- PPL-devel - BUGSENG Mailman

[GIT] ppl/ppl(sparse_matrices): Merge remote branch 'origin/master' into sparse_matrices
by Marco Poletti 18 Sep '10

18 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: d5385797ea5f8c5c07c67d9aa8c07aa081f5c963 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=d5385797ea5f… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Sat Sep 18 14:48:22 2010 +0200 Merge remote branch 'origin/master' into sparse_matrices ---

1 0

[GIT] ppl/ppl(sparse_matrices): The Solaris sed bug affects all language interfaces (not just the Java one ).
by Enea Zaffanella 18 Sep '10

18 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: 7f6f8d36f5f925c7ef440823d0223a9fa1b72962 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=7f6f8d36f5f9… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Sat Sep 18 09:49:08 2010 +0200 The Solaris sed bug affects all language interfaces (not just the Java one). --- README.solaris | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/README.solaris b/README.solaris index e454198..9d4d024 100644 --- a/README.solaris +++ b/README.solaris @@ -1,5 +1,5 @@ Build problems have been reported for Solaris 10, affecting the build -of the Java language interface, which are due to the use of an +of the (C, Java, ...) language interfaces, which are due to the use of an incompatible version of sed: https://www.cs.unipr.it/mantis/view.php?id=103

1 0

[GIT] ppl/ppl(MPI): Revert "Distributed_Sparse_Matrix: print the number of pivot operations in the destructor."
by Marco Poletti 18 Sep '10

18 Sep '10

Module: ppl/ppl Branch: MPI Commit: c51907509e1e93f5d7972bb1414092a50cb3ff7e URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=c51907509e1e… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Sat Sep 18 12:21:33 2010 +0200 Revert "Distributed_Sparse_Matrix: print the number of pivot operations in the destructor." This reverts commit 845523ef94033ab908ea709b87c6bc511235f1af. --- src/Distributed_Sparse_Matrix.cc | 2 -- src/Distributed_Sparse_Matrix.defs.hh | 2 -- src/Distributed_Sparse_Matrix.inlines.hh | 5 ----- 3 files changed, 0 insertions(+), 9 deletions(-) diff --git a/src/Distributed_Sparse_Matrix.cc b/src/Distributed_Sparse_Matrix.cc index 2e6e99a..b3f0b7e 100644 --- a/src/Distributed_Sparse_Matrix.cc +++ b/src/Distributed_Sparse_Matrix.cc @@ -1873,8 +1873,6 @@ PPL::Distributed_Sparse_Matrix dimension_type& exiting_var_index, Sparse_Row& working_cost) { - ++pivot_count; - broadcast_operation(GET_EXITING_AND_PIVOT, id, entering_index); // 1. exiting_index(entering_index): diff --git a/src/Distributed_Sparse_Matrix.defs.hh b/src/Distributed_Sparse_Matrix.defs.hh index 59907bb..7396ee3 100644 --- a/src/Distributed_Sparse_Matrix.defs.hh +++ b/src/Distributed_Sparse_Matrix.defs.hh @@ -470,8 +470,6 @@ private: If base[i] is not 0, it is the column associated with that row. */ std::vector<dimension_type> base; - - dimension_type pivot_count; }; namespace std { diff --git a/src/Distributed_Sparse_Matrix.inlines.hh b/src/Distributed_Sparse_Matrix.inlines.hh index 12c5a23..95301a8 100644 --- a/src/Distributed_Sparse_Matrix.inlines.hh +++ b/src/Distributed_Sparse_Matrix.inlines.hh @@ -55,7 +55,6 @@ Distributed_Sparse_Matrix::Distributed_Sparse_Matrix() reverse_mapping(comm_size), next_rank(0), local_rows(0), base(0) { PPL_ASSERT(comm().rank() == 0); PPL_ASSERT(OK()); - pivot_count = 0; } inline @@ -65,7 +64,6 @@ Distributed_Sparse_Matrix PPL_ASSERT(comm().rank() == 0); init(num_rows1, num_cols1); - pivot_count = 0; } inline @@ -82,8 +80,6 @@ Distributed_Sparse_Matrix broadcast_operation(COPY_MATRIX_OPERATION, matrix.id, id); - pivot_count = 0; - PPL_ASSERT(OK()); } @@ -111,7 +107,6 @@ Distributed_Sparse_Matrix::~Distributed_Sparse_Matrix() { broadcast_operation(DELETE_MATRIX_OPERATION, id); // The rows stored at the root will be destroyed automatically. - std::cout << pivot_count << std::endl; } inline dimension_type

1 0

[GIT] ppl/ppl(master): The Solaris sed bug affects all language interfaces (not just the Java one).
by Enea Zaffanella 18 Sep '10

18 Sep '10

Module: ppl/ppl Branch: master Commit: 7f6f8d36f5f925c7ef440823d0223a9fa1b72962 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=7f6f8d36f5f9… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Sat Sep 18 09:49:08 2010 +0200 The Solaris sed bug affects all language interfaces (not just the Java one). --- README.solaris | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/README.solaris b/README.solaris index e454198..9d4d024 100644 --- a/README.solaris +++ b/README.solaris @@ -1,5 +1,5 @@ Build problems have been reported for Solaris 10, affecting the build -of the Java language interface, which are due to the use of an +of the (C, Java, ...) language interfaces, which are due to the use of an incompatible version of sed: https://www.cs.unipr.it/mantis/view.php?id=103

1 0

[GIT] ppl/ppl(MPI): New commits
by Roberto Bagnara 17 Sep '10

17 Sep '10

URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=237252ca6540… Merge: 176f106074935a5735288cb540ee813b54dde36c bc20af849dbe0aa805e4e0216a73a94c837b515f Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Fri Sep 17 21:34:02 2010 +0200 Merge remote branch 'origin/sparse_matrices' into MPI URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=bc20af849dbe… Merge: 90d9338c35f62fa89a86c3b7192b2fa8172d1b20 6cab0f23dfda89b788bf1f7cbb92b5fcbec9dc83 Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Fri Sep 17 21:00:57 2010 +0200 Merge remote branch 'origin/master' into sparse_matrices Conflicts: src/MIP_Problem.cc URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=6cab0f23dfda… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 17:42:19 2010 +0200 Do destroy the MIP object before returning from solve_with_simplex. Do destroy the Polyhedron object before returning from solve_with_generators. URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=f1b3b29529f9… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 16:35:35 2010 +0200 Prefer neg_assign to the use of unary minus on Coefficient objects. Use add_mul_assign even in code only activated during debugging. URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=719f973998fc… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 15:42:03 2010 +0200 Mention the correction of an efficiency bug in the C interface. URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=90d9338c35f6… Merge: ff67a6bed878e178880fcb16019ebe662e56ea51 d91c33b33dbc8b59332d279550366ed55ae9ae38 Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Thu Sep 16 21:08:52 2010 +0200 Merge remote branch 'origin/master' into sparse_matrices URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=d91c33b33dbc… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Thu Sep 16 21:08:27 2010 +0200 MIP_Problem: fix performance bug in compute_generator(). URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=ba9f6c92b49a… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Thu Sep 16 21:07:58 2010 +0200 ppl_c_implementation_common: fix performance bug in ppl_Linear_Expression_add_to_coefficient.

1 0

[GIT] ppl/ppl(MPI): Fix compiler errors when using non-distributed matrices .
by Marco Poletti 17 Sep '10

17 Sep '10

Module: ppl/ppl Branch: MPI Commit: 176f106074935a5735288cb540ee813b54dde36c URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=176f10607493… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Fri Sep 17 21:31:24 2010 +0200 Fix compiler errors when using non-distributed matrices. --- src/MIP_Problem.defs.hh | 7 +------ 1 files changed, 1 insertions(+), 6 deletions(-) diff --git a/src/MIP_Problem.defs.hh b/src/MIP_Problem.defs.hh index 4480603..5ab651d 100644 --- a/src/MIP_Problem.defs.hh +++ b/src/MIP_Problem.defs.hh @@ -27,11 +27,8 @@ site: http://www.cs.unipr.it/ppl/ . */ #include "globals.types.hh" #include "Row.defs.hh" -#ifdef USE_PPL_DISTRIBUTED_SPARSE_MATRIX #include "Distributed_Sparse_Matrix.defs.hh" -#else #include "Matrix.defs.hh" -#endif #include "Linear_Expression.defs.hh" #include "Constraint.types.hh" @@ -451,10 +448,8 @@ private: Matrix tableau; #endif - typedef Row working_cost_type; - //! The working cost function. - working_cost_type working_cost; + Row working_cost; //! A map between the variables of `input_cs' and `tableau'. /*!

1 0

[GIT] ppl/ppl(sparse_matrices): Merge remote branch 'origin/master' into sparse_matrices
by Marco Poletti 17 Sep '10

17 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: bc20af849dbe0aa805e4e0216a73a94c837b515f URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=bc20af849dbe… Author: Marco Poletti <poletti.marco(a)gmail.com> Date: Fri Sep 17 21:00:57 2010 +0200 Merge remote branch 'origin/master' into sparse_matrices Conflicts: src/MIP_Problem.cc ---

1 0

[GIT] ppl/ppl(sparse_matrices): Do destroy the MIP object before returning from solve_with_simplex.
by Enea Zaffanella 17 Sep '10

17 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: 6cab0f23dfda89b788bf1f7cbb92b5fcbec9dc83 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=6cab0f23dfda… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 17:42:19 2010 +0200 Do destroy the MIP object before returning from solve_with_simplex. Do destroy the Polyhedron object before returning from solve_with_generators. --- demos/ppl_lpsol/ppl_lpsol.c | 45 ++++++++++++++++++++++-------------------- 1 files changed, 24 insertions(+), 21 deletions(-) diff --git a/demos/ppl_lpsol/ppl_lpsol.c b/demos/ppl_lpsol/ppl_lpsol.c index abf9fbd..20157cd 100644 --- a/demos/ppl_lpsol/ppl_lpsol.c +++ b/demos/ppl_lpsol/ppl_lpsol.c @@ -773,10 +773,10 @@ solve_with_generators(ppl_Constraint_System_t ppl_cs, ppl_Coefficient_t optimum_d, ppl_Generator_t point) { ppl_Polyhedron_t ppl_ph; + int optimum_found = 0; int empty; int unbounded; int included; - int ok; /* Create the polyhedron (recycling the data structures of ppl_cs). */ ppl_new_C_Polyhedron_recycle_Constraint_System(&ppl_ph, ppl_cs); @@ -809,7 +809,7 @@ solve_with_generators(ppl_Constraint_System_t ppl_cs, if (verbosity >= 1) fprintf(output_file, "Unfeasible problem.\n"); maybe_check_results(PPL_MIP_PROBLEM_STATUS_UNFEASIBLE, 0.0); - return 0; + goto exit; } if (!empty && no_optimization) { @@ -818,7 +818,7 @@ solve_with_generators(ppl_Constraint_System_t ppl_cs, /* Kludge: let's pass PPL_MIP_PROBLEM_STATUS_OPTIMIZED, to let work `maybe_check_results'. */ maybe_check_results(PPL_MIP_PROBLEM_STATUS_OPTIMIZED, 0.0); - return 0; + goto exit; } /* Check whether the problem is unbounded. */ @@ -841,21 +841,16 @@ solve_with_generators(ppl_Constraint_System_t ppl_cs, if (verbosity >= 1) fprintf(output_file, "Unbounded problem.\n"); maybe_check_results(PPL_MIP_PROBLEM_STATUS_UNBOUNDED, 0.0); - return 0; + goto exit; } - ok = maximize + optimum_found = maximize ? ppl_Polyhedron_maximize_with_point(ppl_ph, ppl_objective_le, - optimum_n, optimum_d, &included, - point) + optimum_n, optimum_d, &included, + point) : ppl_Polyhedron_minimize_with_point(ppl_ph, ppl_objective_le, - optimum_n, optimum_d, &included, - point); - - if (!ok) - fatal("internal error"); - - ppl_delete_Polyhedron(ppl_ph); + optimum_n, optimum_d, &included, + point); #ifdef PPL_LPSOL_SUPPORTS_TIMINGS @@ -868,10 +863,15 @@ solve_with_generators(ppl_Constraint_System_t ppl_cs, #endif /* defined(PPL_LPSOL_SUPPORTS_TIMINGS) */ + if (!optimum_found) + fatal("internal error"); + if (!included) fatal("internal error"); - return 1; + exit: + ppl_delete_Polyhedron(ppl_ph); + return optimum_found; } static int @@ -881,6 +881,7 @@ solve_with_simplex(ppl_const_Constraint_System_t cs, ppl_Coefficient_t optimum_d, ppl_Generator_t point) { ppl_MIP_Problem_t ppl_mip; + int optimum_found = 0; int pricing = 0; int status = 0; int satisfiable = 0; @@ -967,7 +968,7 @@ solve_with_simplex(ppl_const_Constraint_System_t cs, if (verbosity >= 1) fprintf(output_file, "Unfeasible problem.\n"); maybe_check_results(status, 0.0); - return 0; + goto exit; } else if (no_optimization && satisfiable) { if (verbosity >= 1) @@ -975,25 +976,27 @@ solve_with_simplex(ppl_const_Constraint_System_t cs, /* Kludge: let's pass PPL_MIP_PROBLEM_STATUS_OPTIMIZED, to let work `maybe_check_results'. */ maybe_check_results(PPL_MIP_PROBLEM_STATUS_OPTIMIZED, 0.0); - return 0; + goto exit; } else if (status == PPL_MIP_PROBLEM_STATUS_UNBOUNDED) { if (verbosity >= 1) fprintf(output_file, "Unbounded problem.\n"); maybe_check_results(status, 0.0); - return 0; + goto exit; } else if (status == PPL_MIP_PROBLEM_STATUS_OPTIMIZED) { ppl_MIP_Problem_optimal_value(ppl_mip, optimum_n, optimum_d); ppl_MIP_Problem_optimizing_point(ppl_mip, &g); ppl_assign_Generator_from_Generator(point, g); - return 1; + optimum_found = 1; + goto exit; } else fatal("internal error"); - /* This is just to avoid a compiler warning. */ - return 0; + exit: + ppl_delete_MIP_Problem(ppl_mip); + return optimum_found; } static void

1 0

[GIT] ppl/ppl(sparse_matrices): Prefer neg_assign to the use of unary minus on Coefficient objects.
by Enea Zaffanella 17 Sep '10

17 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: f1b3b29529f9602a0ea7e36b7a5ae62518e125dc URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=f1b3b29529f9… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 16:35:35 2010 +0200 Prefer neg_assign to the use of unary minus on Coefficient objects. Use add_mul_assign even in code only activated during debugging. --- src/MIP_Problem.cc | 6 +++--- src/PIP_Tree.cc | 10 +++++----- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/src/MIP_Problem.cc b/src/MIP_Problem.cc index 33ead88..78e5ff1 100644 --- a/src/MIP_Problem.cc +++ b/src/MIP_Problem.cc @@ -1730,12 +1730,12 @@ PPL::MIP_Problem::compute_generator() const { Coefficient_traits::const_reference t_row_split_var = t_row.get(split_var); if (t_row_split_var > 0) { - split_num = -t_row.get(0); + neg_assign(split_num, t_row.get(0)); split_den = t_row_split_var; } else { split_num = t_row.get(0); - split_den = -t_row_split_var; + neg_assign(split_den, t_row_split_var); } // We compute the lcm to compute subsequently the difference // between the 2 variables. @@ -1808,7 +1808,7 @@ PPL::MIP_Problem::second_phase() { const dimension_type split_var = mapping[i].second; working_cost[original_var] = new_cost[i]; if (mapping[i].second != 0) - working_cost[split_var] = - new_cost[i]; + neg_assign(working_cost[split_var], new_cost[i]); } // Here the first phase problem succeeded with optimum value zero. // Express the old cost function in terms of the computed base. diff --git a/src/PIP_Tree.cc b/src/PIP_Tree.cc index a5af8f7..2e102d4 100644 --- a/src/PIP_Tree.cc +++ b/src/PIP_Tree.cc @@ -2691,7 +2691,7 @@ PIP_Solution_Node::solve(const PIP_Problem& pip, dimension_type j = 1; for (Variables_Set::const_iterator p = all_params.begin(), p_end = all_params.end(); p != p_end; ++p, ++j) - expr += t_test.get(j) * Variable(*p); + add_mul_assign(expr, t_test.get(j), Variable(*p)); using namespace IO_Operators; std::cerr << "Found mixed parameter sign row: " << best_i << ".\n" << "Solution depends on sign of parameter " @@ -3108,8 +3108,8 @@ PIP_Solution_Node::generate_cut(const dimension_type index, Linear_Expression expr1(ctx1.get(0)); Linear_Expression expr2(ctx2.get(0)); for (dimension_type j = 1; j <= num_params; ++j, ++p) { - expr1 += ctx1.get(j) * Variable(*p); - expr2 += ctx2.get(j) * Variable(*p); + add_mul_assign(expr1, ctx1.get(j), Variable(*p)); + add_mul_assign(expr2, ctx2.get(j), Variable(*p)); } std::cout << "Inserting into context: " << Constraint(expr1 >= 0) << " ; " @@ -3158,9 +3158,9 @@ PIP_Solution_Node::generate_cut(const dimension_type index, dimension_type si = 0; for (dimension_type j = 0; j < space_dimension; ++j) { if (parameters.count(j) == 1) - expr += cut_t.get(ti++) * Variable(j); + add_mul_assign(expr, cut_t.get(ti++), Variable(j)); else - expr += cut_s.get(si++) * Variable(j); + add_mul_assign(expr, cut_s.get(si++), Variable(j)); } std::cout << "Adding cut: " << Constraint(expr + cut_t.get(0) >= 0)

1 0

[GIT] ppl/ppl(sparse_matrices): Mention the correction of an efficiency bug in the C interface.
by Enea Zaffanella 17 Sep '10

17 Sep '10

Module: ppl/ppl Branch: sparse_matrices Commit: 719f973998fc26bafb723586da46eaf3b72de66c URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=719f973998fc… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Fri Sep 17 15:42:03 2010 +0200 Mention the correction of an efficiency bug in the C interface. --- NEWS | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/NEWS b/NEWS index 5ac4a8f..93de84b 100644 --- a/NEWS +++ b/NEWS @@ -27,6 +27,9 @@ o The input routine for PPL numeric datatypes has been extended to Bugfixes ======== +o Corrected an efficiency bug in the C language interface function + int ppl_Linear_Expression_add_to_coefficient() + o Corrected a bug affecting the input routine of ppl_lpsol, whereby the inhomogeneous term of the objective function was disregarded.

1 0