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
219
|
[comment {-*- tcl -*- doctools manpage}]
[manpage_begin simulation::montecarlo n 0.1]
[keywords math]
[keywords {montecarlo simulation}]
[keywords {stochastic modelling}]
[copyright {2008 Arjen Markus <arjenmarkus@users.sourceforge.net>}]
[moddesc {Tcl Simulation Tools}]
[titledesc {Monte Carlo simulations}]
[category Mathematics]
[require Tcl [opt 8.4]]
[require simulation::montecarlo 0.1]
[require simulation::random]
[require math::statistics]
[description]
[para]
The technique of [emph "Monte Carlo simulations"] is basically simple:
[list_begin itemized]
[item]
generate random values for one or more parameters.
[item]
evaluate the model of some system you are interested in and record the
interesting results for each realisation of these parameters.
[item]
after a suitable number of such trials, deduce an overall characteristic
of the model.
[list_end]
You can think of a model of a network of computers, an ecosystem of some
kind or in fact anything that can be quantitatively described and has
some stochastic element in it.
[para]
The package [term simulation::montecarlo] offers a basic framework for
such a modelling technique:
[example {
#
# MC experiments:
# Determine the mean and median of a set of points and compare them
#
::simulation::montecarlo::singleExperiment -init {
package require math::statistics
set prng [::simulation::random::prng_Normal 0.0 1.0]
} -loop {
set numbers {}
for { set i 0 } { $i < [getOption samples] } { incr i } {
lappend numbers [$prng]
}
set mean [::math::statistics::mean $numbers]
set median [::math::statistics::median $numbers] ;# ? Exists?
setTrialResult [list $mean $median]
} -final {
set result [getTrialResults]
set means {}
set medians {}
foreach r $result {
foreach {m M} $r break
lappend means $m
lappend medians $M
}
puts [getOption reportfile] "Correlation: [::math::statistics::corr $means $medians]"
} -trials 100 -samples 10 -verbose 1 -columns {Mean Median}
}]
This example attemps to find out how well the median value and the mean
value of a random set of numbers correlate. Sometimes a median value is
a more robust characteristic than a mean value - especially if you have
a statistical distribution with "fat" tails.
[section "PROCEDURES"]
The package defines the following auxiliary procedures:
[list_begin definitions]
[call [cmd ::simulation::montecarlo::getOption] [arg keyword]]
Get the value of an option given as part of the [term singeExperiment] command.
[list_begin arguments]
[arg_def string keyword] Given keyword (without leading minus)
[list_end]
[para]
[call [cmd ::simulation::montecarlo::hasOption] [arg keyword]]
Returns 1 if the option is available, 0 if not.
[list_begin arguments]
[arg_def string keyword] Given keyword (without leading minus)
[list_end]
[para]
[call [cmd ::simulation::montecarlo::setOption] [arg keyword] [arg value]]
Set the value of the given option.
[list_begin arguments]
[arg_def string keyword] Given keyword (without leading minus)
[arg_def string value] (New) value for the option
[list_end]
[para]
[call [cmd ::simulation::montecarlo::setTrialResult] [arg values]]
Store the results of the trial for later analysis
[list_begin arguments]
[arg_def list values] List of values to be stored
[list_end]
[para]
[call [cmd ::simulation::montecarlo::setExpResult] [arg values]]
Set the results of the entire experiment (typically used in the final
phase).
[list_begin arguments]
[arg_def list values] List of values to be stored
[list_end]
[para]
[call [cmd ::simulation::montecarlo::getTrialResults]]
Get the results of all individual trials for analysis (typically used in
the final phase or after completion of the command).
[para]
[call [cmd ::simulation::montecarlo::getExpResult]]
Get the results of the entire experiment (typically used in the final
phase or even after completion of the [term singleExperiment] command).
[para]
[call [cmd ::simulation::montecarlo::transposeData] [arg values]]
Interchange columns and rows of a list of lists and return the result.
[list_begin arguments]
[arg_def list values] List of lists of values
[list_end]
[list_end]
There are two main procedures: [term integral2D] and [term singleExperiment].
[list_begin definitions]
[call [cmd ::simulation::montecarlo::integral2D] [arg ...]]
Integrate a function over a two-dimensional region using a Monte Carlo
approach.
[para]
Arguments PM
[para]
[call [cmd ::simulation::montecarlo::singleExperiment] [arg args]]
Iterate code over a number of trials and store the results. The
iteration is gouverned by parameters given via a list of
keyword-value pairs.
[list_begin args]
[arg_def int n] List of keyword-value pairs, all of which are available
during the execution via the [term getOption] command.
[list_end]
[list_end]
The [term singleExperiment] command predefines the following
options:
[list_begin itemized]
[item]
[term "-init code"]: code to be run at start up
[item]
[term "-loop body"]: body of code that defines the computation to
be run time and again. The code should use [term setTrialResult]
to store the results of each trial (typically a list of numbers,
but the interpretation is up to the implementation). Note: Required keyword.
[item]
[term "-final code"]: code to be run at the end
[item]
[term "-trials n"]: number of trials in the experiment (required)
[item]
[term "-reportfile file"]: opened file to send the output to (default: stdout)
[item]
[term "-verbose"]: write the intermediate results (1) or not (0) (default: 0)
[item]
[term "-analysis proc"]: either "none" (no automatic analysis), standard
(basic statistics of the trial results and a correlation matrix) or the
name of a procedure that will take care of the analysis.
[item]
[term "-columns list"]: list of column names, useful for verbose output
and the analysis
[list_end]
Any other options can be used via the getOption procedure
in the body.
[section TIPS]
The procedure [term singleExperiment] works by constructing a
temporary procedure that does the actual work. It loops for the given
number of trials.
[para]
As it constructs a temporary procedure, local variables defined at the
start continue to exist in the loop.
[manpage_end]
|
