * w_upper) + (nz * v_upper)
print("...SETTINGS...")
print("npop = %s" % npop)
print("maxiter = %s" % maxiter)
print("maxfun = %s" % maxfun)
print("convergence_tol = %s" % convergence_tol)
print("crossover = %s" % crossover)
print("percent_change = %s" % percent_change)
print("..............\n")
print(" model: f(x) = %s(x)" % function_name)
print(" target: %s" % str(target))
print(" error: %s" % str(error))
print(" npts: %s" % str((nx,ny,nz)))
print("..............\n")
param_string = "["
for i in range(nx):
param_string += "'wx%s', " % str(i+1)
for i in range(nx):
param_string += "'x%s', " % str(i+1)
for i in range(ny):
param_string += "'wy%s', " % str(i+1)
for i in range(ny):
param_string += "'y%s', " % str(i+1)
for i in range(nz):
param_string += "'wz%s', " % str(i+1)
for i in range(nz):
param_string += "'z%s', " % str(i+1)
param_string = param_string[:-2] + "]"
print(" parameters: %s" % param_string)
print(" lower bounds: %s" % lower_bounds)
print(" upper bounds: %s" % upper_bounds)
# print(" ...")
pars = (target,error)
npts = (nx,ny,nz)
bounds = (lower_bounds,upper_bounds)
solved, diameter = maximize(pars,npts,bounds)
from numpy import array
from mystic.math.discrete import product_measure
c = product_measure().load(solved,npts)
print("solved: [wx,x]\n%s" % array(list(zip(c[0].weights,c[0].positions))))
print("solved: [wy,y]\n%s" % array(list(zip(c[1].weights,c[1].positions))))
print("solved: [wz,z]\n%s" % array(list(zip(c[2].weights,c[2].positions))))
print("expect: %s" % str( c.expect(model) ))
# EOF
|
