| 1 | /*------------------------------------------------------------------------- |
| 2 | * |
| 3 | * geqo_selection.c |
| 4 | * linear selection scheme for the genetic query optimizer |
| 5 | * |
| 6 | * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group |
| 7 | * Portions Copyright (c) 1994, Regents of the University of California |
| 8 | * |
| 9 | * src/backend/optimizer/geqo/geqo_selection.c |
| 10 | * |
| 11 | *------------------------------------------------------------------------- |
| 12 | */ |
| 13 | |
| 14 | /* contributed by: |
| 15 | =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= |
| 16 | * Martin Utesch * Institute of Automatic Control * |
| 17 | = = University of Mining and Technology = |
| 18 | * utesch@aut.tu-freiberg.de * Freiberg, Germany * |
| 19 | =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= |
| 20 | */ |
| 21 | |
| 22 | /* this is adopted from D. Whitley's Genitor algorithm */ |
| 23 | |
| 24 | /*************************************************************/ |
| 25 | /* */ |
| 26 | /* Copyright (c) 1990 */ |
| 27 | /* Darrell L. Whitley */ |
| 28 | /* Computer Science Department */ |
| 29 | /* Colorado State University */ |
| 30 | /* */ |
| 31 | /* Permission is hereby granted to copy all or any part of */ |
| 32 | /* this program for free distribution. The author's name */ |
| 33 | /* and this copyright notice must be included in any copy. */ |
| 34 | /* */ |
| 35 | /*************************************************************/ |
| 36 | |
| 37 | #include "postgres.h" |
| 38 | |
| 39 | #include <math.h> |
| 40 | |
| 41 | #include "optimizer/geqo_copy.h" |
| 42 | #include "optimizer/geqo_random.h" |
| 43 | #include "optimizer/geqo_selection.h" |
| 44 | |
| 45 | static int linear_rand(PlannerInfo *root, int max, double bias); |
| 46 | |
| 47 | |
| 48 | /* |
| 49 | * geqo_selection |
| 50 | * according to bias described by input parameters, |
| 51 | * first and second genes are selected from the pool |
| 52 | */ |
| 53 | void |
| 54 | geqo_selection(PlannerInfo *root, Chromosome *momma, Chromosome *daddy, |
| 55 | Pool *pool, double bias) |
| 56 | { |
| 57 | int first, |
| 58 | second; |
| 59 | |
| 60 | first = linear_rand(root, pool->size, bias); |
| 61 | second = linear_rand(root, pool->size, bias); |
| 62 | |
| 63 | /* |
| 64 | * Ensure we have selected different genes, except if pool size is only |
| 65 | * one, when we can't. |
| 66 | * |
| 67 | * This code was observed to hang up in an infinite loop when the |
| 68 | * platform's implementation of erand48() was broken. We now always use |
| 69 | * our own version. |
| 70 | */ |
| 71 | if (pool->size > 1) |
| 72 | { |
| 73 | while (first == second) |
| 74 | second = linear_rand(root, pool->size, bias); |
| 75 | } |
| 76 | |
| 77 | geqo_copy(root, momma, &pool->data[first], pool->string_length); |
| 78 | geqo_copy(root, daddy, &pool->data[second], pool->string_length); |
| 79 | } |
| 80 | |
| 81 | /* |
| 82 | * linear_rand |
| 83 | * generates random integer between 0 and input max number |
| 84 | * using input linear bias |
| 85 | * |
| 86 | * bias is y-intercept of linear distribution |
| 87 | * |
| 88 | * probability distribution function is: f(x) = bias - 2(bias - 1)x |
| 89 | * bias = (prob of first rule) / (prob of middle rule) |
| 90 | */ |
| 91 | static int |
| 92 | linear_rand(PlannerInfo *root, int pool_size, double bias) |
| 93 | { |
| 94 | double index; /* index between 0 and pop_size */ |
| 95 | double max = (double) pool_size; |
| 96 | |
| 97 | /* |
| 98 | * If geqo_rand() returns exactly 1.0 then we will get exactly max from |
| 99 | * this equation, whereas we need 0 <= index < max. Also it seems |
| 100 | * possible that roundoff error might deliver values slightly outside the |
| 101 | * range; in particular avoid passing a value slightly less than 0 to |
| 102 | * sqrt(). If we get a bad value just try again. |
| 103 | */ |
| 104 | do |
| 105 | { |
| 106 | double sqrtval; |
| 107 | |
| 108 | sqrtval = (bias * bias) - 4.0 * (bias - 1.0) * geqo_rand(root); |
| 109 | if (sqrtval > 0.0) |
| 110 | sqrtval = sqrt(sqrtval); |
| 111 | index = max * (bias - sqrtval) / 2.0 / (bias - 1.0); |
| 112 | } while (index < 0.0 || index >= max); |
| 113 | |
| 114 | return (int) index; |
| 115 | } |
| 116 | |