lemon/random.h
author Alpar Juttner <alpar@cs.elte.hu>
Tue, 06 Apr 2010 06:52:48 +0200
branch1.2
changeset 887 77704bf2a4ae
parent 559 c5fd2d996909
permissions -rw-r--r--
Merge bugfix #364 to branch 1.2
alpar@209
     1
/* -*- mode: C++; indent-tabs-mode: nil; -*-
alpar@10
     2
 *
alpar@209
     3
 * This file is a part of LEMON, a generic C++ optimization library.
alpar@10
     4
 *
alpar@440
     5
 * Copyright (C) 2003-2009
alpar@10
     6
 * Egervary Jeno Kombinatorikus Optimalizalasi Kutatocsoport
alpar@10
     7
 * (Egervary Research Group on Combinatorial Optimization, EGRES).
alpar@10
     8
 *
alpar@10
     9
 * Permission to use, modify and distribute this software is granted
alpar@10
    10
 * provided that this copyright notice appears in all copies. For
alpar@10
    11
 * precise terms see the accompanying LICENSE file.
alpar@10
    12
 *
alpar@10
    13
 * This software is provided "AS IS" with no warranty of any kind,
alpar@10
    14
 * express or implied, and with no claim as to its suitability for any
alpar@10
    15
 * purpose.
alpar@10
    16
 *
alpar@10
    17
 */
alpar@10
    18
alpar@10
    19
/*
alpar@10
    20
 * This file contains the reimplemented version of the Mersenne Twister
alpar@10
    21
 * Generator of Matsumoto and Nishimura.
alpar@10
    22
 *
alpar@10
    23
 * See the appropriate copyright notice below.
alpar@209
    24
 *
alpar@10
    25
 * Copyright (C) 1997 - 2002, Makoto Matsumoto and Takuji Nishimura,
alpar@209
    26
 * All rights reserved.
alpar@10
    27
 *
alpar@10
    28
 * Redistribution and use in source and binary forms, with or without
alpar@10
    29
 * modification, are permitted provided that the following conditions
alpar@10
    30
 * are met:
alpar@10
    31
 *
alpar@10
    32
 * 1. Redistributions of source code must retain the above copyright
alpar@10
    33
 *    notice, this list of conditions and the following disclaimer.
alpar@10
    34
 *
alpar@10
    35
 * 2. Redistributions in binary form must reproduce the above copyright
alpar@10
    36
 *    notice, this list of conditions and the following disclaimer in the
alpar@10
    37
 *    documentation and/or other materials provided with the distribution.
alpar@10
    38
 *
alpar@209
    39
 * 3. The names of its contributors may not be used to endorse or promote
alpar@209
    40
 *    products derived from this software without specific prior written
alpar@10
    41
 *    permission.
alpar@10
    42
 *
alpar@10
    43
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
alpar@10
    44
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
alpar@10
    45
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
alpar@10
    46
 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
alpar@10
    47
 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
alpar@10
    48
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
alpar@10
    49
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
alpar@10
    50
 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
alpar@10
    51
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
alpar@10
    52
 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
alpar@10
    53
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
alpar@10
    54
 * OF THE POSSIBILITY OF SUCH DAMAGE.
alpar@10
    55
 *
alpar@10
    56
 *
alpar@10
    57
 * Any feedback is very welcome.
alpar@10
    58
 * http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html
alpar@10
    59
 * email: m-mat @ math.sci.hiroshima-u.ac.jp (remove space)
alpar@10
    60
 */
alpar@10
    61
alpar@10
    62
#ifndef LEMON_RANDOM_H
alpar@10
    63
#define LEMON_RANDOM_H
alpar@10
    64
alpar@10
    65
#include <algorithm>
alpar@10
    66
#include <iterator>
alpar@10
    67
#include <vector>
deba@110
    68
#include <limits>
deba@177
    69
#include <fstream>
alpar@10
    70
alpar@68
    71
#include <lemon/math.h>
alpar@10
    72
#include <lemon/dim2.h>
alpar@68
    73
deba@177
    74
#ifndef WIN32
deba@177
    75
#include <sys/time.h>
deba@177
    76
#include <ctime>
deba@177
    77
#include <sys/types.h>
deba@177
    78
#include <unistd.h>
deba@177
    79
#else
alpar@482
    80
#include <lemon/bits/windows.h>
deba@177
    81
#endif
deba@177
    82
alpar@10
    83
///\ingroup misc
alpar@10
    84
///\file
alpar@10
    85
///\brief Mersenne Twister random number generator
alpar@10
    86
alpar@10
    87
namespace lemon {
alpar@10
    88
alpar@10
    89
  namespace _random_bits {
alpar@209
    90
alpar@10
    91
    template <typename _Word, int _bits = std::numeric_limits<_Word>::digits>
alpar@10
    92
    struct RandomTraits {};
alpar@10
    93
alpar@10
    94
    template <typename _Word>
alpar@10
    95
    struct RandomTraits<_Word, 32> {
alpar@10
    96
alpar@10
    97
      typedef _Word Word;
alpar@10
    98
      static const int bits = 32;
alpar@10
    99
alpar@10
   100
      static const int length = 624;
alpar@10
   101
      static const int shift = 397;
alpar@209
   102
alpar@10
   103
      static const Word mul = 0x6c078965u;
alpar@10
   104
      static const Word arrayInit = 0x012BD6AAu;
alpar@10
   105
      static const Word arrayMul1 = 0x0019660Du;
alpar@10
   106
      static const Word arrayMul2 = 0x5D588B65u;
alpar@10
   107
alpar@10
   108
      static const Word mask = 0x9908B0DFu;
alpar@10
   109
      static const Word loMask = (1u << 31) - 1;
alpar@10
   110
      static const Word hiMask = ~loMask;
alpar@10
   111
alpar@10
   112
alpar@10
   113
      static Word tempering(Word rnd) {
alpar@10
   114
        rnd ^= (rnd >> 11);
alpar@10
   115
        rnd ^= (rnd << 7) & 0x9D2C5680u;
alpar@10
   116
        rnd ^= (rnd << 15) & 0xEFC60000u;
alpar@10
   117
        rnd ^= (rnd >> 18);
alpar@10
   118
        return rnd;
alpar@10
   119
      }
alpar@10
   120
alpar@10
   121
    };
alpar@10
   122
alpar@10
   123
    template <typename _Word>
alpar@10
   124
    struct RandomTraits<_Word, 64> {
alpar@10
   125
alpar@10
   126
      typedef _Word Word;
alpar@10
   127
      static const int bits = 64;
alpar@10
   128
alpar@10
   129
      static const int length = 312;
alpar@10
   130
      static const int shift = 156;
alpar@10
   131
alpar@10
   132
      static const Word mul = Word(0x5851F42Du) << 32 | Word(0x4C957F2Du);
alpar@10
   133
      static const Word arrayInit = Word(0x00000000u) << 32 |Word(0x012BD6AAu);
alpar@10
   134
      static const Word arrayMul1 = Word(0x369DEA0Fu) << 32 |Word(0x31A53F85u);
alpar@10
   135
      static const Word arrayMul2 = Word(0x27BB2EE6u) << 32 |Word(0x87B0B0FDu);
alpar@10
   136
alpar@10
   137
      static const Word mask = Word(0xB5026F5Au) << 32 | Word(0xA96619E9u);
alpar@10
   138
      static const Word loMask = (Word(1u) << 31) - 1;
alpar@10
   139
      static const Word hiMask = ~loMask;
alpar@10
   140
alpar@10
   141
      static Word tempering(Word rnd) {
alpar@10
   142
        rnd ^= (rnd >> 29) & (Word(0x55555555u) << 32 | Word(0x55555555u));
alpar@10
   143
        rnd ^= (rnd << 17) & (Word(0x71D67FFFu) << 32 | Word(0xEDA60000u));
alpar@10
   144
        rnd ^= (rnd << 37) & (Word(0xFFF7EEE0u) << 32 | Word(0x00000000u));
alpar@10
   145
        rnd ^= (rnd >> 43);
alpar@10
   146
        return rnd;
alpar@10
   147
      }
alpar@10
   148
alpar@10
   149
    };
alpar@10
   150
alpar@10
   151
    template <typename _Word>
alpar@10
   152
    class RandomCore {
alpar@10
   153
    public:
alpar@10
   154
alpar@10
   155
      typedef _Word Word;
alpar@10
   156
alpar@10
   157
    private:
alpar@10
   158
alpar@10
   159
      static const int bits = RandomTraits<Word>::bits;
alpar@10
   160
alpar@10
   161
      static const int length = RandomTraits<Word>::length;
alpar@10
   162
      static const int shift = RandomTraits<Word>::shift;
alpar@10
   163
alpar@10
   164
    public:
alpar@10
   165
alpar@10
   166
      void initState() {
alpar@10
   167
        static const Word seedArray[4] = {
alpar@10
   168
          0x12345u, 0x23456u, 0x34567u, 0x45678u
alpar@10
   169
        };
alpar@209
   170
alpar@10
   171
        initState(seedArray, seedArray + 4);
alpar@10
   172
      }
alpar@10
   173
alpar@10
   174
      void initState(Word seed) {
alpar@10
   175
alpar@10
   176
        static const Word mul = RandomTraits<Word>::mul;
alpar@10
   177
alpar@209
   178
        current = state;
alpar@10
   179
alpar@10
   180
        Word *curr = state + length - 1;
alpar@10
   181
        curr[0] = seed; --curr;
alpar@10
   182
        for (int i = 1; i < length; ++i) {
alpar@10
   183
          curr[0] = (mul * ( curr[1] ^ (curr[1] >> (bits - 2)) ) + i);
alpar@10
   184
          --curr;
alpar@10
   185
        }
alpar@10
   186
      }
alpar@10
   187
alpar@10
   188
      template <typename Iterator>
alpar@10
   189
      void initState(Iterator begin, Iterator end) {
alpar@10
   190
alpar@10
   191
        static const Word init = RandomTraits<Word>::arrayInit;
alpar@10
   192
        static const Word mul1 = RandomTraits<Word>::arrayMul1;
alpar@10
   193
        static const Word mul2 = RandomTraits<Word>::arrayMul2;
alpar@10
   194
alpar@10
   195
alpar@10
   196
        Word *curr = state + length - 1; --curr;
alpar@10
   197
        Iterator it = begin; int cnt = 0;
alpar@10
   198
        int num;
alpar@10
   199
alpar@10
   200
        initState(init);
alpar@10
   201
alpar@10
   202
        num = length > end - begin ? length : end - begin;
alpar@10
   203
        while (num--) {
alpar@209
   204
          curr[0] = (curr[0] ^ ((curr[1] ^ (curr[1] >> (bits - 2))) * mul1))
alpar@10
   205
            + *it + cnt;
alpar@10
   206
          ++it; ++cnt;
alpar@10
   207
          if (it == end) {
alpar@10
   208
            it = begin; cnt = 0;
alpar@10
   209
          }
alpar@10
   210
          if (curr == state) {
alpar@10
   211
            curr = state + length - 1; curr[0] = state[0];
alpar@10
   212
          }
alpar@10
   213
          --curr;
alpar@10
   214
        }
alpar@10
   215
alpar@10
   216
        num = length - 1; cnt = length - (curr - state) - 1;
alpar@10
   217
        while (num--) {
alpar@10
   218
          curr[0] = (curr[0] ^ ((curr[1] ^ (curr[1] >> (bits - 2))) * mul2))
alpar@10
   219
            - cnt;
alpar@10
   220
          --curr; ++cnt;
alpar@10
   221
          if (curr == state) {
alpar@10
   222
            curr = state + length - 1; curr[0] = state[0]; --curr;
alpar@10
   223
            cnt = 1;
alpar@10
   224
          }
alpar@10
   225
        }
alpar@209
   226
alpar@10
   227
        state[length - 1] = Word(1) << (bits - 1);
alpar@10
   228
      }
alpar@209
   229
alpar@10
   230
      void copyState(const RandomCore& other) {
alpar@10
   231
        std::copy(other.state, other.state + length, state);
alpar@10
   232
        current = state + (other.current - other.state);
alpar@10
   233
      }
alpar@10
   234
alpar@10
   235
      Word operator()() {
alpar@10
   236
        if (current == state) fillState();
alpar@10
   237
        --current;
alpar@10
   238
        Word rnd = *current;
alpar@10
   239
        return RandomTraits<Word>::tempering(rnd);
alpar@10
   240
      }
alpar@10
   241
alpar@10
   242
    private:
alpar@10
   243
alpar@209
   244
alpar@10
   245
      void fillState() {
alpar@10
   246
        static const Word mask[2] = { 0x0ul, RandomTraits<Word>::mask };
alpar@10
   247
        static const Word loMask = RandomTraits<Word>::loMask;
alpar@10
   248
        static const Word hiMask = RandomTraits<Word>::hiMask;
alpar@10
   249
alpar@209
   250
        current = state + length;
alpar@10
   251
alpar@10
   252
        register Word *curr = state + length - 1;
alpar@10
   253
        register long num;
alpar@209
   254
alpar@10
   255
        num = length - shift;
alpar@10
   256
        while (num--) {
alpar@10
   257
          curr[0] = (((curr[0] & hiMask) | (curr[-1] & loMask)) >> 1) ^
alpar@10
   258
            curr[- shift] ^ mask[curr[-1] & 1ul];
alpar@10
   259
          --curr;
alpar@10
   260
        }
alpar@10
   261
        num = shift - 1;
alpar@10
   262
        while (num--) {
alpar@10
   263
          curr[0] = (((curr[0] & hiMask) | (curr[-1] & loMask)) >> 1) ^
alpar@10
   264
            curr[length - shift] ^ mask[curr[-1] & 1ul];
alpar@10
   265
          --curr;
alpar@10
   266
        }
deba@62
   267
        state[0] = (((state[0] & hiMask) | (curr[length - 1] & loMask)) >> 1) ^
alpar@10
   268
          curr[length - shift] ^ mask[curr[length - 1] & 1ul];
alpar@10
   269
alpar@10
   270
      }
alpar@10
   271
alpar@209
   272
alpar@10
   273
      Word *current;
alpar@10
   274
      Word state[length];
alpar@209
   275
alpar@10
   276
    };
alpar@10
   277
alpar@10
   278
alpar@209
   279
    template <typename Result,
alpar@10
   280
              int shift = (std::numeric_limits<Result>::digits + 1) / 2>
alpar@10
   281
    struct Masker {
alpar@10
   282
      static Result mask(const Result& result) {
alpar@10
   283
        return Masker<Result, (shift + 1) / 2>::
alpar@10
   284
          mask(static_cast<Result>(result | (result >> shift)));
alpar@10
   285
      }
alpar@10
   286
    };
alpar@209
   287
alpar@10
   288
    template <typename Result>
alpar@10
   289
    struct Masker<Result, 1> {
alpar@10
   290
      static Result mask(const Result& result) {
alpar@10
   291
        return static_cast<Result>(result | (result >> 1));
alpar@10
   292
      }
alpar@10
   293
    };
alpar@10
   294
alpar@209
   295
    template <typename Result, typename Word,
alpar@209
   296
              int rest = std::numeric_limits<Result>::digits, int shift = 0,
alpar@10
   297
              bool last = rest <= std::numeric_limits<Word>::digits>
alpar@10
   298
    struct IntConversion {
alpar@10
   299
      static const int bits = std::numeric_limits<Word>::digits;
alpar@209
   300
alpar@10
   301
      static Result convert(RandomCore<Word>& rnd) {
alpar@10
   302
        return static_cast<Result>(rnd() >> (bits - rest)) << shift;
alpar@10
   303
      }
alpar@10
   304
alpar@209
   305
    };
alpar@209
   306
alpar@209
   307
    template <typename Result, typename Word, int rest, int shift>
alpar@10
   308
    struct IntConversion<Result, Word, rest, shift, false> {
alpar@10
   309
      static const int bits = std::numeric_limits<Word>::digits;
alpar@10
   310
alpar@10
   311
      static Result convert(RandomCore<Word>& rnd) {
alpar@209
   312
        return (static_cast<Result>(rnd()) << shift) |
alpar@10
   313
          IntConversion<Result, Word, rest - bits, shift + bits>::convert(rnd);
alpar@10
   314
      }
alpar@10
   315
    };
alpar@10
   316
alpar@10
   317
alpar@10
   318
    template <typename Result, typename Word,
alpar@209
   319
              bool one_word = (std::numeric_limits<Word>::digits <
alpar@209
   320
                               std::numeric_limits<Result>::digits) >
alpar@10
   321
    struct Mapping {
alpar@10
   322
      static Result map(RandomCore<Word>& rnd, const Result& bound) {
alpar@10
   323
        Word max = Word(bound - 1);
alpar@10
   324
        Result mask = Masker<Result>::mask(bound - 1);
alpar@10
   325
        Result num;
alpar@10
   326
        do {
alpar@209
   327
          num = IntConversion<Result, Word>::convert(rnd) & mask;
alpar@10
   328
        } while (num > max);
alpar@10
   329
        return num;
alpar@10
   330
      }
alpar@10
   331
    };
alpar@10
   332
alpar@10
   333
    template <typename Result, typename Word>
alpar@10
   334
    struct Mapping<Result, Word, false> {
alpar@10
   335
      static Result map(RandomCore<Word>& rnd, const Result& bound) {
alpar@10
   336
        Word max = Word(bound - 1);
alpar@10
   337
        Word mask = Masker<Word, (std::numeric_limits<Result>::digits + 1) / 2>
alpar@10
   338
          ::mask(max);
alpar@10
   339
        Word num;
alpar@10
   340
        do {
alpar@10
   341
          num = rnd() & mask;
alpar@10
   342
        } while (num > max);
alpar@10
   343
        return num;
alpar@10
   344
      }
alpar@10
   345
    };
alpar@10
   346
kpeter@487
   347
    template <typename Result, int exp>
alpar@10
   348
    struct ShiftMultiplier {
alpar@10
   349
      static const Result multiplier() {
alpar@10
   350
        Result res = ShiftMultiplier<Result, exp / 2>::multiplier();
alpar@10
   351
        res *= res;
alpar@10
   352
        if ((exp & 1) == 1) res *= static_cast<Result>(0.5);
alpar@209
   353
        return res;
alpar@10
   354
      }
alpar@10
   355
    };
alpar@10
   356
alpar@10
   357
    template <typename Result>
kpeter@487
   358
    struct ShiftMultiplier<Result, 0> {
alpar@10
   359
      static const Result multiplier() {
alpar@209
   360
        return static_cast<Result>(1.0);
alpar@10
   361
      }
alpar@10
   362
    };
alpar@10
   363
alpar@10
   364
    template <typename Result>
kpeter@487
   365
    struct ShiftMultiplier<Result, 20> {
alpar@10
   366
      static const Result multiplier() {
alpar@209
   367
        return static_cast<Result>(1.0/1048576.0);
alpar@10
   368
      }
alpar@10
   369
    };
alpar@209
   370
alpar@10
   371
    template <typename Result>
kpeter@487
   372
    struct ShiftMultiplier<Result, 32> {
alpar@10
   373
      static const Result multiplier() {
kpeter@487
   374
        return static_cast<Result>(1.0/4294967296.0);
alpar@10
   375
      }
alpar@10
   376
    };
alpar@10
   377
alpar@10
   378
    template <typename Result>
kpeter@487
   379
    struct ShiftMultiplier<Result, 53> {
alpar@10
   380
      static const Result multiplier() {
alpar@209
   381
        return static_cast<Result>(1.0/9007199254740992.0);
alpar@10
   382
      }
alpar@10
   383
    };
alpar@10
   384
alpar@10
   385
    template <typename Result>
kpeter@487
   386
    struct ShiftMultiplier<Result, 64> {
alpar@10
   387
      static const Result multiplier() {
alpar@209
   388
        return static_cast<Result>(1.0/18446744073709551616.0);
alpar@10
   389
      }
alpar@10
   390
    };
alpar@10
   391
alpar@10
   392
    template <typename Result, int exp>
alpar@10
   393
    struct Shifting {
alpar@10
   394
      static Result shift(const Result& result) {
alpar@10
   395
        return result * ShiftMultiplier<Result, exp>::multiplier();
alpar@10
   396
      }
alpar@10
   397
    };
alpar@10
   398
alpar@10
   399
    template <typename Result, typename Word,
alpar@209
   400
              int rest = std::numeric_limits<Result>::digits, int shift = 0,
alpar@10
   401
              bool last = rest <= std::numeric_limits<Word>::digits>
alpar@209
   402
    struct RealConversion{
alpar@10
   403
      static const int bits = std::numeric_limits<Word>::digits;
alpar@10
   404
alpar@10
   405
      static Result convert(RandomCore<Word>& rnd) {
kpeter@487
   406
        return Shifting<Result, shift + rest>::
alpar@10
   407
          shift(static_cast<Result>(rnd() >> (bits - rest)));
alpar@10
   408
      }
alpar@10
   409
    };
alpar@10
   410
alpar@10
   411
    template <typename Result, typename Word, int rest, int shift>
alpar@209
   412
    struct RealConversion<Result, Word, rest, shift, false> {
alpar@10
   413
      static const int bits = std::numeric_limits<Word>::digits;
alpar@10
   414
alpar@10
   415
      static Result convert(RandomCore<Word>& rnd) {
kpeter@487
   416
        return Shifting<Result, shift + bits>::
alpar@10
   417
          shift(static_cast<Result>(rnd())) +
alpar@10
   418
          RealConversion<Result, Word, rest-bits, shift + bits>::
alpar@10
   419
          convert(rnd);
alpar@10
   420
      }
alpar@10
   421
    };
alpar@10
   422
alpar@10
   423
    template <typename Result, typename Word>
alpar@10
   424
    struct Initializer {
alpar@10
   425
alpar@10
   426
      template <typename Iterator>
alpar@10
   427
      static void init(RandomCore<Word>& rnd, Iterator begin, Iterator end) {
alpar@10
   428
        std::vector<Word> ws;
alpar@10
   429
        for (Iterator it = begin; it != end; ++it) {
alpar@10
   430
          ws.push_back(Word(*it));
alpar@10
   431
        }
alpar@10
   432
        rnd.initState(ws.begin(), ws.end());
alpar@10
   433
      }
alpar@10
   434
alpar@10
   435
      static void init(RandomCore<Word>& rnd, Result seed) {
alpar@10
   436
        rnd.initState(seed);
alpar@10
   437
      }
alpar@10
   438
    };
alpar@10
   439
alpar@10
   440
    template <typename Word>
alpar@10
   441
    struct BoolConversion {
alpar@10
   442
      static bool convert(RandomCore<Word>& rnd) {
alpar@10
   443
        return (rnd() & 1) == 1;
alpar@10
   444
      }
alpar@10
   445
    };
alpar@10
   446
alpar@10
   447
    template <typename Word>
alpar@10
   448
    struct BoolProducer {
alpar@10
   449
      Word buffer;
alpar@10
   450
      int num;
alpar@209
   451
alpar@10
   452
      BoolProducer() : num(0) {}
alpar@10
   453
alpar@10
   454
      bool convert(RandomCore<Word>& rnd) {
alpar@10
   455
        if (num == 0) {
alpar@10
   456
          buffer = rnd();
alpar@10
   457
          num = RandomTraits<Word>::bits;
alpar@10
   458
        }
alpar@10
   459
        bool r = (buffer & 1);
alpar@10
   460
        buffer >>= 1;
alpar@10
   461
        --num;
alpar@10
   462
        return r;
alpar@10
   463
      }
alpar@10
   464
    };
alpar@10
   465
alpar@10
   466
  }
alpar@10
   467
alpar@10
   468
  /// \ingroup misc
alpar@10
   469
  ///
alpar@10
   470
  /// \brief Mersenne Twister random number generator
alpar@10
   471
  ///
alpar@10
   472
  /// The Mersenne Twister is a twisted generalized feedback
alpar@10
   473
  /// shift-register generator of Matsumoto and Nishimura. The period
alpar@10
   474
  /// of this generator is \f$ 2^{19937} - 1 \f$ and it is
alpar@10
   475
  /// equi-distributed in 623 dimensions for 32-bit numbers. The time
alpar@10
   476
  /// performance of this generator is comparable to the commonly used
alpar@10
   477
  /// generators.
alpar@10
   478
  ///
alpar@10
   479
  /// This implementation is specialized for both 32-bit and 64-bit
alpar@10
   480
  /// architectures. The generators differ sligthly in the
alpar@10
   481
  /// initialization and generation phase so they produce two
alpar@10
   482
  /// completly different sequences.
alpar@10
   483
  ///
alpar@10
   484
  /// The generator gives back random numbers of serveral types. To
alpar@10
   485
  /// get a random number from a range of a floating point type you
alpar@10
   486
  /// can use one form of the \c operator() or the \c real() member
alpar@10
   487
  /// function. If you want to get random number from the {0, 1, ...,
alpar@10
   488
  /// n-1} integer range use the \c operator[] or the \c integer()
alpar@10
   489
  /// method. And to get random number from the whole range of an
alpar@10
   490
  /// integer type you can use the argumentless \c integer() or \c
alpar@10
   491
  /// uinteger() functions. After all you can get random bool with
alpar@10
   492
  /// equal chance of true and false or given probability of true
alpar@10
   493
  /// result with the \c boolean() member functions.
alpar@10
   494
  ///
alpar@10
   495
  ///\code
alpar@10
   496
  /// // The commented code is identical to the other
alpar@10
   497
  /// double a = rnd();                     // [0.0, 1.0)
alpar@10
   498
  /// // double a = rnd.real();             // [0.0, 1.0)
alpar@10
   499
  /// double b = rnd(100.0);                // [0.0, 100.0)
alpar@10
   500
  /// // double b = rnd.real(100.0);        // [0.0, 100.0)
alpar@10
   501
  /// double c = rnd(1.0, 2.0);             // [1.0, 2.0)
alpar@10
   502
  /// // double c = rnd.real(1.0, 2.0);     // [1.0, 2.0)
alpar@10
   503
  /// int d = rnd[100000];                  // 0..99999
alpar@10
   504
  /// // int d = rnd.integer(100000);       // 0..99999
alpar@10
   505
  /// int e = rnd[6] + 1;                   // 1..6
alpar@10
   506
  /// // int e = rnd.integer(1, 1 + 6);     // 1..6
alpar@10
   507
  /// int b = rnd.uinteger<int>();          // 0 .. 2^31 - 1
alpar@10
   508
  /// int c = rnd.integer<int>();           // - 2^31 .. 2^31 - 1
alpar@10
   509
  /// bool g = rnd.boolean();               // P(g = true) = 0.5
alpar@10
   510
  /// bool h = rnd.boolean(0.8);            // P(h = true) = 0.8
alpar@10
   511
  ///\endcode
alpar@10
   512
  ///
kpeter@49
   513
  /// LEMON provides a global instance of the random number
alpar@10
   514
  /// generator which name is \ref lemon::rnd "rnd". Usually it is a
alpar@10
   515
  /// good programming convenience to use this global generator to get
alpar@10
   516
  /// random numbers.
alpar@10
   517
  class Random {
alpar@10
   518
  private:
alpar@10
   519
kpeter@16
   520
    // Architecture word
alpar@10
   521
    typedef unsigned long Word;
alpar@209
   522
alpar@10
   523
    _random_bits::RandomCore<Word> core;
alpar@10
   524
    _random_bits::BoolProducer<Word> bool_producer;
alpar@209
   525
alpar@10
   526
alpar@10
   527
  public:
alpar@10
   528
deba@177
   529
    ///\name Initialization
deba@177
   530
    ///
deba@177
   531
    /// @{
deba@177
   532
kpeter@49
   533
    /// \brief Default constructor
alpar@10
   534
    ///
alpar@10
   535
    /// Constructor with constant seeding.
alpar@10
   536
    Random() { core.initState(); }
alpar@10
   537
kpeter@49
   538
    /// \brief Constructor with seed
alpar@10
   539
    ///
alpar@10
   540
    /// Constructor with seed. The current number type will be converted
alpar@10
   541
    /// to the architecture word type.
alpar@10
   542
    template <typename Number>
alpar@209
   543
    Random(Number seed) {
alpar@10
   544
      _random_bits::Initializer<Number, Word>::init(core, seed);
alpar@10
   545
    }
alpar@10
   546
kpeter@49
   547
    /// \brief Constructor with array seeding
alpar@10
   548
    ///
alpar@10
   549
    /// Constructor with array seeding. The given range should contain
alpar@10
   550
    /// any number type and the numbers will be converted to the
alpar@10
   551
    /// architecture word type.
alpar@10
   552
    template <typename Iterator>
alpar@209
   553
    Random(Iterator begin, Iterator end) {
alpar@10
   554
      typedef typename std::iterator_traits<Iterator>::value_type Number;
alpar@10
   555
      _random_bits::Initializer<Number, Word>::init(core, begin, end);
alpar@10
   556
    }
alpar@10
   557
alpar@10
   558
    /// \brief Copy constructor
alpar@10
   559
    ///
alpar@10
   560
    /// Copy constructor. The generated sequence will be identical to
alpar@10
   561
    /// the other sequence. It can be used to save the current state
alpar@10
   562
    /// of the generator and later use it to generate the same
alpar@10
   563
    /// sequence.
alpar@10
   564
    Random(const Random& other) {
alpar@10
   565
      core.copyState(other.core);
alpar@10
   566
    }
alpar@10
   567
alpar@10
   568
    /// \brief Assign operator
alpar@10
   569
    ///
alpar@10
   570
    /// Assign operator. The generated sequence will be identical to
alpar@10
   571
    /// the other sequence. It can be used to save the current state
alpar@10
   572
    /// of the generator and later use it to generate the same
alpar@10
   573
    /// sequence.
alpar@10
   574
    Random& operator=(const Random& other) {
alpar@10
   575
      if (&other != this) {
alpar@10
   576
        core.copyState(other.core);
alpar@10
   577
      }
alpar@10
   578
      return *this;
alpar@10
   579
    }
alpar@10
   580
deba@102
   581
    /// \brief Seeding random sequence
deba@102
   582
    ///
deba@102
   583
    /// Seeding the random sequence. The current number type will be
deba@102
   584
    /// converted to the architecture word type.
deba@102
   585
    template <typename Number>
alpar@209
   586
    void seed(Number seed) {
deba@102
   587
      _random_bits::Initializer<Number, Word>::init(core, seed);
deba@102
   588
    }
deba@102
   589
deba@102
   590
    /// \brief Seeding random sequence
deba@102
   591
    ///
deba@102
   592
    /// Seeding the random sequence. The given range should contain
deba@102
   593
    /// any number type and the numbers will be converted to the
deba@102
   594
    /// architecture word type.
deba@102
   595
    template <typename Iterator>
alpar@209
   596
    void seed(Iterator begin, Iterator end) {
deba@102
   597
      typedef typename std::iterator_traits<Iterator>::value_type Number;
deba@102
   598
      _random_bits::Initializer<Number, Word>::init(core, begin, end);
deba@102
   599
    }
deba@102
   600
deba@177
   601
    /// \brief Seeding from file or from process id and time
deba@177
   602
    ///
deba@177
   603
    /// By default, this function calls the \c seedFromFile() member
alpar@178
   604
    /// function with the <tt>/dev/urandom</tt> file. If it does not success,
deba@177
   605
    /// it uses the \c seedFromTime().
kpeter@559
   606
    /// \return Currently always \c true.
deba@177
   607
    bool seed() {
deba@177
   608
#ifndef WIN32
deba@177
   609
      if (seedFromFile("/dev/urandom", 0)) return true;
deba@177
   610
#endif
deba@177
   611
      if (seedFromTime()) return true;
deba@177
   612
      return false;
deba@177
   613
    }
alpar@209
   614
deba@177
   615
    /// \brief Seeding from file
deba@177
   616
    ///
deba@177
   617
    /// Seeding the random sequence from file. The linux kernel has two
deba@177
   618
    /// devices, <tt>/dev/random</tt> and <tt>/dev/urandom</tt> which
deba@177
   619
    /// could give good seed values for pseudo random generators (The
deba@177
   620
    /// difference between two devices is that the <tt>random</tt> may
deba@177
   621
    /// block the reading operation while the kernel can give good
deba@177
   622
    /// source of randomness, while the <tt>urandom</tt> does not
deba@177
   623
    /// block the input, but it could give back bytes with worse
deba@177
   624
    /// entropy).
deba@177
   625
    /// \param file The source file
deba@177
   626
    /// \param offset The offset, from the file read.
kpeter@559
   627
    /// \return \c true when the seeding successes.
deba@177
   628
#ifndef WIN32
alpar@209
   629
    bool seedFromFile(const std::string& file = "/dev/urandom", int offset = 0)
deba@177
   630
#else
alpar@209
   631
    bool seedFromFile(const std::string& file = "", int offset = 0)
deba@177
   632
#endif
deba@177
   633
    {
deba@177
   634
      std::ifstream rs(file.c_str());
deba@177
   635
      const int size = 4;
deba@177
   636
      Word buf[size];
deba@177
   637
      if (offset != 0 && !rs.seekg(offset)) return false;
deba@177
   638
      if (!rs.read(reinterpret_cast<char*>(buf), sizeof(buf))) return false;
deba@177
   639
      seed(buf, buf + size);
deba@177
   640
      return true;
deba@177
   641
    }
deba@177
   642
deba@177
   643
    /// \brief Seding from process id and time
deba@177
   644
    ///
deba@177
   645
    /// Seding from process id and time. This function uses the
deba@177
   646
    /// current process id and the current time for initialize the
deba@177
   647
    /// random sequence.
kpeter@559
   648
    /// \return Currently always \c true.
alpar@209
   649
    bool seedFromTime() {
deba@177
   650
#ifndef WIN32
deba@177
   651
      timeval tv;
deba@177
   652
      gettimeofday(&tv, 0);
deba@177
   653
      seed(getpid() + tv.tv_sec + tv.tv_usec);
deba@177
   654
#else
alpar@482
   655
      seed(bits::getWinRndSeed());
deba@177
   656
#endif
deba@177
   657
      return true;
deba@177
   658
    }
deba@177
   659
deba@177
   660
    /// @}
deba@177
   661
kpeter@584
   662
    ///\name Uniform Distributions
deba@177
   663
    ///
deba@177
   664
    /// @{
deba@177
   665
alpar@10
   666
    /// \brief Returns a random real number from the range [0, 1)
alpar@10
   667
    ///
alpar@10
   668
    /// It returns a random real number from the range [0, 1). The
kpeter@49
   669
    /// default Number type is \c double.
alpar@10
   670
    template <typename Number>
alpar@10
   671
    Number real() {
alpar@10
   672
      return _random_bits::RealConversion<Number, Word>::convert(core);
alpar@10
   673
    }
alpar@10
   674
alpar@10
   675
    double real() {
alpar@10
   676
      return real<double>();
alpar@10
   677
    }
alpar@10
   678
alpar@10
   679
    /// \brief Returns a random real number from the range [0, 1)
alpar@10
   680
    ///
alpar@10
   681
    /// It returns a random double from the range [0, 1).
alpar@10
   682
    double operator()() {
alpar@10
   683
      return real<double>();
alpar@10
   684
    }
alpar@10
   685
alpar@10
   686
    /// \brief Returns a random real number from the range [0, b)
alpar@10
   687
    ///
alpar@10
   688
    /// It returns a random real number from the range [0, b).
alpar@377
   689
    double operator()(double b) {
alpar@377
   690
      return real<double>() * b;
alpar@10
   691
    }
alpar@10
   692
alpar@10
   693
    /// \brief Returns a random real number from the range [a, b)
alpar@10
   694
    ///
alpar@10
   695
    /// It returns a random real number from the range [a, b).
alpar@377
   696
    double operator()(double a, double b) {
alpar@377
   697
      return real<double>() * (b - a) + a;
alpar@10
   698
    }
alpar@10
   699
alpar@10
   700
    /// \brief Returns a random integer from a range
alpar@10
   701
    ///
alpar@10
   702
    /// It returns a random integer from the range {0, 1, ..., b - 1}.
alpar@10
   703
    template <typename Number>
alpar@10
   704
    Number integer(Number b) {
alpar@10
   705
      return _random_bits::Mapping<Number, Word>::map(core, b);
alpar@10
   706
    }
alpar@10
   707
alpar@10
   708
    /// \brief Returns a random integer from a range
alpar@10
   709
    ///
alpar@10
   710
    /// It returns a random integer from the range {a, a + 1, ..., b - 1}.
alpar@10
   711
    template <typename Number>
alpar@10
   712
    Number integer(Number a, Number b) {
alpar@10
   713
      return _random_bits::Mapping<Number, Word>::map(core, b - a) + a;
alpar@10
   714
    }
alpar@10
   715
alpar@10
   716
    /// \brief Returns a random integer from a range
alpar@10
   717
    ///
alpar@10
   718
    /// It returns a random integer from the range {0, 1, ..., b - 1}.
alpar@10
   719
    template <typename Number>
alpar@10
   720
    Number operator[](Number b) {
alpar@10
   721
      return _random_bits::Mapping<Number, Word>::map(core, b);
alpar@10
   722
    }
alpar@10
   723
alpar@10
   724
    /// \brief Returns a random non-negative integer
alpar@10
   725
    ///
alpar@10
   726
    /// It returns a random non-negative integer uniformly from the
kpeter@49
   727
    /// whole range of the current \c Number type. The default result
kpeter@49
   728
    /// type of this function is <tt>unsigned int</tt>.
alpar@10
   729
    template <typename Number>
alpar@10
   730
    Number uinteger() {
alpar@10
   731
      return _random_bits::IntConversion<Number, Word>::convert(core);
alpar@10
   732
    }
alpar@10
   733
alpar@10
   734
    unsigned int uinteger() {
alpar@10
   735
      return uinteger<unsigned int>();
alpar@10
   736
    }
alpar@10
   737
alpar@10
   738
    /// \brief Returns a random integer
alpar@10
   739
    ///
alpar@10
   740
    /// It returns a random integer uniformly from the whole range of
alpar@10
   741
    /// the current \c Number type. The default result type of this
kpeter@49
   742
    /// function is \c int.
alpar@10
   743
    template <typename Number>
alpar@10
   744
    Number integer() {
alpar@209
   745
      static const int nb = std::numeric_limits<Number>::digits +
alpar@10
   746
        (std::numeric_limits<Number>::is_signed ? 1 : 0);
alpar@10
   747
      return _random_bits::IntConversion<Number, Word, nb>::convert(core);
alpar@10
   748
    }
alpar@10
   749
alpar@10
   750
    int integer() {
alpar@10
   751
      return integer<int>();
alpar@10
   752
    }
alpar@209
   753
alpar@10
   754
    /// \brief Returns a random bool
alpar@10
   755
    ///
alpar@10
   756
    /// It returns a random bool. The generator holds a buffer for
alpar@10
   757
    /// random bits. Every time when it become empty the generator makes
alpar@10
   758
    /// a new random word and fill the buffer up.
alpar@10
   759
    bool boolean() {
alpar@10
   760
      return bool_producer.convert(core);
alpar@10
   761
    }
alpar@10
   762
deba@177
   763
    /// @}
deba@177
   764
kpeter@584
   765
    ///\name Non-uniform Distributions
alpar@10
   766
    ///
alpar@10
   767
    ///@{
alpar@209
   768
kpeter@340
   769
    /// \brief Returns a random bool with given probability of true result.
alpar@10
   770
    ///
kpeter@23
   771
    /// It returns a random bool with given probability of true result.
alpar@10
   772
    bool boolean(double p) {
alpar@10
   773
      return operator()() < p;
alpar@10
   774
    }
alpar@10
   775
kpeter@340
   776
    /// Standard normal (Gauss) distribution
alpar@10
   777
kpeter@340
   778
    /// Standard normal (Gauss) distribution.
alpar@10
   779
    /// \note The Cartesian form of the Box-Muller
alpar@10
   780
    /// transformation is used to generate a random normal distribution.
alpar@209
   781
    double gauss()
alpar@10
   782
    {
alpar@10
   783
      double V1,V2,S;
alpar@10
   784
      do {
alpar@209
   785
        V1=2*real<double>()-1;
alpar@209
   786
        V2=2*real<double>()-1;
alpar@209
   787
        S=V1*V1+V2*V2;
alpar@10
   788
      } while(S>=1);
alpar@10
   789
      return std::sqrt(-2*std::log(S)/S)*V1;
alpar@10
   790
    }
kpeter@340
   791
    /// Normal (Gauss) distribution with given mean and standard deviation
alpar@10
   792
kpeter@340
   793
    /// Normal (Gauss) distribution with given mean and standard deviation.
alpar@10
   794
    /// \sa gauss()
alpar@10
   795
    double gauss(double mean,double std_dev)
alpar@10
   796
    {
alpar@10
   797
      return gauss()*std_dev+mean;
alpar@10
   798
    }
alpar@10
   799
alpar@339
   800
    /// Lognormal distribution
alpar@339
   801
alpar@339
   802
    /// Lognormal distribution. The parameters are the mean and the standard
alpar@339
   803
    /// deviation of <tt>exp(X)</tt>.
alpar@339
   804
    ///
alpar@339
   805
    double lognormal(double n_mean,double n_std_dev)
alpar@339
   806
    {
alpar@339
   807
      return std::exp(gauss(n_mean,n_std_dev));
alpar@339
   808
    }
alpar@339
   809
    /// Lognormal distribution
alpar@339
   810
alpar@339
   811
    /// Lognormal distribution. The parameter is an <tt>std::pair</tt> of
alpar@339
   812
    /// the mean and the standard deviation of <tt>exp(X)</tt>.
alpar@339
   813
    ///
alpar@339
   814
    double lognormal(const std::pair<double,double> &params)
alpar@339
   815
    {
alpar@339
   816
      return std::exp(gauss(params.first,params.second));
alpar@339
   817
    }
alpar@339
   818
    /// Compute the lognormal parameters from mean and standard deviation
alpar@339
   819
alpar@339
   820
    /// This function computes the lognormal parameters from mean and
alpar@339
   821
    /// standard deviation. The return value can direcly be passed to
alpar@339
   822
    /// lognormal().
alpar@339
   823
    std::pair<double,double> lognormalParamsFromMD(double mean,
kpeter@340
   824
                                                   double std_dev)
alpar@339
   825
    {
alpar@339
   826
      double fr=std_dev/mean;
alpar@339
   827
      fr*=fr;
alpar@339
   828
      double lg=std::log(1+fr);
alpar@339
   829
      return std::pair<double,double>(std::log(mean)-lg/2.0,std::sqrt(lg));
alpar@339
   830
    }
alpar@339
   831
    /// Lognormal distribution with given mean and standard deviation
kpeter@340
   832
alpar@339
   833
    /// Lognormal distribution with given mean and standard deviation.
alpar@339
   834
    ///
alpar@339
   835
    double lognormalMD(double mean,double std_dev)
alpar@339
   836
    {
alpar@339
   837
      return lognormal(lognormalParamsFromMD(mean,std_dev));
alpar@339
   838
    }
kpeter@340
   839
alpar@10
   840
    /// Exponential distribution with given mean
alpar@10
   841
alpar@10
   842
    /// This function generates an exponential distribution random number
alpar@10
   843
    /// with mean <tt>1/lambda</tt>.
alpar@10
   844
    ///
alpar@10
   845
    double exponential(double lambda=1.0)
alpar@10
   846
    {
alpar@11
   847
      return -std::log(1.0-real<double>())/lambda;
alpar@10
   848
    }
alpar@10
   849
alpar@10
   850
    /// Gamma distribution with given integer shape
alpar@10
   851
alpar@10
   852
    /// This function generates a gamma distribution random number.
alpar@209
   853
    ///
alpar@10
   854
    ///\param k shape parameter (<tt>k>0</tt> integer)
alpar@209
   855
    double gamma(int k)
alpar@10
   856
    {
alpar@10
   857
      double s = 0;
alpar@10
   858
      for(int i=0;i<k;i++) s-=std::log(1.0-real<double>());
alpar@10
   859
      return s;
alpar@10
   860
    }
alpar@209
   861
alpar@10
   862
    /// Gamma distribution with given shape and scale parameter
alpar@10
   863
alpar@10
   864
    /// This function generates a gamma distribution random number.
alpar@209
   865
    ///
alpar@10
   866
    ///\param k shape parameter (<tt>k>0</tt>)
alpar@10
   867
    ///\param theta scale parameter
alpar@10
   868
    ///
alpar@10
   869
    double gamma(double k,double theta=1.0)
alpar@10
   870
    {
alpar@10
   871
      double xi,nu;
alpar@10
   872
      const double delta = k-std::floor(k);
alpar@68
   873
      const double v0=E/(E-delta);
alpar@10
   874
      do {
alpar@209
   875
        double V0=1.0-real<double>();
alpar@209
   876
        double V1=1.0-real<double>();
alpar@209
   877
        double V2=1.0-real<double>();
alpar@209
   878
        if(V2<=v0)
alpar@209
   879
          {
alpar@209
   880
            xi=std::pow(V1,1.0/delta);
alpar@209
   881
            nu=V0*std::pow(xi,delta-1.0);
alpar@209
   882
          }
alpar@209
   883
        else
alpar@209
   884
          {
alpar@209
   885
            xi=1.0-std::log(V1);
alpar@209
   886
            nu=V0*std::exp(-xi);
alpar@209
   887
          }
alpar@10
   888
      } while(nu>std::pow(xi,delta-1.0)*std::exp(-xi));
alpar@116
   889
      return theta*(xi+gamma(int(std::floor(k))));
alpar@10
   890
    }
alpar@209
   891
alpar@11
   892
    /// Weibull distribution
alpar@11
   893
alpar@11
   894
    /// This function generates a Weibull distribution random number.
alpar@209
   895
    ///
alpar@11
   896
    ///\param k shape parameter (<tt>k>0</tt>)
alpar@11
   897
    ///\param lambda scale parameter (<tt>lambda>0</tt>)
alpar@11
   898
    ///
alpar@11
   899
    double weibull(double k,double lambda)
alpar@11
   900
    {
alpar@11
   901
      return lambda*pow(-std::log(1.0-real<double>()),1.0/k);
alpar@209
   902
    }
alpar@209
   903
alpar@11
   904
    /// Pareto distribution
alpar@11
   905
alpar@11
   906
    /// This function generates a Pareto distribution random number.
alpar@209
   907
    ///
alpar@12
   908
    ///\param k shape parameter (<tt>k>0</tt>)
alpar@11
   909
    ///\param x_min location parameter (<tt>x_min>0</tt>)
alpar@11
   910
    ///
alpar@12
   911
    double pareto(double k,double x_min)
alpar@11
   912
    {
alpar@116
   913
      return exponential(gamma(k,1.0/x_min))+x_min;
alpar@209
   914
    }
alpar@209
   915
alpar@92
   916
    /// Poisson distribution
alpar@92
   917
alpar@92
   918
    /// This function generates a Poisson distribution random number with
alpar@92
   919
    /// parameter \c lambda.
alpar@209
   920
    ///
alpar@92
   921
    /// The probability mass function of this distribusion is
alpar@92
   922
    /// \f[ \frac{e^{-\lambda}\lambda^k}{k!} \f]
alpar@92
   923
    /// \note The algorithm is taken from the book of Donald E. Knuth titled
alpar@92
   924
    /// ''Seminumerical Algorithms'' (1969). Its running time is linear in the
alpar@92
   925
    /// return value.
alpar@209
   926
alpar@92
   927
    int poisson(double lambda)
alpar@92
   928
    {
alpar@92
   929
      const double l = std::exp(-lambda);
alpar@92
   930
      int k=0;
alpar@92
   931
      double p = 1.0;
alpar@92
   932
      do {
alpar@209
   933
        k++;
alpar@209
   934
        p*=real<double>();
alpar@92
   935
      } while (p>=l);
alpar@92
   936
      return k-1;
alpar@209
   937
    }
alpar@209
   938
alpar@10
   939
    ///@}
alpar@209
   940
kpeter@584
   941
    ///\name Two Dimensional Distributions
alpar@10
   942
    ///
alpar@10
   943
    ///@{
alpar@209
   944
kpeter@23
   945
    /// Uniform distribution on the full unit circle
kpeter@16
   946
kpeter@16
   947
    /// Uniform distribution on the full unit circle.
kpeter@16
   948
    ///
alpar@209
   949
    dim2::Point<double> disc()
alpar@10
   950
    {
alpar@10
   951
      double V1,V2;
alpar@10
   952
      do {
alpar@209
   953
        V1=2*real<double>()-1;
alpar@209
   954
        V2=2*real<double>()-1;
alpar@209
   955
alpar@10
   956
      } while(V1*V1+V2*V2>=1);
alpar@10
   957
      return dim2::Point<double>(V1,V2);
alpar@10
   958
    }
kpeter@340
   959
    /// A kind of two dimensional normal (Gauss) distribution
alpar@10
   960
alpar@10
   961
    /// This function provides a turning symmetric two-dimensional distribution.
alpar@10
   962
    /// Both coordinates are of standard normal distribution, but they are not
alpar@10
   963
    /// independent.
alpar@10
   964
    ///
alpar@10
   965
    /// \note The coordinates are the two random variables provided by
alpar@10
   966
    /// the Box-Muller method.
alpar@10
   967
    dim2::Point<double> gauss2()
alpar@10
   968
    {
alpar@10
   969
      double V1,V2,S;
alpar@10
   970
      do {
alpar@209
   971
        V1=2*real<double>()-1;
alpar@209
   972
        V2=2*real<double>()-1;
alpar@209
   973
        S=V1*V1+V2*V2;
alpar@10
   974
      } while(S>=1);
alpar@10
   975
      double W=std::sqrt(-2*std::log(S)/S);
alpar@10
   976
      return dim2::Point<double>(W*V1,W*V2);
alpar@10
   977
    }
alpar@10
   978
    /// A kind of two dimensional exponential distribution
alpar@10
   979
alpar@10
   980
    /// This function provides a turning symmetric two-dimensional distribution.
alpar@10
   981
    /// The x-coordinate is of conditionally exponential distribution
alpar@209
   982
    /// with the condition that x is positive and y=0. If x is negative and
alpar@10
   983
    /// y=0 then, -x is of exponential distribution. The same is true for the
alpar@10
   984
    /// y-coordinate.
alpar@209
   985
    dim2::Point<double> exponential2()
alpar@10
   986
    {
alpar@10
   987
      double V1,V2,S;
alpar@10
   988
      do {
alpar@209
   989
        V1=2*real<double>()-1;
alpar@209
   990
        V2=2*real<double>()-1;
alpar@209
   991
        S=V1*V1+V2*V2;
alpar@10
   992
      } while(S>=1);
alpar@10
   993
      double W=-std::log(S)/S;
alpar@10
   994
      return dim2::Point<double>(W*V1,W*V2);
alpar@10
   995
    }
alpar@10
   996
alpar@209
   997
    ///@}
alpar@10
   998
  };
alpar@10
   999
alpar@10
  1000
alpar@10
  1001
  extern Random rnd;
alpar@10
  1002
alpar@10
  1003
}
alpar@10
  1004
alpar@10
  1005
#endif