X-Git-Url: http://info.iut-bm.univ-fcomte.fr/pub/gitweb/simgrid.git/blobdiff_plain/a4a0b7a61683f71695edc5b66c554d209311699f..65e200cd6cca5e000376f689bd6b51d0a1fa886b:/src/surf/lagrange.cpp diff --git a/src/surf/lagrange.cpp b/src/surf/lagrange.cpp index e82b60f5d7..e81ea8e96f 100644 --- a/src/surf/lagrange.cpp +++ b/src/surf/lagrange.cpp @@ -1,5 +1,4 @@ -/* Copyright (c) 2007-2014. The SimGrid Team. - * All rights reserved. */ +/* Copyright (c) 2007-2017. The SimGrid Team. All rights reserved. */ /* This program is free software; you can redistribute it and/or modify it * under the terms of the license (GNU LGPL) which comes with this package. */ @@ -8,19 +7,26 @@ * Modeling the proportional fairness using the Lagrangian Optimization Approach. For a detailed description see: * "ssh://username@scm.gforge.inria.fr/svn/memo/people/pvelho/lagrange/ppf.ps". */ +#include "surf/maxmin.hpp" #include "xbt/log.h" #include "xbt/sysdep.h" -#include "maxmin_private.hpp" -#include +#include +#include #ifndef MATH -#include +#include #endif XBT_LOG_NEW_DEFAULT_SUBCATEGORY(surf_lagrange, surf, "Logging specific to SURF (lagrange)"); XBT_LOG_NEW_SUBCATEGORY(surf_lagrange_dichotomy, surf_lagrange, "Logging specific to SURF (lagrange dichotomy)"); #define SHOW_EXPR(expr) XBT_CDEBUG(surf_lagrange,#expr " = %g",expr); +#define VEGAS_SCALING 1000.0 +#define RENO_SCALING 1.0 +#define RENO2_SCALING 1.0 + +namespace simgrid { +namespace surf { double (*func_f_def) (lmm_variable_t, double); double (*func_fp_def) (lmm_variable_t, double); @@ -33,27 +39,27 @@ double (*func_fpi_def) (lmm_variable_t, double); void lagrange_solve(lmm_system_t sys); //computes the value of the dichotomy using a initial values, init, with a specific variable or constraint static double dichotomy(double init, double diff(double, void *), void *var_cnst, double min_error); -//computes the value of the differential of constraint param_cnst applied to lambda +//computes the value of the differential of constraint param_cnst applied to lambda static double partial_diff_lambda(double lambda, void *param_cnst); static int __check_feasible(xbt_swag_t cnst_list, xbt_swag_t var_list, int warn) { - void *_cnst, *_elem, *_var; + void* _cnst; + void* _elem; + void* _var; xbt_swag_t elem_list = nullptr; lmm_element_t elem = nullptr; lmm_constraint_t cnst = nullptr; lmm_variable_t var = nullptr; - double tmp; - xbt_swag_foreach(_cnst, cnst_list) { - cnst = static_cast(_cnst); - tmp = 0; + cnst = static_cast(_cnst); + double tmp = 0; elem_list = &(cnst->enabled_element_set); xbt_swag_foreach(_elem, elem_list) { elem = static_cast(_elem); var = elem->variable; - xbt_assert(var->weight > 0); + xbt_assert(var->sharing_weight > 0); tmp += var->value; } @@ -68,7 +74,7 @@ static int __check_feasible(xbt_swag_t cnst_list, xbt_swag_t var_list, int warn) xbt_swag_foreach(_var, var_list) { var = static_cast(_var); - if (not var->weight) + if (not var->sharing_weight) break; if (var->bound < 0) continue; @@ -87,12 +93,12 @@ static double new_value(lmm_variable_t var) { double tmp = 0; - for (int i = 0; i < var->cnsts_number; i++) { - tmp += (var->cnsts[i].constraint)->lambda; + for (s_lmm_element_t const& elem : var->cnsts) { + tmp += elem.constraint->lambda; } if (var->bound > 0) tmp += var->mu; - XBT_DEBUG("\t Working on var (%p). cost = %e; Weight = %e", var, tmp, var->weight); + XBT_DEBUG("\t Working on var (%p). cost = %e; Weight = %e", var, tmp, var->sharing_weight); //uses the partial differential inverse function return var->func_fpi(var, tmp); } @@ -102,8 +108,8 @@ static double new_mu(lmm_variable_t var) double mu_i = 0.0; double sigma_i = 0.0; - for (int j = 0; j < var->cnsts_number; j++) { - sigma_i += (var->cnsts[j].constraint)->lambda; + for (s_lmm_element_t const& elem : var->cnsts) { + sigma_i += elem.constraint->lambda; } mu_i = var->func_fp(var, var->bound) - sigma_i; if (mu_i < 0.0) @@ -124,11 +130,11 @@ static double dual_objective(xbt_swag_t var_list, xbt_swag_t cnst_list) var = static_cast(_var); double sigma_i = 0.0; - if (not var->weight) + if (not var->sharing_weight) break; - for (int j = 0; j < var->cnsts_number; j++) - sigma_i += (var->cnsts[j].constraint)->lambda; + for (s_lmm_element_t const& elem : var->cnsts) + sigma_i += elem.constraint->lambda; if (var->bound > 0) sigma_i += var->mu; @@ -157,81 +163,63 @@ void lagrange_solve(lmm_system_t sys) double dichotomy_min_error = 1e-14; double overall_modification = 1; - /* Variables to manipulate the data structure proposed to model the maxmin fairness. See documentation for details. */ - xbt_swag_t cnst_list = nullptr; - void *_cnst; - lmm_constraint_t cnst = nullptr; - - xbt_swag_t var_list = nullptr; - void *_var; - lmm_variable_t var = nullptr; - - /* Auxiliary variables. */ - int iteration = 0; - double tmp = 0; - int i; - double obj; - double new_obj; - XBT_DEBUG("Iterative method configuration snapshot =====>"); XBT_DEBUG("#### Maximum number of iterations : %d", max_iterations); XBT_DEBUG("#### Minimum error tolerated : %e", epsilon_min_error); XBT_DEBUG("#### Minimum error tolerated (dichotomy) : %e", dichotomy_min_error); if (XBT_LOG_ISENABLED(surf_lagrange, xbt_log_priority_debug)) { - lmm_print(sys); + sys->print(); } if (not sys->modified) return; /* Initialize lambda. */ - cnst_list = &(sys->active_constraint_set); + xbt_swag_t cnst_list = &(sys->active_constraint_set); + void* _cnst; xbt_swag_foreach(_cnst, cnst_list) { - cnst = (lmm_constraint_t)_cnst; + lmm_constraint_t cnst = (lmm_constraint_t)_cnst; cnst->lambda = 1.0; cnst->new_lambda = 2.0; XBT_DEBUG("#### cnst(%p)->lambda : %e", cnst, cnst->lambda); } - /* - * Initialize the var list variable with only the active variables. + /* + * Initialize the var list variable with only the active variables. * Associate an index in the swag variables. Initialize mu. */ - var_list = &(sys->variable_set); - i = 0; + xbt_swag_t var_list = &(sys->variable_set); + void* _var; xbt_swag_foreach(_var, var_list) { - var = static_cast(_var); - if (not var->weight) - var->value = 0.0; - else { - int nb = 0; - if (var->bound < 0.0) { - XBT_DEBUG("#### NOTE var(%d) is a boundless variable", i); - var->mu = -1.0; + lmm_variable_t var = static_cast(_var); + if (not var->sharing_weight) + var->value = 0.0; + else { + if (var->bound < 0.0) { + XBT_DEBUG("#### NOTE var(%p) is a boundless variable", var); + var->mu = -1.0; + } else { + var->mu = 1.0; + var->new_mu = 2.0; + } var->value = new_value(var); - } else { - var->mu = 1.0; - var->new_mu = 2.0; - var->value = new_value(var); - } - XBT_DEBUG("#### var(%p) ->weight : %e", var, var->weight); - XBT_DEBUG("#### var(%p) ->mu : %e", var, var->mu); - XBT_DEBUG("#### var(%p) ->weight: %e", var, var->weight); - XBT_DEBUG("#### var(%p) ->bound: %e", var, var->bound); - for (i = 0; i < var->cnsts_number; i++) { - if (var->cnsts[i].value == 0.0) - nb++; - } - if (nb == var->cnsts_number) - var->value = 1.0; + XBT_DEBUG("#### var(%p) ->weight : %e", var, var->sharing_weight); + XBT_DEBUG("#### var(%p) ->mu : %e", var, var->mu); + XBT_DEBUG("#### var(%p) ->weight: %e", var, var->sharing_weight); + XBT_DEBUG("#### var(%p) ->bound: %e", var, var->bound); + auto weighted = std::find_if(begin(var->cnsts), end(var->cnsts), + [](s_lmm_element_t const& x) { return x.consumption_weight != 0.0; }); + if (weighted == end(var->cnsts)) + var->value = 1.0; } } /* Compute dual objective. */ - obj = dual_objective(var_list, cnst_list); + double obj = dual_objective(var_list, cnst_list); /* While doesn't reach a minimum error or a number maximum of iterations. */ + int iteration = 0; while (overall_modification > epsilon_min_error && iteration < max_iterations) { iteration++; XBT_DEBUG("************** ITERATION %d **************", iteration); @@ -239,18 +227,14 @@ void lagrange_solve(lmm_system_t sys) /* Improve the value of mu_i */ xbt_swag_foreach(_var, var_list) { - var = static_cast(_var); - if (not var->weight) - break; - if (var->bound >= 0) { + lmm_variable_t var = static_cast(_var); + if (var->sharing_weight && var->bound >= 0) { XBT_DEBUG("Working on var (%p)", var); var->new_mu = new_mu(var); -/* dual_updated += (fabs(var->new_mu-var->mu)>dichotomy_min_error); */ -/* XBT_DEBUG("dual_updated (%d) : %1.20f",dual_updated,fabs(var->new_mu-var->mu)); */ XBT_DEBUG("Updating mu : var->mu (%p) : %1.20f -> %1.20f", var, var->mu, var->new_mu); var->mu = var->new_mu; - new_obj = dual_objective(var_list, cnst_list); + double new_obj = dual_objective(var_list, cnst_list); XBT_DEBUG("Improvement for Objective (%g -> %g) : %g", obj, new_obj, obj - new_obj); xbt_assert(obj - new_obj >= -epsilon_min_error, "Our gradient sucks! (%1.20f)", obj - new_obj); obj = new_obj; @@ -259,15 +243,13 @@ void lagrange_solve(lmm_system_t sys) /* Improve the value of lambda_i */ xbt_swag_foreach(_cnst, cnst_list) { - cnst = static_cast(_cnst); + lmm_constraint_t cnst = static_cast(_cnst); XBT_DEBUG("Working on cnst (%p)", cnst); cnst->new_lambda = dichotomy(cnst->lambda, partial_diff_lambda, cnst, dichotomy_min_error); -/* dual_updated += (fabs(cnst->new_lambda-cnst->lambda)>dichotomy_min_error); */ -/* XBT_DEBUG("dual_updated (%d) : %1.20f",dual_updated,fabs(cnst->new_lambda-cnst->lambda)); */ XBT_DEBUG("Updating lambda : cnst->lambda (%p) : %1.20f -> %1.20f", cnst, cnst->lambda, cnst->new_lambda); cnst->lambda = cnst->new_lambda; - new_obj = dual_objective(var_list, cnst_list); + double new_obj = dual_objective(var_list, cnst_list); XBT_DEBUG("Improvement for Objective (%g -> %g) : %g", obj, new_obj, obj - new_obj); xbt_assert(obj - new_obj >= -epsilon_min_error, "Our gradient sucks! (%1.20f)", obj - new_obj); obj = new_obj; @@ -277,13 +259,13 @@ void lagrange_solve(lmm_system_t sys) XBT_DEBUG("-------------- Check convergence ----------"); overall_modification = 0; xbt_swag_foreach(_var, var_list) { - var = static_cast(_var); - if (var->weight <= 0) + lmm_variable_t var = static_cast(_var); + if (var->sharing_weight <= 0) var->value = 0.0; else { - tmp = new_value(var); + double tmp = new_value(var); - overall_modification = MAX(overall_modification, fabs(var->value - tmp)); + overall_modification = std::max(overall_modification, fabs(var->value - tmp)); var->value = tmp; XBT_DEBUG("New value of var (%p) = %e, overall_modification = %e", var, var->value, overall_modification); @@ -291,13 +273,9 @@ void lagrange_solve(lmm_system_t sys) } XBT_DEBUG("-------------- Check feasability ----------"); - if (!__check_feasible(cnst_list, var_list, 0)) + if (not __check_feasible(cnst_list, var_list, 0)) overall_modification = 1.0; XBT_DEBUG("Iteration %d: overall_modification : %f", iteration, overall_modification); - /* if(not dual_updated) { */ - /* XBT_WARN("Could not improve the convergence at iteration %d. Drop it!",iteration); */ - /* break; */ - /* } */ } __check_feasible(cnst_list, var_list, 1); @@ -310,7 +288,7 @@ void lagrange_solve(lmm_system_t sys) } if (XBT_LOG_ISENABLED(surf_lagrange, xbt_log_priority_debug)) { - lmm_print(sys); + sys->print(); } } @@ -320,7 +298,7 @@ void lagrange_solve(lmm_system_t sys) * * @param init initial value for \mu or \lambda * @param diff a function that computes the differential of with respect a \mu or \lambda - * @param var_cnst a pointer to a variable or constraint + * @param var_cnst a pointer to a variable or constraint * @param min_erro a minimum error tolerated * * @return a double corresponding to the result of the dichotomy process @@ -394,25 +372,20 @@ static double dichotomy(double init, double diff(double, void *), void *var_cnst min = middle; overall_error = max_diff - middle_diff; min_diff = middle_diff; -/* SHOW_EXPR(overall_error); */ } else if (middle_diff > 0) { XBT_CDEBUG(surf_lagrange_dichotomy, "Decreasing max"); max = middle; overall_error = max_diff - middle_diff; max_diff = middle_diff; -/* SHOW_EXPR(overall_error); */ } else { overall_error = 0; -/* SHOW_EXPR(overall_error); */ } } else if (fabs(min_diff) < 1e-20) { max = min; overall_error = 0; -/* SHOW_EXPR(overall_error); */ } else if (fabs(max_diff) < 1e-20) { min = max; overall_error = 0; -/* SHOW_EXPR(overall_error); */ } else if (min_diff > 0 && max_diff < 0) { XBT_CWARN(surf_lagrange_dichotomy, "The impossible happened, partial_diff(min) > 0 && partial_diff(max) < 0"); xbt_abort(); @@ -431,31 +404,26 @@ static double dichotomy(double init, double diff(double, void *), void *var_cnst static double partial_diff_lambda(double lambda, void *param_cnst) { - int j; - void *_elem; - xbt_swag_t elem_list = nullptr; - lmm_element_t elem = nullptr; - lmm_variable_t var = nullptr; lmm_constraint_t cnst = static_cast(param_cnst); double diff = 0.0; - double sigma_i = 0.0; XBT_IN(); - elem_list = &(cnst->enabled_element_set); XBT_CDEBUG(surf_lagrange_dichotomy, "Computing diff of cnst (%p)", cnst); + xbt_swag_t elem_list = &(cnst->enabled_element_set); + void* _elem; xbt_swag_foreach(_elem, elem_list) { - elem = static_cast(_elem); - var = elem->variable; - xbt_assert(var->weight > 0); + lmm_element_t elem = static_cast(_elem); + lmm_variable_t var = elem->variable; + xbt_assert(var->sharing_weight > 0); XBT_CDEBUG(surf_lagrange_dichotomy, "Computing sigma_i for var (%p)", var); // Initialize the summation variable - sigma_i = 0.0; + double sigma_i = 0.0; - // Compute sigma_i - for (j = 0; j < var->cnsts_number; j++) { - sigma_i += (var->cnsts[j].constraint)->lambda; + // Compute sigma_i + for (s_lmm_element_t const& elem : var->cnsts) { + sigma_i += elem.constraint->lambda; } //add mu_i if this flow has a RTT constraint associated @@ -476,9 +444,9 @@ static double partial_diff_lambda(double lambda, void *param_cnst) } /** \brief Attribute the value bound to var->bound. - * + * * \param func_fpi inverse of the partial differential of f (f prime inverse, (f')^{-1}) - * + * * Set default functions to the ones passed as parameters. This is a polymorphism in C pure, enjoy the roots of * programming. * @@ -500,24 +468,22 @@ void lmm_set_default_protocol_function(double (*func_f) (lmm_variable_t var, dou * Therefore: $fp(x) = \frac{\alpha D_f}{x}$ * Therefore: $fpi(x) = \frac{\alpha D_f}{x}$ */ -#define VEGAS_SCALING 1000.0 - double func_vegas_f(lmm_variable_t var, double x) { xbt_assert(x > 0.0, "Don't call me with stupid values! (%1.20f)", x); - return VEGAS_SCALING * var->weight * log(x); + return VEGAS_SCALING * var->sharing_weight * log(x); } double func_vegas_fp(lmm_variable_t var, double x) { xbt_assert(x > 0.0, "Don't call me with stupid values! (%1.20f)", x); - return VEGAS_SCALING * var->weight / x; + return VEGAS_SCALING * var->sharing_weight / x; } double func_vegas_fpi(lmm_variable_t var, double x) { xbt_assert(x > 0.0, "Don't call me with stupid values! (%1.20f)", x); - return var->weight / (x / VEGAS_SCALING); + return var->sharing_weight / (x / VEGAS_SCALING); } /* @@ -525,30 +491,29 @@ double func_vegas_fpi(lmm_variable_t var, double x) * Therefore: $fp(x) = \frac{3}{3 D_f^2 x^2+2}$ * Therefore: $fpi(x) = \sqrt{\frac{1}{{D_f}^2 x} - \frac{2}{3{D_f}^2}}$ */ -#define RENO_SCALING 1.0 double func_reno_f(lmm_variable_t var, double x) { - xbt_assert(var->weight > 0.0, "Don't call me with stupid values!"); + xbt_assert(var->sharing_weight > 0.0, "Don't call me with stupid values!"); - return RENO_SCALING * sqrt(3.0 / 2.0) / var->weight * atan(sqrt(3.0 / 2.0) * var->weight * x); + return RENO_SCALING * sqrt(3.0 / 2.0) / var->sharing_weight * atan(sqrt(3.0 / 2.0) * var->sharing_weight * x); } double func_reno_fp(lmm_variable_t var, double x) { - return RENO_SCALING * 3.0 / (3.0 * var->weight * var->weight * x * x + 2.0); + return RENO_SCALING * 3.0 / (3.0 * var->sharing_weight * var->sharing_weight * x * x + 2.0); } double func_reno_fpi(lmm_variable_t var, double x) { double res_fpi; - xbt_assert(var->weight > 0.0, "Don't call me with stupid values!"); + xbt_assert(var->sharing_weight > 0.0, "Don't call me with stupid values!"); xbt_assert(x > 0.0, "Don't call me with stupid values!"); - res_fpi = 1.0 / (var->weight * var->weight * (x / RENO_SCALING)) - 2.0 / (3.0 * var->weight * var->weight); + res_fpi = 1.0 / (var->sharing_weight * var->sharing_weight * (x / RENO_SCALING)) - + 2.0 / (3.0 * var->sharing_weight * var->sharing_weight); if (res_fpi <= 0.0) return 0.0; -/* xbt_assert(res_fpi>0.0,"Don't call me with stupid values!"); */ return sqrt(res_fpi); } @@ -557,22 +522,22 @@ double func_reno_fpi(lmm_variable_t var, double x) * Therefore: $fp(x) = 2/(Weight*x + 2) * Therefore: $fpi(x) = (2*Weight)/x - 4 */ -#define RENO2_SCALING 1.0 double func_reno2_f(lmm_variable_t var, double x) { - xbt_assert(var->weight > 0.0, "Don't call me with stupid values!"); - return RENO2_SCALING * (1.0 / var->weight) * log((x * var->weight) / (2.0 * x * var->weight + 3.0)); + xbt_assert(var->sharing_weight > 0.0, "Don't call me with stupid values!"); + return RENO2_SCALING * (1.0 / var->sharing_weight) * + log((x * var->sharing_weight) / (2.0 * x * var->sharing_weight + 3.0)); } double func_reno2_fp(lmm_variable_t var, double x) { - return RENO2_SCALING * 3.0 / (var->weight * x * (2.0 * var->weight * x + 3.0)); + return RENO2_SCALING * 3.0 / (var->sharing_weight * x * (2.0 * var->sharing_weight * x + 3.0)); } double func_reno2_fpi(lmm_variable_t var, double x) { xbt_assert(x > 0.0, "Don't call me with stupid values!"); - double tmp = x * var->weight * var->weight; + double tmp = x * var->sharing_weight * var->sharing_weight; double res_fpi = tmp * (9.0 * x + 24.0); if (res_fpi <= 0.0) @@ -581,3 +546,5 @@ double func_reno2_fpi(lmm_variable_t var, double x) res_fpi = RENO2_SCALING * (-3.0 * tmp + sqrt(res_fpi)) / (4.0 * tmp); return res_fpi; } +} +}