Sun Nov 08 12:00:00 2015 Andrew Makhorin * GLPK 4.57 (38:0:2) has been released. * src/simplex/spy*.* New, more efficient implementation of the dual simplex method was included in the package. * src/simplex/spxprim.c Incorrect directive #ifdef USE_AT was changed to #if USE_AT. * src/simplex/simplex.h New header for simplex-based drivers was added. * src/glpspx02.c Old implementation of the dual simplex method was removed from the package. * src/glpspx.h Old header for simplex-based drivers was removed. * src/glpapi06.c Call to spx_dual was changed to spy_dual. tol_piv = 1e-10 in glp_init_smcp was changed to 1e-9 (this new tolerance seems to be more adequate for new implementations of the primal and dual simplex methods). * src/glpk.h, src/glpapi09.c, src/glpios03.c Option iocp.sr_heur to enable/disable simple rounding heuristic was added. Thanks to Chris Matrakidis for suggestion. * src/env/env.h src/env/error.c API routine glp_at_error was added and documented. Thanks to Jeroen Demeyer for suggestion. * src/glpapi19.c Translation of MiniSat solver (src/minisat) from C++ to C is made by a non-skilled programmer who converts pointers to ints and vice versa, so MiniSat works on 32-bit platforms only. Fixing this portability issue would require rewriting the code, so currently to prevent crashing the API routine glp_minisat1 was changed just to report failure if 64-bit version of GLPK is used. Thanks to Heinrich Schuchardt for bug report. * doc/glpk.tex, doc/gmpl.tex Some material was added according to changes in GLPK API. Also some minor typos were corrected. Thanks to Anton Voropaev for typo report. * w32/makefiles, w64/makefiles Option /Zi was added to pass to MSVC compiler. examples/tsp/*.* An example application program TSPSOL was added. This program is intended to solve the Symmetric Traveling Salesman Problem. For more details see examples/tsp/README. Thu Oct 01 12:00:00 2015 Andrew Makhorin * GLPK 4.56 (37:3:1) has been released. * src/simplex/*.c, *.h New, more efficient and robust implementation of the primal simplex method was included in the package. * src/glpspx01.c Old implementation of the primal simplex method was removed from the package. * src/bflib/sgf.c A bug was fixed in routine sgf_reduce_nuc. (The bug appeared if the basis matrix was structurally singular.) Thanks to Martin Jacob for bug report. * w32/*.*, w64/*.* Scripts to build GLPK with Microsoft Visual Studio 2015 were added. Thanks to Xypron for contribution and testing. Fri Aug 22 12:00:00 2014 Andrew Makhorin * GLPK 4.55 (37:2:1) has been released. * src/bflib/luf.c, src/bflib/luf.h Two routines luf_vt_solve1 and luf_estimate_norm were added to estimate 1-norm of matrix inverse. * src/bflib/btf.c, src/bflib/btf.h Two routines btf_at_solve1 and btf_estimate_norm were added to estimate 1-norm of block triangular matrix inverse. * src/bflib/fhvint.c, src/bflib/fhvint.h The routine fhvint_estimate was added to estimate 1-norm of the basis matrix inverse (FHV-factorization). * src/bflib/scfint.c, src/bflib/scfint.h The routine scfint_estimate was added to estimate 1-norm of the basis matrix inverse (Schur-complement-based factorization). * src/bfd.c, src/bfd.c The routine bfd_condest was added to estimate the condition of the basis matrix. * src/bflib/scfint.h The prefix '_glp' was added to keep the namespace clean. * src/env/stream.c Two open modes "a" and "ab" were added. Thanks to Pedro P. Wong for bug report. * src/glpapi21.c (glpsol) Minor bug was fixed (command-line options --btf, --cbg, --cgr didn't work properly). * src/bflib/sgf.c A serious bug was fixed in a basis factorization routine used on the dense phase. (The bug might appear only if the number of rows exceeded sqrt(2**31) ~= 46,340 and caused access violtaion exception because of integer overflow.) Thanks to Mark Meketon for bug report. * doc/glpk.tex Two API routines glp_alloc and glp_realloc were documented. Thanks to Brian Gladman for suggestion. * doc/gmpl_es.tex, doc/gmpl_es.pdf Translation of the document "Modeling Language GNU MathProg" to Spanish was included (in LaTeX and pdf formats). Thanks to Pablo Yapura for contribution. Fri Mar 28 12:00:00 2014 Andrew Makhorin * GLPK 4.54 (37:1:1) has been released. * src/bflib/fhvint.h, fhvint.c Some technical changes were made. * src/bflib/btf.h, btf.c, btfint.h, btfint.c The module BTF was added to the package. It implements sparse block triangular factorization of a square non-singular matrix and is based on LU-factorization (see the LUF module). * src/bflib/scf.h, scf.c, scfint.h, scfint.c The module SCF was added to the package. It implements sparse factorization of a square non-singular matrix based on Schur complement and LU-factorization (plain or block triangular; see the LUF and BTF modules). * src/bfd.c LP basis factorization driver was changed to use most recent versions of the FHV and SCF modules. * src/glplpf.h, glplpf.c Old version of the Schur-complement-based factorization module was removed from the package. * src/glpk.h New flags GLP_BF_LUF, GLP_BF_BTF were added for glp_bfcp. * src/glpapi12.c The API routines glp_get_bfcp, glp_set_bfcp were replaced due to new version of the BFD module. * src/glpapi21.c New command-line options --luf, --btf, --ft, --cbg, and --cgr were added for the glpsol solver. * src/* According to a new version of the GNU Coding Standards in all messages a grave accent character (`) was replaced by a single apostrophe character ('). * src/glpapi09.c mip status bug was fixed. Thanks to Remy Roy for bug report. * doc/glpk.tex Some comments about invalidating the basis factorization were added. Thanks to Xypron for suggestion. * configure.ac "iodbc-config --cflags" was added to detect iodbc flags. Thanks to Sebastien Villemot for patch. Thu Feb 13 12:00:00 2014 Andrew Makhorin * GLPK 4.53 (37:0:1) has been released. * src/glpmps.c (glp_read_mps) The code was changed to remove free rows at the end. * src/glpcpx.c (glp_read_lp) A bug was fixed (explicit bounds for binaries not set). Thanks to Gabriel Hackebeil for bug report. * src/glpenv07.c (z_fgetc) A bug was fixed (Z_STREAM_END -> Z_OK). Thanks to Achim Gaedke for bug report. * src/glpenv07.c Replaced by src/env/stream.c. * src/glpenv08.c Replaced by src/env/dlsup.c. * src/bflib/ifu.h, src/bflib/ifu.c Re-implemented. * src/glpscf.h, src/glpscf.c Replaced by IFU. * src/glplpf.h, src/glplpf.c Changed due to IFU. * src/glpbfd.c Changed due to LPF. * src/glpapi06.c Two API routines glp_get_it_cnt and glp_set_it_cnt were added. Thanks to Joey Rios for suggestion. * src/glplpx.h, src/glplpx01.c, src/glplpx02.c, src/glplpx03.c All obsolete API routines were completely removed. * examples/oldapi/lpx.h, examples/oldapi/lpx.c A set of routines that simulate the old GLPK API (as defined in 4.48) were added. Thanks to Jan Engelhardt for suggestion. * src/zlib/* zlib 1.2.7 was downgraded to zlib 1.2.5 (from glpk 4.50) due to bugs detected in zlib 1.2.7 on some 64-bit platforms. Thanks to Carlo Baldassi for bug report. * src/glpsql.c Alignment bug was fixed. Thanks to Xypron for suggestion. * src/glpsql.c #include and #include were commented out to fix a namespace bug on compiling with MariaDB. Thanks to Xypron for suggestion. Sun Jul 28 12:00:00 2013 Andrew Makhorin * GLPK 4.52.1 (36:1:0) has been released. * src/Makefile.am Version information bug was fixed. Thanks to Sebastien Villemot for bug report. * src/proxy/proxy.c A minor bug (incorrect use of glp_term_out) was fixed. * src/glpios03.c The simple rounding heuristic routine was changed to check only global constraints. * src/glpcpx.c The code was changed to issue the warning message 'lower/upper bound redefined' only once. Thu Jul 18 12:00:00 2013 Andrew Makhorin * GLPK 4.52 (36:0:1) has been released. * src/misc/wclique1.h, src/misc/wclique1.c Greedy heuristic to find maximum weight clique was implemented. * src/cglib/cfg.h, src/cglib/cfg.c, src/cglib/cfg1.c Conflict graph routines (used to generate clique cuts) were implemented. * src/glpios08.c New version of the clique cut generator was implemented. Now it is able to efficiently process very large and/or dense conflict graphs. Old rudimentary version was removed from the package. * examples/misp1.dat, examples/misp2.dat Two data files for the maximum independent set problem (MISP) were added to illustrate efficiency of using the clique cuts. * src/glpios03.c Simple rounding heuristic applied for every node subproblem was implemented. * proxy/proxy.c Some bugs were fixed in the proximity search heuristic routine. Thanks to Giorgio Sartor <0gioker0@gmail.com>. * src/glpapi21.c New command-line option '--proxy [nnn]' was added to glpsol to enable using the proximity search heuristic. * src/glpspx02.c A feature was added to switch to the primal simplex in case of stalling due to dual degeneracy (for GLP_DUALP only). * src/glpmps.c A bug (incorrect processing of LI column indicator) was fixed in the mps format reading routine. Thanks to Charles Brixko for bug report. Wed Jun 19 12:00:00 2013 Andrew Makhorin * GLPK 4.51 (36:0:0) has been released. * src/bflib/sgf.c, src/bflib/sgf.h Singleton and dense phases of sparse Gaussian factorizer were implemented. * src/bflib/lufint.c, src/bflib/lufint.h Interface routines to LU-factorization were added. * src/bflib/fhvint.c, src/bflib/fhvint.h Interface routines to FHV-factorization were changed to use lufint.c routines. * src/glplpf.c, src/glplpf.h Routines implementing Schur-complement version of the LP basis factorization were changed to use lufint.c routines. * src/glpbfd.c, src/glpbfd.h Interface routines to the LP basis factorization were changed due to changes in glplpf.c routines. * src/glpluf.c, src/glpluf.c Old version of LU-factorization was removed from the package. (This old version was used for 10 years since 3.0.6.) * src/misc/triang.c, src/misc/triang.h Routine to find maximal triangular part of rectangular matrix was added. * src/glpini01.c The API routine glp_adv_basis that constructs advanced initial LP basis was replaced by an improved version, which (unlike the old version) takes into account numerical values of constraint coefficients. * src/proxy/* Routines that implement the proximity search heuristic for MIP were added. Thanks to Giorgio Sartor <0gioker0@gmail.com> for contribution. * src/glpk.h iocp.ps_heur was added to enable/disable proxy heuristic. * glpsol --proxy command-line option was added. * src/zlib/*.* zlib general purpose compression library, version 1.2.7, was ANSIfied and modified according to GLPK requirements and included in the distribution as an external software module. This version replaced the old one (1.2.5). For details please see src/zlib/README. * src/glpk.h, src/env/time.c The API routine glp_time was changed to return double rather than glp_long. * src/glplib02.c Routines that implement glp_long type (64-bit arithmetic) were removed from the package. * src/glpk.h, src/env/alloc.c New API routine glp_alloc was added. It makes obsolete two API routines glp_malloc and glp_calloc which were replaced by macros to use glp_alloc (hence 36:0:0). * src/glpios10.c A bug was fixed that caused numerical instability in the FPUMP heuristic (the bug was introduced in glpk 4.40). Fri May 24 12:00:00 2013 Andrew Makhorin * GLPK 4.50 (35:0:0) has been released * src/Makefile.am Though API was not changed since 4.49, symbols _glp_lpx_* were removed from the export list, hence 35:0:0. * src/glpfhv.h, src/glpfhv.c Old routines for FHV-factorization were removed. * src/bflib/*.h, src/bflib/*.c New version of basis factorization routines, including routines for FHV-factorization, were added. * src/glpbfd.c LP basis factorization driver was changed according to the new routines for FHV-factorization. * doc/glpk.tex Some clarifications about using the name index routines were added. Thanks to Xypron for suggestion. * doc/gmpl.tex Some typos were corrected. Thanks to Jeffrey Kantor for report. * src/glprlx.c A serious bug was *tentatively* fixed. This bug is inherited from the original Fortran version of the RELAX-IV code. Unfortunately, the code is very intricate, so this bug is still under investigation. Thanks to Sylvain Fournier for bug report. RELAX-IV bug details -------------------- In the original RELAX-IV code there are four similar fragments in subroutines ascnt1 and ascnt2 like this: C C DECREASE THE PRICES OF THE SCANNED NODES BY DELPRC. C ADJUST FLOW TO MAINTAIN COMPLEMENTARY SLACKNESS WITH C THE PRICES. C NB = 0 DO 6 I=1,NSAVE . . . IF (RC(ARC).EQ.0) THEN DELX=DELX+U(ARC) NB = NB + 1 PRDCSR(NB) = ARC END IF . . . On some instances the variable NB becomes greater than N (the number of nodes) that leads to indexing error, because the array PRDCSR is declared as array of N elements (more precisely, as array of MAXNN elements, however, NB becomes even much greater than MAXNN). Tue Apr 16 12:00:00 2013 Andrew Makhorin * GLPK 4.49 (34:0:0) has been released * glprlx.c, glprlx.h C version of the Fortran code RELAX-IV (relaxation method of Bertsekas and Tseng) was added. * glpapi17.c API routine glp_mincost_relax4, which is a driver to RELAX-IV code, was added and documented. * glpnet03.c API routine glp_netgen_prob (Klingman's standard network problems) was added and documented. * glpapi12.c A bug (wrong dual feasibility test) was fixed in API routine glp_warm_up. Thanks to David T. Price for bug report. * glpapi10.c Obsolete API routine lpx_check_kkt was replaced by API routine glp_check_kkt. * glpk.h All old API routines whose names begin with 'lpx_' were removed from API and no more available. * glpk.tex, graphs.tex These documents were essentially reformatted. Mon Jan 28 12:00:00 2013 Andrew Makhorin * GLPK 4.48 (33:0:0) has been released * src/glpmps.c A bug in glp_write_mps was fixed. Thanks to Raniere Gaia Costa da Silva for bug report. * glpk.h glp_prob declaration changed (now it is incomplete struct); glp_tree declaration changed (now it is incomplete struct); glp_tran declaration changed (now it is incomplete struct); glp_long declaration removed; glp_time declaration removed; glp_difftime removed from API; glp_data removed from API; glp_sdf_* removed from API; glp_mem_usage declaration changed (glp_long -> size_t); glp_realloc declaration added (not documented yet). * glpenv.h, glpenv05.c Dynamic memory allocation routines were reimplemented. * glpk.h, glpnet03.c Routine glp_netgen_prob was added (not documented yet). * configure.ac Check for gmp.h was added. Thanks to Heinrich Schuchardt for suggestion. * w32/glpk.def, w64/glpk.def Changes were made to export only API symbols. Fri Sep 09 12:00:00 2011 Andrew Makhorin * GLPK 4.47 (32:0:32) has been released * src/glpapi20.c New API routine glp_infeas1 to solve 0-1 feasibility problem was added and documented (see doc/cnfsat.pdf). * src/glpnpp06.c Some new preprocessor routines for SAT-CNF encoding, which are used by the routine glp_intfeas1, were added. * src/glpnpp.h The header was modified due to additions. * src/glpapi21.c The glpsol solver was modified to bypass model postprocessing if the solution reported is neither optimal nor feasible. * src/glpapi21.c New glpsol options (--minisat and --objbnd) were added. * examples/pbn/*.* The paint-by-numbers puzzle model (pbn.mod) was modified to print solution in PostScript format and to check for multiple solutions. Some benchmark examples from encoded in MathProg were included in the distribution. For more details see examples/pbn/README and examples/pbn/pbn.pdf. * examples/Makefile.am A minor bug was fixed to correctly build glpk in a separate directory. Thanks to Marco Atzeri for bug report. Tue Aug 09 12:00:00 2011 Andrew Makhorin * GLPK 4.46 (31:0:31) has been released * src/glpk.h, src/Makefile glpk.h was relocated from 'include' to 'src', and 'include' subdir was removed; src/Makefile.am was changed appropriately. * src/zlib/*.* zlib general purpose compression library, version 1.2.5, was ANSIfied and modified according to GLPK requirements and included in the distribution as an external software module. For details see src/zlib/README. * src/glpdmx.c The following new API routines were added: glp_read_cnfsat - read CNF-SAT problem data in DIMACS format glp_check_cnfsat - check for CNF-SAT problem instance glp_write_cnfsat - write CNF-SAT problem data in DIMACS format * src/minisat/*.* MiniSat, a CNF-SAT solver, version 1.14.1, was ANSIfied and modified according to GLPK requirements and included in the distribution as an external software module. For details see minisat/README and minisat/LICENSE. * src/glpapi19.c The API routine glp_minisat1, which is a driver to the MiniSat solver, was included in the package. * doc/satcnf.* The document "CNF Satisfiability Problem" was included in the package. It is a supplement to the GLPK Reference Manual. * src/glpapi20.c New glpsol options (--cnf, --wcnf, and --minisat) was added. * glpsql.c Some bugs were fixed. Thanks to Xypron . Sun Dec 05 12:00:00 2010 Andrew Makhorin * GLPK 4.45 (30:0:30) has been released * glplpx01.c A bug (it_cnt) in routine reset_parms was fixed. Thanks to Ali Baharev for report. * glpmpl03.c A bug (print "text\") was fixed. Thanks to Xypron for report. * glpsql.c A precision bug was fixed. Thanks to Xypron . * glpk.tex Some typos were corrected. Thanks to Robbie Morrison . Thu Jun 03 12:00:00 2010 Andrew Makhorin * GLPK 4.44 (29:0:29) has been released * glpapi14.c glpmpl.h glpmpl01.c glpmpl03.c glpmpl04.c Implemented suffixes for variables and constraints. * glpmpl06.c Made changes to allow comment records in CSV files. * glpapi17.c Added and documented new API routine glp_cpp to solve Critical Path Problem. Sat Feb 20 12:00:00 2010 Andrew Makhorin * GLPK 4.43 (28:0:28) has been released * glplib.h, glplib.c, glpenv.h, glpenv.c The module glpenv was split into two modules glpenv and glplib. * glpenv01.c, glpenv03.c, glpenv04.c, glpenv06.c The following new API routines were added and documented: glp_init_env, glp_free_env, glp_open_tee, glp_close_tee, glp_error (macro), glp_difftime. * glpapi16.c New API routine glp_top_sort (topological sorting of ayclic digraph) was added and documented. * glpapi17.c A serious bug was fixed in the routine glp_asn_prob_hall. * glpnpp05.c A bug was fixed in the LP/MIP preprocessor (hidden covering inequalities). * glpsql.c Some improvements were made in the table driver (NULL data). Thanks to Xypron for contribution. * configure.ac Changes were made to use .dylib rather than .so under Mac OS. Thanks to Noli Sicad for testing Wed Jan 13 12:00:00 2010 Andrew Makhorin * GLPK 4.42 (27:0:27) has been released * glpapi01.c, glpapi11.c, glpapi12.c, glpdmx.c The following new API routines were added and documented: glp_check_dup (check for duplicate elements in sparse matrix); glp_sort_matrix (sort elements of the constraint matrix); glp_read_prob (read problem data in GLPK format); glp_write_prob (write problem data in GLPK format); glp_analyze_bound (analyze active bound of non-basic variable); glp_analyze_coef (analyze obj. coefficient at basic variable); glp_print_ranges (print sensitivity analysis report; replaces lpx_print_sens_bnds). * glpapi20.c New command-line options were added to glpsol: --glp (read problem data in GLPK format); --wglp (write problem data in GLPK format); --lp (replaces --cpxlp); --wlp (replaces --wcpxlp); --ranges (print sensitivity analysis report). * glpapi06.c In the routine glp_init_smcp default value of the parameter out_frq was changed to 500 (it was 200). * glpipp.h, glpipp01.c, glpipp02.c The old MIP preprocessor module was removed. * glpapi09.c Now the MIP solver uses the new MIP preprocessor (NPP). * glplpx03.c lpx_write_opb was disabled due to replacing IPP with NPP. * glpnet09.c Kellerman's heuristic to cover edges by cliques was added. * glplib08.c Recognition of special filenames "/dev/stdin", "/dev/stdout", and "/dev/stderr" was added. * glpk.tex Chapter "Graph and network routines" was carried out from the reference manual as a separate document. Mon Dec 21 12:00:00 2009 Andrew Makhorin * GLPK 4.41 (26:0:26) has been released * glpapi12.c The following new API routines were added: glp_transform_row (replaces lpx_transform_row); glp_transform_col (replaces lpx_transform_col); glp_prim_rtest (replaces lpx_prim_ratio_test); glp_dual_rtest (replaces lpx_dual_ratio_test). Note that values returned by glp_prim_rtest and glp_dual_rtest differ from the ones retutned by the deprecated routines. * glpnpp*.* The LP/MIP preprocessor was essentially re-implemented. * glpios03.c The feature to remove inactive cuts from the active subproblem was implemented. * glpios11.c The feature processing cuts stored in the cut pool was improved (now it uses estimation of objective degradation). * glpscg.* Obsolete implemetation of the conflict graph was removed. * glpmpl.h, glpmpl03.c, glpmpl04.c FILE was replaced by XFILE to allow using GLPK I/O routines. * glpsql.c, examples/sql, doc/tables.tex The SQL table driver was changed to allow multiple arguments separated by semicolon in SQL statements. Thanks to Xypron . * glpk.h, glpapi14.c New API routine glp_time was added (not documented yet). * glpapi20.c Two new options were added to glpsol: --seed value (initialize pseudo-random number generator used in MathProg model with specified seed value), and --ini filename (use as initial basis previously saved with -w option). * examples/xyacfs.mod Thanks to Nigel Galloway for contribution. * examples/dbf/*.* Thanks to Noli Sicad for contribution. * w32/*.*, w64/*.* Scripts to build GLPK with Microsoft Visual Studio 2010 were added. Thanks to Xypron for contribution and testing. Tue Nov 03 12:00:00 2009 Andrew Makhorin * GLPK 4.40 (25:0:25) has been released * glpdmx.c Two new API routines were added: glp_read_ccdata (read graph in DIMACS clique/coloring format); glp_write_ccdata (write graph in DIMACS clique/coloring format). Also an example file examples/sample.col was added. * glpapi19.c, glpnet08.c New API routine glp_wclique_exact was added. It is intended to find a maximum weight clique with the exact algorithm developed by Prof. P. Ostergard. * glpnpp02.c A bug was fixed in the LP preprocessor (routine npp_empty_col). Thanks to Stefan Vigerske for the bug report. * glpios10.c A bug was fixed and some improvements were made in the FPUMP heuristic module. Thanks to Xypron . * glpapi12.c A bug was fixed in the API routine glp_warm_up (dual feasibility test was incorrect in maximization case). Thanks to Uday Venkatadri for the bug report. * glpapi16.c Two new API routines were added: glp_del_vertices (remove vertices from graph); glp_del_arc (remove arc from graph). * glpios09.c The hybrid pseudocost branching heuristic was included in the MIP solver. It is available on API level (iocp.br_tech should be set to GLP_BR_PCH) and in the stand-alone solver glpsol (via the command-line option --pcost). This heuristic may be useful on solving hard MIP instances. * glpios03.c The branching heuristic by Driebeck and Tomlin (used in the MIP solver by default) was changed to switch to branching on most fractional variable if an lower bound of degradation of the objective is close to zero for all branching candidates. Sun Jul 26 12:00:00 2009 Andrew Makhorin * GLPK 4.39 (24:0:24) has been released * glpsdf.c New API routines to read plain data files were added. * glpcpx.h, glpini.h, glpscl.h These headers were removed. * glpcpx.c API routines glp_read_lp and glp_write_lp to read/write files in CPLEX LP format were re-implemented. Now glp_write_lp correctly writes double-bounded (ranged) rows by introducing slack variables rather than by duplicating the rows. The data structure glp_cpxcp and routine glp_init_cpxcp were added. * amd/* The 'xfree(NULL)' bug was fixed in the AMD routines. Thanks to Niels Klitgord for the bug report. * glpapi16.c New API routines glp_set_vertex_name, glp_create_v_index, glp_find_vertex, and glp_delete_v_index were added. * glpdmx.c New API routines glp_read_asnprob, glp_write_asnprob, glp_read_ccformat, and glp_write_ccformat were added (the two latter routines are not documented yet). * glpapi18.c New API routines glp_check_asnprob, glp_asnprob_lp, glp_asnprob_okalg, and glp_asnprob_hall were added. * glpini01.c, glpini02.c The message "Crashing..." was changed to "Constructing initial basis..." due to suggestion by Thomas Kahle . * glpapi14.c New API routines glp_printf, glp_vprintf, glp_malloc, glp_calloc, glp_free, and glp_assert were added. * glplpp.h, glplpp01.c, glplpp02.c Old LP presolver routines were removed. Now glp_simplex uses new preprocessing routines (see glpnpp). * glpapi12.c New API routine glp_warm_up was added; it replaces the routine lpx_warm_up. Sat May 02 12:00:00 2009 Andrew Makhorin * GLPK 4.38 (23:0:23) has been released * glpmps.c API routines to read/write MPS files were re-implemented. * glpspx02.c Some improvements were made in the dual simplex routine. * glpk.h New structure glp_iptcp was added. * glpnpp.h, glpnpp01.c, glpnpp02.c New LP/MIP preprocessor. Currently it includes only some basic routines and used only in the interior-point solver. * glpapi08.c API routine glp_interior was replaced by an improved version (new LP/MIP preprocessor, new ordering algorithms). New API routine glp_init_iptcp was added. API routine glp_ipt_status may return two new statuses due to changes in glp_interior. * glpsol.c New command-line options were added (ordering algorithm used in the interior-point solver). * amd/*.*, colamd/*.* Two external software modules AMD and COLAMD/SYMAMD used in the interior-point solver were included in the distribution. For details see amd/README and colamd/README. * glpnet03.c, glpnet04.c, glpnet05.c A minor bug was fixed (_G => G_). Thanks to Nelson H. F. Beebe for bug report. Sun Mar 29 12:00:00 2009 Andrew Makhorin * GLPK 4.37 (22:0:22) has been released * glpk.h iocp.fp_heur was added to enable/disable fpump heuristic. * glpios10.c ios_feas_pump was added (feasibility pump heuristic). * glpsol.c --fpump command-line option was added. * glpsds.c Plain data set routines were added to facilitate reading plain data in application programs. Currently these routines are not in API, though declared in glpk.h. * glpapi08.c A bug was fixed in the internal routine restore. (Due to this bug dual solution components were computed incorrectly if the problem was scaled.) * glpapi10.c, glpapi11.c The following new API routines were added: glp_print_sol (replaces lpx_print_sol); glp_print_ipt (replaces lpx_print_ips); glp_print_mip (replaces lpx_print_mip); _glp_check_kkt (replaces lpx_check_kkt, lpx_check_int). Now the routine lpx_print_prob (deprecated) is equivalent to the routine glp_write_lp. * glpapi18.c, glpapi19.c The following new API routines were added: glp_read_graph (read graph from plain text file); glp_write_graph (write graph to plain text file); glp_weak_comp (find all weakly connected components); glp_strong_comp (find all strongly connected components). * configure.ac, Makefile.am Changes were made: (a) to allow using autoreconf/autoheader; (b) to allow building glpk in a directory other than its source directory. Thanks to Marco Atzeri for bug report. * examples/shiftcover.mod An example model in MathProg was added. Thanks to Larry D'Agostino for contribution. Fri Feb 06 12:00:00 2009 Andrew Makhorin * GLPK 4.36 (21:0:21) has been released * glpnet06.c, glpnet07.c, glpapi19.c The following new API routines were added: glp_mincost_okalg find minimum-cost flow with out-of-kilter algorithm glp_maxflow_ffalg find maximal flow with Ford-Fulkerson algorithm * glpsol.c Two new command-line options were added: --mincost read min-cost flow data in DIMACS format --maxflow read maximum flow data in DIMACS format * doc/glpk.* New edition of the reference manual was included. * glpk.h Duplicate symbols were removed to allow using swig. Thanks to Kelly Westbrooks and Nigel Galloway for suggestion. * glpcpx.c A minor defect was fixed in the routine glp_write_lp. Thanks to Sebastien Briais for bug report. * glpsql.c A minor bug was fixed. Thanks to Xypron for patch. * examples/hashi.mod, examples/shikaku.mod Two example models in MathProg were added. Thanks to Sebastian Nowozin for contribution. * examples/qfit.mod, examples/yacfs.mod Two example models in MathProg were added. Thanks to Nigel Galloway for contribution. Fri Jan 09 12:00:00 2009 Andrew Makhorin * GLPK 4.35 (20:0:20) has been released * glpk.h, glpapi.c, glpnet.c The following new API routines were added: glp_create_graph create graph glp_set_graph_name assign (change) graph name glp_add_vertices add new vertices to graph glp_add_arc add new arc to graph glp_erase_graph erase graph content glp_delete_graph delete graph glp_read_mincost read minimum cost flow problem data in DIMACS format glp_write_mincost write minimum cost flow problem data in DIMACS format glp_mincost_lp convert minimum cost flow problem to LP glp_netgen Klingman's network problem generator glp_gridgen grid-like network problem generator glp_read_maxflow read maximum flow problem data in DIMACS format glp_write_maxflow write maximum flow problem data in DIMACS format glp_maxflow_lp convert maximum flow problem to LP glp_rmfgen Goldfarb's maximum flow problem generator * doc/glpk.* New edition of the reference manual was included. * examples/sample.min, examples/sample.max Two example data files in DIMACS format were added. * glplib04.c The statement "if (c = '\n') fflush(stdout)" was added to the internal routine xputc to provide "real-time" terminal output. Thanks to Luiz Bettoni for suggestion. * glpmpl05.c A minor bug was fixed in the internal routine mpl_fn_time2str. Thanks to Stefan Vigerske for bug report. * w32/makefile, w64/makefile The flag -O2 (/O2) was added to some makefiles. Thu Dec 04 12:00:00 2008 Andrew Makhorin * GLPK 4.34 (19:0:19) has been released * src/glpios03.c A bug was fixed in the internal routine branch_on. Thanks to Nigel Galloway for bug report. * src/glpmpl05.c Three new MathProg functions were included: gmtime obtaining current calendar time str2time converting character string to calendar time time2str converting calendar time to character string Thanks to Xypron . * doc/glpk.*, doc/gmpl.* A new edition of the GLPK reference manual and GNU MathProg language description were included. Thu Oct 30 12:00:00 2008 Andrew Makhorin * GLPK 4.33 (18:0:18) has been released * glpapi*.* The following new API routines were added: glp_copy_prob copy problem object content glp_exact solve LP in exact arithmetic (makes lpx_exact deprecated) glp_get_unbnd_ray determine variable causing unboundedness (makes lpx_get_ray_info deprecated) glp_interior solve LP with interior-point method (makes lpx_interior deprecated) * glpapi*.* The following new API routines for processing models written in the GNU Mathprog language were added to the package: glp_mpl_alloc_wksp allocate the translator workspace glp_mpl_read_model read and translate model section glp_mpl_read_data read and translate data section glp_mpl_generate generate the model glp_mpl_build_prob build LP/MIP instance from the model glp_mpl_postsolve postsolve the model glp_mpl_free_wksp deallocate the translator workspace (These routines make lpx_read_model deprecated.) * src/glpapi17.c, examples/glpsol.c The stand-alone solver glpsol was re-implemented with new API routines. * src/glpsql.c Some bugs were fixed in the SQL table driver. Thanks to Xypron . * examples/cplex/*.* A crude implementation of CPLEX-like interface to GLPK API was added to the package. See examples/cplex/README. Fri Oct 03 12:00:00 2008 Andrew Makhorin * GLPK 4.32 (17:0:17) has been released * glpmpl01.c A bug was fixed. Due to this bug iterated expressions having an indexing expression whose dummy indices are bound to some values, i.e. like sum{(i+1,j,k-1) in E} x[i,j,k] are evaluated incorrectly. Namely, current value of such expressions is not invalidated when corresponding dummy indices (like i and k in the example above) are changed, that erroneously results in the same value evaluated for the first time. * glpios03.c Euclidean reduction of the local objective bound was added in the routine glpios03.c. * glpapi11.c The following new branch-and-cut API routines were added: glp_ios_row_attr determine additional row attributes; glp_ios_pool_size determine current size of the cut pool; glp_ios_add_row add constraint to the cut pool; glp_ios_del_row delete constraint from the cut pool; glp_ios_clear_pool delete all constraints from the cut pool. * glpapi08.c The following new features were included in the branch-and-cut solver (the API routine glp_intopt): MIP presolver; mixed cover cut generator; clique cut generator. Due to the MIP presolver glp_intopt may additionally return GLP_ENOPFS and GLP_ENODFS, if primal or dual infeasibility of LP relaxation is detected by the presolver. Also the return code GLP_EMIPGAP was introduced to correctly indicate that the mip gap tolerance is reached. * glplpx01.c Now the obsolete API routines lpx_integer and lpx_intopt are completely superseded by the API routine glp_intopt that makes them deprecated. * glpmpl05.c Now the table driver name "iODBC" can be specified as "ODBC". * glpmpl03.c A bug fixed in the routine free_dca. Thanks to Xypron . * glpsql.c A bug was fixed in the SQL table driver. Thanks to Xypron . * examples/glpsol.c Changes were made to allow multiple MathProg data files. * doc/glpk.* A new edition of the GLPK reference manual was included. * doc/tables.* A new edition of the supplement "Using data tables in the GNU MathProg language" was included. Tue Sep 02 12:00:00 2008 Andrew Makhorin * GLPK 4.31 (16:0:16) has been released * glpspx.h, glpspx01.c, glpspx02.c, glpapi06.c The dual simplex solver (spx_dual_opt) was replaced by a new implementation of the two-phase dual simplex method (spx_dual). Old simplex method routines (spx_prim_opt, spx_prim_feas, and spx_dual_opt) were removed from the package. * glpk.h, glpscl.h, glpscl.c, glpapi04.c New API routine glp_scale_prob was added. It replaces routine lpx_scale_prob which is deprecated. * glpk.h, glpini.h, glpini01.c, glpini02.c, glpapi05.c New API routines glp_std_basis, glp_adv_basis, glp_cpx_basis were added. They replace routines lpx_std_basis, lpx_adv_basis, lpx_cpx_basis which are deprecated. * glpdmp.c 8-byte data alignment was added to the module (sufficient for both ILP32 and LP64 environments). * glplib07.c 16-byte data alignment was added to the module to provide compatibility with LP64 environment (8-byte is not sufficient due to jmp_buf; thanks to Xypron for investigation). * glplpx16.c New version of the routine lpx_write_pb was added. Thanks to Oscar Gustafsson for the contribution. * w32/VC9, w64/VC9 Makefiles and batch files were added to build GLPK under 32- and 64-bit Windows with Microsoft Visual Studio Express 2008. Thanks to Heinrich Schuchardt for the contribution and testing. * w32/DM, w32/OWC Makefiles and batch files were added to build GLPK with Digital Mars C/C++ 8.50 and Open Watcom C/C++ 1.6 (32-bit Windows). Wed Aug 13 12:00:00 2008 Andrew Makhorin * GLPK 4.30 (15:0:15) has been released * glpspx.h, glpspx03.c, glpapi06.c The primal simplex solver (spx_prim_opt, spx_prim_feas) was replaced by a new implementation (spx_primal), which currently provides the same features as the old version. * glpmpl01.c, glpmpl03.c Some changes were made in the MathProg translator to allow <, <=, >=, and > on comparing symbolic values. Thanks to Heinrich Schuchardt for patches. * glplpx10.c Internal routine set_d_eps in the exact LP solver was changed to prevent approximation errors in case of integral data. Thanks to Markus Pilz for bug report. XXX XXX XX 12:00:00 2008 Andrew Makhorin * GLPK 4.29 (14:0:14) has been released * configure.ac The configure script was changed to disable optional features by default. For details see file INSTALL. * glpipp02.c A bug was fixed in the internal routine reduce_bounds. Thanks to Anne-Laurence Putz for the bug report. * glpapi01.c New API routine glp_erase_prob was added. * glpapi13.c New API routines glp_read_mps and glp_write_mps were added. They replace API routines lpx_read_mps, lpx_read_freemps, lpx_write_mps, and lpx_write_freemps, which are deprecated. * glpapi14.c New API routines glp_read_lp and glp_write_lp were added. They replace API routines lpx_read_cpxlp and lpx_write_cpxlp, which are deprecated. * glpsql.c Minor bug was fixed. Thanks to Xypron for the bug report. Tue Mar 25 12:00:00 2008 Andrew Makhorin * GLPK 4.28 (13:0:13) has been released * glplib.h, glplib.c Three wrapper routines xdlopen, xdlsym, and xdlclose, which provide the shared library support, were added. A particular version of these routines depends on the option --enable-dl passed to the configure script (see file INSTALL for details). Thanks to Rafael Laboissiere for useful advices concerning the shared library support. * glpsql.c A static linking to iODBC and MySQL libraries used in the MathProg table drivers was replaced by a dynamic linking to corresponding shared libraries. Many thanks to Heinrich Schuchardt for the contribution and to Vijay Patil for testing this feature under Windows XP. * glpgmp.h, glpgmp.c A bug (which appeared only on 64-bit platforms) was fixed. Thanks to Axel Simon for the bug report. * glpapi.c A bug was fixed in the api routine glp_add_cols. (If the basis is valid, adding column keeps it valid, however, col->bind was set to -1 rather to 0.) Thanks to Cedric[FR] for the bug report. * glplib.c 64-bit unsigned int type glp_ulong and corresponding routines were replaced by 64-bit signed int type xlong_t. * glpk.h, glpapi.c The type glp_ulong was replaced by glp_long. This affects only the api routine glp_mem_usage. * glplib.c Compressed data file support was added. This feature requires the zlib data compression libraries and allows compressing and decompressing .gz files "on the fly". * glpcli.h, glpcli.c Command-line interface routines were added. (This feature is incomplete so far.) Sun Mar 02 12:00:00 2008 Andrew Makhorin * GLPK 4.27 (12:0:12) has been released * glpsql.h, glpsql.c Two MathProg table drivers for iODBC and MySQL contributed by Heinrich Schuchardt were added to the package. * glpmpl05.c Mathprog table driver for xBASE was added to the package. * glpmpl03.c A minor was fixed in the MathProg translator (if some field specified in the table statement is missing in corresponding input table, the bug causes abnormal termination). Thanks to Heinrich Schuchardt for the bug report. * glpmpl.h, glpmpl.c STRING data type was replaced by plain character strings. Sun Feb 17 12:00:00 2008 Andrew Makhorin * GLPK 4.26 (11:0:11) has been released * glpmpl.h, glpmpl01.c, glpmpl03.c, glpmpl05.c The table statement was implemented. Description of this new feature is given in file doc/tables.txt. * glpios03.c A bug causing zero divide error on computing euclidean norm of the cut coefficient vector was fixed. Wed Dec 19 12:00:00 2007 Andrew Makhorin * GLPK 4.25 (10:0:10) has been released * glpapi10.c Routines lpx_eval_tab_row and lpx_eval_tab_col were replaced by glp_eval_tab_row and glp_eval_tab_col. * glpios03.c, glpios05.c Gomory's mixed integer cuts were implemented. * glpscs.h, glpscs.c Segmented character string routines are no longer used and were removed from the package. Wed Nov 21 12:00:00 2007 Andrew Makhorin * GLPK 4.24 (9:0:9) has been released * src/glplpx16.c A bug was fixed in the routine lpx_write_cpxlp. If a variable x has upper bound and no lower bound, it should appear in the bounds section as "-inf <= x <= u", not as "x <= u". Thanks to Enric Rodriguez for the bug report. * src/glpios03.c, src/glpios04.c, src/glpios05.c MIR (mixed integer rounding) cuts were implemented. Sun Oct 28 12:00:00 2007 Andrew Makhorin * GLPK 4.23 (8:0:8) has been released * src/glplib05.c, configure.ac Check for vsnprintf was added. * include/glppds.h, src/glppds.c A module to scan plain data was added. * src/glpapi09.c The following new API routines were added: glp_read_sol read basic solution from text file; glp_write_sol write basic solution to text file; glp_read_ipt read interior-point solution from text file; glp_write_ipt write interior-point solution to text file; glp_read_mip read MIP solution from text file; glp_write_mip write MIP solution to text file. * src/glpapi12.c Advanced API routine glp_free_env was added. * examples/glpsol.c The following three command-line options were added: --mipgap tol set relative MIP gap tolerance -r filename read solution from filename -w filename write solution to filename Wed Sep 19 12:00:00 2007 Andrew Makhorin * GLPK 4.22 (7:0:7) has been released * src/glpios02.c A bug was fixed in the MIP preprocessor (ios_preprocess_node). Thanks to Roberto Bagnara (Department of Mathematics, University of Parma, Italy) for the bug report. * src/glpios02.c A bug was fixed in the MIP preprocessor (col_implied_bounds), due to which constraint coefficients with small magnitude could lead to wrong implied bounds of structural variables. * src/glpipp02.c A similar bug was fixed in the routine reduce_bounds. * src/glpapi01.c A bug was fixed in the routines glp_set_mat_row and glp_set_mat_col. (The bug appeared due to incorrect removing zero elements from the row/column lists.) * src/glplpx14.c A bug was fixed in the API routines lpx_read_mps and lpx_read_freemps, due to which bounds of type LI specified in BOUNDS section were incorrectly processed. * src/glplib05.c A call to standard function vsprintf was replaced by a call to vsnprintf for security reasons. Many thanks to Peter T. Breuer and Rafael Laboissiere . Tue Aug 28 12:00:00 2007 Andrew Makhorin * GLPK 4.21 (6:0:6) has been released * glpscg.h, glpscg.c Routines to maintain sparse cliqued graph were added. * glpios02.c MIP preprocessing routines were added. * glpk.h, glpios.h, glpios03.c New reasons for calling the callback routine were introduced in the MIP solver. * glpapi08.c Default backtracking strategy was changed to best local bound. * glpapi11.c New API routine glp_term_out to enable/disable terminal output was added. * glprng.h, glprng02.c Two routines to generate uniformly distributed pseudo-random floating-point numbers were added. Thu Jul 26 12:00:00 2007 Andrew Makhorin * GLPK 4.20 (5:0:5) has been released * glpk.h, glpapi08.c The routine lpx_integer was replaced by an equivalent routine glp_intopt. Also new API routine glp_init_iocp was added. * glpiet.h, glpiet.c Routines implementing the implicit enumeration tree are no longer used and therefore were removed from the package. * glpios.h, glpios01.c, glpios02, glpios03 Routines implementing the integer optimization suite being replaced by a new version were removed from the package. * glpmip.h, glpmip01.c, glpmip02.c Routines implementing the B&B method being replaced by a new version were removed from the package. * glpios.h, glpios01.c, glpios02.c Routines implementing a new version of the integer optimization suite (IOS) based on the B&B method were added to the package. * glpk.h, glpapi10.c Branch-and-bound interface routines were added to the package. * examples/tspsol.c The TSP solver based on old version of the integer optimization suite is no more supported and was removed from the package. * glpipp02.c An error in the routine reduce_bounds was fixed; thanks to Graham Rockwell for the bug report. * glpk.latex A new edition of the reference manual was included. Thu Jul 05 12:00:00 2007 Andrew Makhorin * GLPK 4.19 (4:0:4) has been released The principal change is upgrading to GPLv3. * glpapi01.c A serious bug in the routine glp_del_cols was fixed; thanks to Cedric[FR] for the bug report. The bug appeared because on deleting non-basic columns the basis header remained valid, however, contained invalid (old) column ordinal numbers. * glpapi10.c A new advanced API routine glp_mem_limit was added. * glplpx01.c The case GLP_EBOUND was added to the routine lpx_simplex. Thanks to Cameron Kellough for the bug report. * glplpx19.c An API routine lpx_write_pb to write the problem instance in OPB (pseudo boolean) format format was added. Thanks to Oscar Gustafsson for the contribution. * glpsol.c Two new options --wpb and --wnpb were added to glpsol to write the problem instance in OPB format. Mon Jun 25 12:00:00 2007 Andrew Makhorin * GLPK 4.18 (3:0:3) has been released * glplib.h Type names ulong_t and uldiv_t were changed to glp_ulong and glp_uldiv to avoid conflicts with standard type names on some platforms. Thanks to Boris Wirtz for the bug report. * glpbfd.*, glpfhv.*, glplpf.* LP basis factorization routines were made tidy. * glpk.h, glpapi04.c The following API routines were added: glp_set_rii, glp_set_sjj, glp_get_rii, glp_get_sjj. * glpk.h, glpapi06.c The routine lpx_simplex was replaced by an equivalent routine glp_simplex. Also new API routine glp_init_smcp was added. * glpk.h, glpapi09.c The following advanced API routines were added: glp_bf_exists, glp_factorize, glp_bf_updated, glp_get_bfcp, glp_set_bfcp, glp_get_bhead, glp_get_row_bind, glp_get_col_bind, glp_ftran, glp_btran. * glpk.latex A new edition of the reference manual was included. * examples/dea.mod, examples/food.mod, examples/food2.mod Three examples in the MathProg language were added. Thanks to Sebastian Nowozin . Sat May 26 12:00:00 2007 Andrew Makhorin * GLPK 4.17 (2:0:2) has been released * glpdmp.h, glpdmp.c Memory pool routines were replaced by a new version. * glpscs.h, glpscs.c Segmented string routines were replaced by a new version. * glplpx08.c, glplpx09.c Now the MIP problem may have no integer columns. * glpapi01.c The routines glp_set_mat_row, glp_set_mat_col, and glp_load_mat were modified to allow zero elements (which are not stored in the constraint matrix). * glpscf.h, glpscf.c Schur complement factorization routines were implemented. * glplpf.h, glplpf.c LP basis factorization routines based on LU-factorization and Schur complement were implemented. * glplpx02.c, glplpx03.c New control parameter LPX_K_BFTYPE was introduced to choose the basis factorization type used by the simplex method routines. * glpsol.c Three new command-line options were added to choose the basis factorization type used by the simplex method routines: --luf, --cbg, and --cgr. * glpk.latex A new edition of the reference manual was included. Sat May 05 12:00:00 2007 Andrew Makhorin * GLPK 4.16 (1:0:1) has been released * glpk.h, glpapi.c, glplpx01.c, glplpx02.c Names of a number basic api routines were changed and now have the prefix 'glp_'. To keep backward compatibility these routines are also available via their old names prefixed with 'lpx_'. * glplpx19.c Three new api routines were added: glp_version, glp_term_hook, and glp_mem_usage. * glpk.latex, gmpl.texi A new edition of the reference manuals was included. * lpglpk40.c This example program is no longer supported and therefore was removed from the package. Sun Feb 18 12:00:00 2007 Andrew Makhorin * GLPK 4.15 (0:0:0) has been released * configure.ac, Makefile.am Autotools specification files were changed to use GNU Libtool that allows building the static as well as shared GLPK library. Thanks to Rafael Laboissiere . Mon Feb 05 08:00:00 2007 Andrew Makhorin * GLPK 4.14 has been released Now GLPK conforms to ILP32, LLP64, and LP64 programming models (the latter seems to be the ultimate choice regarding 64-bit architectures). Note that GLPK itself is a 32-bit application, and the conformity only means that the package works correctly on all these arenae. Nevertheless, on 64-bit platforms it is possible to use more than 4GB of memory, if necessary. * Makefile Starting from this release only the header glpk.h is needed to be installed. * glplib01.c Two routines bigmul and bigdiv which performs multiplication and division of unsigned integers of arbitrary precision were added. * glplib02.c A set of 64-bit arithmetic routines were added. * glplib04.c Some low-level library routines were improved and renamed. * glpcfg.h The macro GLP_TM_SPEC were introduced to specify a version of the time routine depending on the host environment. Mon Nov 13 12:00:00 2006 Andrew Makhorin * GLPK 4.13 has been released * configure.in '-lm' bug was fixed. * glpbfx.h, glpbfx.c Basis factorization interface routines based on exact (bignum) arithmetic were implemented. * glpssx.h, glpssx1.c, glpssx2.c Simplex method routines based on exact (bignum) arithmetic were implemented. * glplpx6e.c The routine lpx_exact, which is an easy-to-use driver to the exact simplex method, was added. * glpsol.c Two command-line options were added: '--exact' and '--xcheck'. Wed Nov 08 12:00:00 2006 Andrew Makhorin * GLPK 4.12 has been released * glpcfg.h The package configuration file was added. * glplib2.c Alternative version of the routines umalloc, ucalloc, and ufree was provided. It does not limit the amount of allocated memory to INT_MAX bytes and therefore can be used on platforms where sizeof(void *) > sizeof(int). To enable this version one should define the preprocessor variable GLP_HUGE_MEM. * glprng.c The routine rng_create_rand was changed to initialize the generator using seed = 1, not 0, to conform ISO C requirements. * glpgmp.h, glpgmp.c A set of bignum arithmetic routines implementing operations on integers and rationals was added. These routines are compatible with the GNU MP library. NOTE: To attain a much better performance it is recommended to use, if possible, the original GNU MP library rather than the GLPK version, by defining the preprocessor variable GLP_USE_GMP. * glplux.h, glplux.c A tentative implementation of sparse LU-factorization based on exact (bignum) arithmetic was added. * glpssx.h, glpssx.c A tentative implementation of some simplex method routines based on exact (bignum) arithmetic was added. * glplpx6f.c A preliminary implementation of the routine lpx_exact_check was added. This routine checks the current basis for primal and dual feasibility using exact (bignum) arithmetic. * examples/glpsol.c The command-line option '--xcheck' was introduced to check the current basis for feasibility using exact (bignum) arithmetic. Tue Jul 25 12:00:00 2006 Andrew Makhorin * GLPK 4.11 has been released. * include/glpbfi.h, src/glpbfi.c Basis factorization interface routines were added. * include/glpluf.h, src/glpluf1.c Hypersparse solution routines were added. * include/glpinv.h, src/glpinv1.c Hypersparse solution routines (fake version) were added. * include/glpmpl.h, src/glpmpl.c Built-in functions card, length, and substr were implemented. Output redirection in the printf statement was implemented. * examples/graph.mod, examples/crypto.mod Two example models illustrating new features of the modeling language were included. Thu May 11 12:00:00 2006 Andrew Makhorin * GLPK 4.10 has been released. * src/glplpx8a.c A fragment was added to the routines lpx_read_mps and lpx_read_freemps to accept LI bound type (it is similar to LO, however, additionally marks the column as integer). * include/glpbfi.h, src/glpbfi.c The module glpbfi which implements the basis factorization interface (BFI) was added. * src/glplpx7a.c The routine lpx_cover_cut to generate mixed cover cuts was added. * src/glplpx7b.c The routine lpx_clique_cut to generate clique cuts and related routines to maintain the conflict graph were added. * include/glplpx.h, src/glplpx5.c The routine lpx_cpx_basis implementing Bixby's algorithm to construct an initial LP basis was added. * examples/glpsol.c Command-line option '--bib' was added which allows building an initial LP basis using Bixby's algorithm. Default command-line option '--mps' was changed to '--freemps'. * examples/cf12a.mod, examples/cf12b.mod Two examples in MathProg (curve fitting problem) were added. Thanks to Dr. Harley Mackenzie . Tue Jan 17 12:00:00 2006 Andrew Makhorin * GLPK 4.9 has been released. * glpipp.h, glpipp1.c, glpipp2.c A MIP presolver were implemented (currently incomplete). It is used internally in the routine lpx_intopt (see below). * glplpx6d.c, glplpx7a.c An advanced branch-and-bound solver (the routine lpx_intopt) were implemented. * glplpx6c.c The routine lpx_check_int to check MIP feasibility conditions was added. * glplpx8a.c The routine lpx_print_mip was changed to print MIP feasibility conditions. * glpmpl.h, glpmpl1.c, glpmpl3.c The built-in functions sin, cos, atan, and atan2 were added to the MathProg language. * doc/lang.* Some typos were fixed. Thanks to Minh Ha Duong (CIRED, CNRS). Wed Jan 12 12:00:00 2005 Andrew Makhorin * GLPK 4.8 has been released. * glpspx.h, glpspx1.c, glpspx2.c, glplpx6a.c Simplex method routines were changed due to a new format of the constraint matrix. * glpmat.h, glpmat.c Sparse matrix routines were re-implemented using storage-by-rows format. * glpipm.h, glpipm.c, glplpx6b.c Interior-point method routines were changed due to a new format of sparse matrices. * glpchol.h, glpchol.c Old version of Cholesky factorization routines being replaced by a new one (see glpmat.c) was removed from the package. * glplpx8c.c Minor bug was fixed in api routine lpx_read_cpxlp. Mon Aug 23 12:00:00 2004 Andrew Makhorin * GLPK 4.7 has been released. * glplpx.h, glplpx1.c New core API routines were added (but not documented yet): lpx_order_matrix, lpx_create_index, lpx_find_row, lpx_find_col, lpx_delete_index. * glplpx8a.c API routine lpx_read_mps was re-implemented, and two new API routines lpx_read_freemps and lpx_write_freemps were added to support free MPS format. * glplpx8c.c Two API routines lpx_read_cpxlp and lpx_write_cpxlp (formerly named lpx_read_lpt and lpx_write_lpt) were re-implemented. * glpmps.h, glpmps.c This module formerly used in lpx_read_mps was removed from the package. * glplpt.h, glplpt.c This module formerly used in lpx_read_lpt was removed from the package. * glpmip.h, glpmip1.h, glpmip2.h New MIP routines mip_best_node and mip_relative_gap were added due to suggestion of Brady Hunsaker . * glpsol.c The following new command-options were added: --freemps to read problem data in free MPS format --wfreemps to write problem data in free MPS format --cpxlp to read problem data in CPLEX LP format --wcpxlp to write problem data in CPLEX LP format --bas to read LP basis from a text file in MPS format --wbas to write LP basis to a text file in MPS format --mostf to use "most fractional" branching heuristic --bestb to use "best bound" backtracking heuristic * contrib/deli/*.* GLPK Delphi interface module was temporarily removed from the distribution due to licensing problems. * contrib/glpkmex/*.* GLPK Matlab interface module was temporarily removed from the distribution due to licensing problems. * contrib/jni/*.* GLPK Java interface module was temporarily removed from the distribution due to licensing problems. Wed Aug 04 12:00:00 2004 Andrew Makhorin * GLPK 4.6 has been released. * glpmpl.h, glpmpl1.c, glpmpl2.c, glpmpl3.c, glpmpl4.c Three new statements were implemented in the GNU MathProg language: solve, printf, and for. Also some bugs were fixed. * glplpx.h, glplpx8e.c Two API routines were added: lpx_read_prob and lpx_write_prob, which allow reading and writing problem data in GNU LP format. * glpsol.c Three new command-line options were added: --glp (to read problem data in GNU LP format), --wglp (to write problem data in GNU LP format), and --name (to change problem name). * glprng.h, glprng.c A portable pseudo-random number generator was implemented as a separate module. * glplib4.c The old implementation of a pseudo-random number generator was removed from the package. * doc/lang.*, doc/refman.* New edition of the GLPK documentation was included. * contrib/glpkmex/*.* A new version of GLPKMEX was included in the distribution. For more details see contrib/glpkmex/ChangeLog. Mon Jul 19 12:00:00 2004 Andrew Makhorin * GLPK 4.5 has been released. * glpmip.h, glpmip1.c, glpmip2.c, glplpx6c.c New implementation of the branch-and-bound method was added. It replaces the old implementation, which was removed from the package. * glpies.h, glpies1.c, glpies2.c, glpies3.c Modules used in the old implementation of the branch-and-bound method were removed from the package. * glplib2.c Now if the preprocessor variable GLPHUGEMEM is defined, other version of the routines umalloc, ucalloc, and ufree is used on compiling the package. This allows avoiding memory allocation problems on platforms where sizeof(void *) > sizeof(int), for example, where addresses are 64-bit while integers are 32-bit. The modification was made due to a bug report provided by Karel Zimmermann and Christophe Caron . Sat Jan 17 12:00:00 2004 Andrew Makhorin * GLPK 4.4 has been released. * glplpx.h, glplpx*.c All API routines were re-implemented using new data structures. Some new API routines were added and some existing API routines became obsolete as shown below: Obsolete API routine Equivalent new API routine lpx_check_name (no more supported) lpx_set_obj_c0 lpx_set_obj_coef lpx_set_row_coef (no more supported) lpx_set_col_coef lpx_set_obj_coef lpx_load_mat (no more supported) lpx_load_mat3 lpx_load_matrix lpx_unmark_all (no more supported) lpx_mark_row (no more supported) lpx_mark_col (no more supported) lpx_clear_mat (no more supported) lpx_del_items lpx_del_rows, lpx_del_cols lpx_get_row_bnds lpx_get_row_type, lpx_get_row_lb, lpx_get_row_ub lpx_get_col_bnds lpx_get_col_type, lpx_get_col_lb, lpx_get_col_ub lpx_get_obj_c0 lpx_get_obj_coef lpx_get_row_coef (no more supported) lpx_get_col_coef lpx_get_obj_coef lpx_get_row_mark (no more supported) lpx_get_col_mark (no more supported) lpx_get_row_info lpx_get_row_stat, lpx_get_row_prim, lpx_get_row_dual lpx_get_col_info lpx_get_col_stat, lpx_get_col_prim, lpx_get_col_dual lpx_get_ips_stat lpx_ipt_status lpx_get_ips_row lpx_ipt_row_prim, lpx_ipt_row_dual lpx_get_ips_col lpx_ipt_col_prim, lpx_ipt_col_dual lpx_get_ips_obj lpx_ipt_obj_val lpx_get_mip_stat lpx_mip_status lpx_get_mip_row lpx_mip_row_val lpx_get_mip_col lpx_mip_col_val lpx_get_mip_obj lpx_mip_obj_val Obsolete API routines were kept for backward compatibility, however, they will be removed in the future. * doc/refman.* New edition of the GLPK reference manual containing description of all new API routines was included. * contrib/glpkmex/*.* GLPKMEX, a Matlab MEX interface to GLPK package, contributed by Nicolo Giorgetti was included. * doc/GLPK_FAQ.txt GLPK FAQ contributed by Harley Mackenzie was included. Fri Dec 12 12:00:00 2003 Andrew Makhorin * GLPK 4.3 has been released. * configure.in The bug, due to which the standard math library is not linked on some platforms, was fixed. * glpmpl3.c The bug (0 ** y) was fixed in the routine fp_power. * glpmpl.h, glpmpl1.c, glpmpl3.c Some new built-in functions (round, trunc, Irand224, Uniform01, Uniform, Normal01, Normal) were added to the MathProg language. * glpmpl1.c The MathProg syntax was changed to allow writing 'subj to'. * glplpx.h, glplpx1.c, glplpx2.c The new api routine lpx_get_ray_info was added. * glplpx8a.c The api routine lpx_print_sol was changed to print the number of non-basic variable, which causes primal unboundness. * glpmps.c The code was changed to avoid errors on compiling the package on Mac OS X. Thanks to Andre Girard for the bug report. * doc/lang.*, doc/refman.* Several typos were fixed and some new material was added in the glpk documentation. Fri Nov 14 12:00:00 2003 Andrew Makhorin * GLPK 4.2 has been released. * glpiet.h, glpiet.c, glpios.h, glpios1.c, glpios2.c, glpios3.c A preliminary implementation of the Integer Optimization Suite (IOS) was included in the package. Eventually IOS will replace the Implicit Enumeration Suite (IES). * glplpx.h, glplpx6d.c A dummy version of the integer optimization routine lpx_intopt was included in the package. Later this routine will replace the routine lpx_integer. * examples/glpsol.c A new command-line option --int-opt was added to the solver to call lpx_intopt rather than lpx_integer. * glpbcs.h, glpbcs1.c, glpbcs2.c Being replaced by IOS routines (see above) the Branch-and-Cut Framework (BCS) routines were removed from the package. * examples/tspsol.c Stand-alone Symmetric TSP solver was completely re-programmed using IOS routines. * glplib.h, glplib2.c, glplib4.c The random-number generator was implemented. It is based on the module GB_FLIB from the Stanford GraphBase originally developed by Donald Knuth. * glphbsm.c, glplpx8a.c, glpmps.c All calls to fopen/fclose were replaced by corresponding calls to ufopen/ufclose due to bug reports provided by Morten Welinder and . * glpmps.c The code was made re-entrant. * glplpx8b.c API routine lpx_print_sens_bnds for bounds sensitivity analysis contributed by Brady Hunsaker was added to the package. This feature is also available in glpsol via the command-line option --bounds. * contrib/jni/*.* New version of GLPK JNI (Java Native Interface) contributed by Chris Rosebrugh was added to the package. * contrib/deli/*.* GLPK DELI (Delphi Interface) contributed by Ivo van Baren was added to the package. * glplpx3.c Default method to scale the problem was changed to equilibration scaling (lp->scale = 1 in lpx_reset_parms). * glplpx6a.c Two minor (non-critical) typos were fixed due to report provided by Andrew Hamilton-Wright . * glplpp2.c An untested case (line 941) had been tested due to bug report provided by Jiri Spitz . * w32bc5.mak, w32vc6.mak, w32vc6d.mak, d32dmc.mak Several makefiles were added to allow building GLPK library for some non-GNU 32-bit platforms. Sat Aug 23 12:00:00 2003 Andrew Makhorin * GLPK 4.1 has been released. * glpmpl1.c, glpmpl3.c Some bugs were fixed in the MathProg translator due to the bug reports provided by Giles Thompson : conditional set expressions were incorrectly parsed; dimen attribute was not set by default when a set was used recursively in its own declaration; logical expressions ... in if ... then ... else ... did not work; displaying set expressions did not free memory allocated for temporary results. * glpmpl3.c (reduce_terms) Implementation of summation of linear forms over domain was improved to reduce complexity of that operation from O(n*n) to O(n*log n). The improvement was made due to a report provided by Sebastien de Menten . * glplpx6a.c (line 1056), glpmip1.c (line 641) Two minor bugs were fixed due to the bug report provided by Kendall Demaree . * glplpx.h, glplpx6a.c The method of one artificial variable implemented in the routine lpx_prim_art and used on the phase I in the glpk simplex solver has a serious defect: for some lp instances it erroneously reports that the problem has no primal feasible solution. This error appears when the column of the artificial variable, which enters the basis to make it primal feasible, has large constraint coefficients, that leads to small reduced costs of non-basic variables and premature termination of the search, i.e. to wrong conclusion that the problem has no primal feasible solution. To avoid this defect the routine lpx_prim_feas was included. It implements the method of implicit artifical variables (based on minimization of the sum of infeasibilities), which is a bit slower but much more robust. The routine lpx_prim_feas having the same functionality now is used instead the routine lpx_prim_art. * glpinv.h, glpinv.c The test used in the routine inv_update to detect low accuracy after updating LU-factorization of the basis matrix was replaced by a new, more robust test. * glplpx6c.c Selecting an active node to be solved next in the routine btrack_bestp was changed. Now, if any integer feasible solution has not been found yet, the routine chooses an active node which has the minimal sum of integer infeasibilities. * glpmip.h, glpmip1.c The additional flag int_obj was included in the structure MIPTREE used by the branch-and-bound. This flag is set in the routine mip_create_tree and used in the routine is_better. It means that the objective is integral, i.e. depends only on integer variables with integer objective coefficients. The test used in the routine check_integrality was also replaced by a new, more reasonable one. * glplpx1.c A minor bug was fixed in the routine lpx_check_name. * glpmpl.h, glpmpl4.c, glplpx8d.c The flag skip_data was added to the parameter list of the routine mpl_read_model. If this flag is set, the data section in the model file is ignored. Corresponding change was made in the routine lpx_read_model. Now, if both model and data files are specified, the data section in the model file is ignored. * glplpx8c.c A minor bug (wrong format used for writing free columns) in the routine lpx_write_lpt was fixed due to the bug report provided by Bernhard Schmidt * sample/glpsol.c The command-line parameter --tmlim, which allows limiting the solution time, was added. * doc/lang.*, doc/refman.* New edition of the GLPK documentation was included. * java-binding/*.* New version of the GLPK JNI (Java Native Interface) package was included in the distribution. * sample/lpglpk40.c A non-trivial example was added. It allows using GLPK as a base LP solver for Concorde, a program for solving Traveling Salesman Problem (TSP). For details see comments in lpglpk40.c. * sample/*.mod Some examples of LP and MIP models written in GNU MathProg were added. Tue May 06 12:00:00 2003 Andrew Makhorin * GLPK 4.0 has been released. * glpmpl.h, glpmpl1.c, glpmpl2.c, glpmpl3.c, glpmpl4.c The model translator for the GNU MathProg modeling language was implemented and included in the package. * glplpx.h, glplpx8d.c The api routine lpx_read_model, which is an interface to the MathProg translator, was included in the package. * glplpx.h, glplpx8a.c The api routine lpx_print_prob for writing LP/MIP problem data in plain text format was included in the package. * sample/glpsol.c New version of the GLPK stand-alone LP/MIP solver that supports the GNU MathProg modeling language was implemented. * doc/lang.latex, doc/lang.dvi, doc/lang.ps The document "GLPK: Modeling Language GNU MathProg" was included in the package. * doc/refman.latex, doc/refman.dvi, doc/refman.ps New edition of the GLPK Reference Manual was included in the package. * glplpx8c.c A bug in the api routine lpx_write_lpt was fixed. Due to that bug an addressing error occured in the routine if the objective function has the non-zero constant term. * glplan.h, glplan1.c, glplan2.c, glplan3.c, glplan4.c, * glplan5.c, glplan6.c, glplan7.c, glplan8.c, glplpx8b.c All modules of the translator for the GLPK/L modeling language were removed from the package, because GLPK/L being completely superseded by GNU MathProg is no more supported. Tue Mar 25 12:00:00 2003 Andrew Makhorin * GLPK 3.3 has been released. * glplpp.h, glplpp1.c, glplpp2.c An implementation of the built-in LP presolver was added to the package. * glplpx.h The flag presol was added to the structure LPX. This flag tells the lpx_simplex whether the built-in LP presolver should be used or not. By default this flag is off. Also three macros (namely LPX_E_NOPFS, LPX_E_NODFS, and LPX_K_PRESOL) that concern using the LP presolver were introduced. * glplpx3.c, glplpx6a.c These modules was changed to use the built-in LP presolver. * sample/glpsol.c Command line options --presol and --nopresol that concern using the LP presolver were added to the stand-alone LP/MIP solver. * glplan1.c This module was changed to allow declaring sets like A[1:10] in the models written in the GLPK/L modeling language. * doc/refman.latex, doc/lang.latex New editions of the documents "GLPK User's Guide" and "GLPK/L Modeling Language" were included in the distribution. * java-binding/*.* The package GLPK JNI (Java Native Interface) implementing Java binding for GLPK was included in the distribution. This package was developed and programmed by Yuri Victorovich . Tue Feb 18 12:00:00 2003 Andrew Makhorin * GLPK 3.2.4 has been released. * glplpx6b.c The code was changed to allow auxiliary variables have non-zero objective coefficients. Also a minor bug was fixed (the constant term was not considered on displaying the objective function value). * sample/glpsol.c The code was changed to fix a bug (the command-line option 'bfs' was not recognized). The bug was fixed due to report provided by Olivier . * glplpt.c The code was changed to fix a bug (binary variables were treated erroneously as integer ones). * glplpx6b.c The code was changed to fix a bug (variables that have no lower bounds were incorrectly processed on converting to the standard formulation). The bug was fixed due to report kindly provided by Kjell Eikland . Mon Nov 11 12:00:00 2002 Andrew Makhorin * GLPK 3.2.3 has been released. * glpmip.h, glpmip1.c A preliminary implementation of the branch-and-bound driver based on the implicit enumeration suite (glpies) was added to the package. This module is not documented yet. * glplpx6c.c A new implementation of the api routine lpx_integer which now is based on the b&b driver (see glpmip above) was included in the package. This new implementation has exactly the same functionality as the old version and therefore all changes are transparent to the api user. * glpbbm.h, glpbbm.c * glprsm.h, glprsm1.c, glprsm2.c * glplp.h, glplp.c These modules were removed from the package, because they were used only in the old version of the routine lpx_integer, which was replaced by the new version (see glplpx6c above). * glplpx.h, glplpx6a.c The api routine lpx_check_kkt was included in the package and its description was added in the reference manual. This routine allows checking Karush-Kuhn-Tucker optimality conditions for an LP solution. * glplpx.h, glplpx8a.c Now the api routine lpx_print_sol also prints information about "solution quality" obtained via the api routine lpx_check_kkt. * glplpx.h, glplpx8a.c New api routines lpx_read_bas and lpx_write_bas were included in the package and documented. The routine lpx_write_bas allows writing a current basis from an LP object to a text file in the MPS format. The routine lpx_read_bas allows reading a basis prepared in the MPS format from a text file into an LP object. * glplpt.c The parsing routine which reads LP problem data prepared in the CPLEX LP format was modified to allow specifying lower bounds of variables also in the form 'variable >= lower bound' (in the bounds section). This modification was made due to a notice provided by Ivan Luzzi . * glplpx.h, glplpx8c.c The api routine lpx_write_lpt which allows writing LP problem data from an LP object to a text file using the CPLEX LP format was included in the package and documented. * glplpx.h, glplpx3.c The control parameter LPX_K_LPTORIG that affects the behavior of the api routine lpx_write_lpt was introduced. * glplan6.c The semantics of the language GLPK/L was changed to allow selection in case when not all mute letters of a predicate (the operand that follows the keyword 'where') are presented in a parameter (the operand that precedes the keyword 'where'), i.e. to allow writing something like this: y[j] := sum(i, x[i] where p[i,j]); The paragraph "Selection" in the langauge description (page 25) was also correspondingly changed. This change of the language semantics was undertaken due to a notice provided by Peter Lee . * sample/hwd.lpm A nice example of LP model written in GLPK/L and contributed by Peter Lee was included in the package. * glplpx6b.c The api routine lpx_interior was modified: a) to compute dual values for all structural as well as auxiliary variables; b) to allow specifying non-zero objective coefficients at auxiliary variables. * sample/glpsol.c Three new command-line options were added to the solver, which are: --plain, --orig, and --wrlpt. Mon Oct 14 12:00:00 2002 Andrew Makhorin * GLPK 3.2.2 has been released. * glplpt.h, glplpt.c A module that reads LP/MIP problem data in CPLEX LP format was implemented. * glplpx8c.c An api routine lpx_read_lpt that reads LP/MIP problem data in CPLEX LP format was implemented. * sample/glpsol.c, sample/plan.lpt A new command-line option '--lpt' that allows reading LP/MIP problem data in CPLEX LP format was added to the solver. * doc/refman.latex, doc/refman.dvi, doc/refman.ps A new edition of the Reference Manual was included. * source/*.c Casting to (unsigned char) was added in some calls to the classification functions (isalpha, etc.). The bug was fixed due to report provided by Morten Welinder . * glplpx8a.c The local routine mps_numb used in the routine lpx_write_mps was modified to correctly format floating-point numbers that have two digits in the decimal exponent. The bug was fixed due to report provided by Vlahos Kiriakos . * glplan.h, glplan1.c, ..., glplan8.c Several serious bugs were fixed in the language processor due to reports provided by : (a) a static search tree used to find sparse array elements was sometimes overwritten that caused the message 'assertion failed' to appear; the bug was fixed by creating separate search trees in parsing routines; (b) a variable declared using the predicate-controlled variable declaration statement had wrong order of domain sets, because the variable array was built as a copy of the predicate array; the bug was fixed by using the internal routine transpose that coordinates mute letters (and therefore domain sets) on copying sparse arrays; (c) sometimes assignment statements like x[#a,#b,#c] := ... was incorrectly processed; the bug was fixed by including an appropriate check into the internal routine assign_stmt. * glp_simplex.c An additional check to see if all lower bounds are not greater than corresponding upper bounds was included in the routine to prevent wrong results to appear. Such incorrectness sometimes was not detected, namely, when variables with such bounds were non-basic and never entered the basis. * glpspx1.c Maximal number of simplex iterations before reinversion was decreased from 100 to 50. This allowed to improve accuracy and, that is more important, to reduce the solution time for many serial lp problems approximately 1.5--2 times. * glpspx2.c A check to see if all elements in the column chosen to enter the basis are close to zero in the routine spx_prim_chuzr was temporarily removed because this check gave wrong conclusion in case when the corresponding non-basic variable had zero column in the constraint matrix. An analogous check to see if all elements in the row chosen to leave the basis are close to zero in the routine spx_dual_chuzc was also temporarily removed on the same reason. The bug was fixed due to reports provided by Flavio Keidi Miyazawa and Vlahos Kiriakos . Mon Aug 12 12:00:00 2002 Andrew Makhorin * GLPK 3.2.1 has been released. * glpbcs.h, glpbcs1.c, glpbcs2.c * glpies.h, glpies1.c, glpies2.c, glpies3.c A preliminary implementation of the branch-and-cut framework was included in the package. * doc/brcut.txt The document "GLPK: A Preliminary Implementation of the Branch-And-Cut Framework" was included in the distribution. * sample/tspsol.c An illustrative program for solving symmetric TSP based on the branch-and-cut method was included in the package. * glpdmp.h, glpdmp.c A new, re-enterable version of routines for managing dynamic memory pools was included in the package. * glpavl.h, glpavl.c A new, re-enterable version of routines for managing AVL search trees was included in the package. * glplib.h, glplib2.c Two new low-level routines ufopen and ufclose were included in the package. * glplpx.h, glplpx7.c The following new api routines were added: lpx_eval_activity, lpx_eval_red_cost, lpx_reduce_form, lpx_mixed_gomory. * glptsp.h, glptsp.c A module for reading TSP data using TSPLIB format was included in the package. Mon Jul 15 12:00:00 2002 Andrew Makhorin * GLPK 3.2 has been released. * glplpx.h, glplpx1.c, glplpx2.c The identifier 'class' (used as a member name in the structure LPX and as an argument name in the routine lpx_set_class) was changed to 'clss' in order to avoid conflicts with C++ reserved words. * glpk.h, glplpx.h, glplpx1.c, glplpx2.c, glplpx6a.c, * glplpx6b.c, glplpx6c.c, glplpx7.c, glplpx8.c The following new api routines were added: lpx_set_obj_name, lpx_get_obj_name, lpx_get_row_mark, lpx_get_col_mark, lpx_transform_row, lpx_transform_col, lpx_prim_ratio_test, lpx_dual_ratio_test, lpx_interior, lpx_get_ips_stat, lpx_get_ips_row, lpx_get_ips_col, lpx_get_ips_obj, lpx_read_lpm, lpx_write_mps, lpx_print_ips. * glpsol.c The solver was completely re-programmed using new api routines. * lang.latex, lang.dvi, lang.ps New edition of the document "GLPK: Modeling Language GLPK/L" was included in the distribution. * refman.latex, refman.dvi, refman.ps New edition of the document "GLPK: Reference Manual" (which contains descriptions of all new api routines) was included in the distribution. * glpapi.h, glpapi1.c, glpapi2.c, glpapi3.c, glpapi4.c These files (which contain old api routines) were removed from the package. * glpipm1.c, glpipm2.c The file glpipm1.c was renamed to glpipm.c. The file glpipm2.c was used only by old api routines and therefore was removed from the package. * language.texinfo Old version of the document "GLPK: Modeling Language GLPK/L" was removed from the distribution. Mon May 27 12:00:00 2002 Andrew Makhorin * GLPK 3.1 has been released. * glplpx.h, glplpx1.c, glplpx2.c, glplpx3.c, glplpx4.c, * glplpx5.c, glplpx6.c, glplpx7.c, glplpx8.c A preliminary implementation of new API routines was completed. * refman.latex, refman.dvi, refman.ps A draft edition of the document "GLPK Reference Manual", which describes new API routines, was included. * glplib3.c A bug in measuring long time intervals was fixed up. * glprsm3.c This module contains some obsolete routines not longer used and therefore it was removed from the package (into the subdirectory 'oldsrc'). * glprsm.h Some declarations related to the module 'glprsm3.c' (see above) were removed. * guide.texinfo The document "GLPK User's Guide" describing old API routines was removed from the package (into the subdirectory 'oldsrc'). * newapi.txt The document "New GLPK API Routines" was removed at all, because it is superseded by the new reference manual (see above). Mon May 13 12:00:00 2002 Andrew Makhorin * GLPK 3.0.8 has been released. * glplpx.h, glplpx1.c, glplpx2.c, glplpx3.c, glplpx4.c, * glplpx5.c, glplpx6.c, glplpx7.c A preliminary (currently incomplete) implementation of new api routines was included. * sample/newsamp.c A sample program for the new api routines was included. * newapi.txt A draft of the document "New GLPK API Routines" was included. * glpapi2.c, glpapi5.c, glpapi6.c These modules (which contain the api routines glp_call_rsm1, glp_simplex1, glp_pivot_in, glp_pivot_out) were removed from the package (to the subdirectory 'oldsrc') since these routines are functionally superseded by the new api routines. * glpk.h, glpapi2.c, glpapi3.c, glpapi4.c The api routines glp_simplex2, glp_call_ipm1, glp_call_bbm1 were renamed to glp_simplex, glp_interior, glp_integer, respectively. * sample/glpsol.c Some command-line options (which got obsolete due to the recent changes in api) were excluded. * doc/guide.texinfo New edition of the document "GLPK User's Guide" was included in the distribution to reflect the changes in some api routines. * doc/libref.texinfo This document was removed from the package (to the subdirectory 'oldsrc') since it describes the library routines, most of which got obsolete and no longer used. * Makefile.in A minor bug was fixed up due to bug report from Hans Schwengeler . Mon Apr 22 12:00:00 2002 Andrew Makhorin * GLPK 3.0.7 has been released. * glpduff.h, glpduff.c, glpspx.h, glpspx1.c, glpspx2.c, * glpapi7.c These modules were replaced by a new implementation of the simplex method and therefore they were removed from the package (however they still can be found in the subdirectory 'oldsrc'). * glprsm1.c The routine crash_aa was replaced by a new implementation and therefore it was removed from the file 'glprsm1.c'. * glplpx.h, glplpx.c, glpspx.h, glpspx1.c, glpspx2.c, glpspx3.c, * glpspx4.c, glpapi7.c New (currently incomplete) implementation of the simplex method components was included in the package. Thu Mar 28 12:00:00 2002 Andrew Makhorin * GLPK 3.0.6 has been released. * glpluf.h, glpluf.c, glpinv.h, glpinv.c New version of LU-factorization and basis maintenance routines (based on Forrest-Tomlin updating technique) was implemented. * glpeta.h, glpeta.c, glpfhv.h, glpfhv.c, glpgel.h, glpgel.c, * glppfi.h, glppfi.c, glprfi.h, glprfi.c These routines implement some other forms of the basis matrix. Now they became obsolete being functionally superseded by the new version of basis maintenance routines (see above) and were removed from the package (however they still can be found in the subdirectory 'oldsrc'). * glpbbm.c, glprsm.h, glprsm1.h, glprsm2.h, glpspx.h, glpspx2.c, * glprsm2.c, glpsol.c Necessary changes were made in order to use the new version of basis maintenance routines. Tue Jan 29 12:00:00 2002 Andrew Makhorin * GLPK 3.0.5 has been released. Structure of the package was re-organized in order to simplify its maintenance. * doc/guide.texinfo New edition of the document "GLPK User's Guide" was included in the distribution. Now the document includes descriptions of some additional API routines recently added to the package. * doc/newapi.txt The document "Additional GLPK API Routines" was removed from the distribution, because the corresponding material was included in the user's guide (see above). Mon Dec 10 12:00:00 2001 Andrew Makhorin * GLPK 3.0.4 has been released. * glpspx.h, glpspx1.c, glpspx2.c, glpapi/glp_simplex2.h A new, more efficient version of the two-phase primal simplex method was implemented (advanced initial basis, projected steepest edge, recursive computations of solution components). * glpapi/glp_call_bbm1.c Now LP relaxation can be solved either using rsm1_driver(), or using glp_simplex2(). The choice is controlled by the parameter 'meth' (a member of struct bbm1). * sample/glpsol.c The new implementation of the simplex method is now used by default. The old version is available via --old-sim option. * glpmat/gm_scaling.c Now this routine displays only two lines: an initial "quality" and a final "quality". * glplp/prepro_lp.c Identifiers 'fmin' and 'fmax' renamed to 'f_min' and 'f_max' in order to avoid conflict with . The bug was fixed due to report provided by Sami Farin . Wed Oct 03 12:00:00 2001 Andrew Makhorin * GLPK 3.0.3 has been released. * glprsm/harris_row.c, glprsm/harris_col.c The relative tolerance used on the first pass of the two-pass ratio test was replaced by the absolute tolerance. * glprsm/rsm_primal.c, glprsm/rsm_feas.c, glprsm/rsm_dual.c The absolute tolerance passed to the two-pass ratio test routine was decaresed (for both primal and dual simplex). These changes were made in order to improve numerical stability of the simplex method. * glprsm/glp_call_rsm1.c, glprsm/glp_call_bbm1.c, * glprsm/glp_simplex1, glprsm/glp_pivoting.c Default form of the inverse was changed from RFI to AFI. Mon Sep 24 12:00:00 2001 Andrew Makhorin * GLPK 3.0.2 has been released. * glpfhv.h, glpfhv.c New version of the basis maintaining routines was implemented. These routines, which are based on so called FHV-factorization (a variety of LU-factorization) and Gustavson's data structures, perform the main operations on the basis matrix faster at the expense of some worsening numerical accuracy. * glprsm.h, glprsm/afi.c The routines, which implement AFI (Advanced Form of the Inverse) based on FHV-factorization, were added to the package. This new form is available via the parameter form = 3 (on API level) or via the option --afi (in GLPSOL solver). * EFI was renamed to PFI In order to correct terminology the acronym EFI (Elimination Form of the Inverse) was replaced by PFI (Product Form of the Inverse) everywhere in the source code and the documentation. * glpset/umalloc.c, glpset/ucalloc.c * glpset/get_atom.c, glpset/get_atomv.c These memory management routines were changed in order *not* to clear allocated memory blocks by binary zeros. Wed Aug 01 12:00:00 2001 Andrew Makhorin * GLPK 3.0.1 has been released. * glpapi/old_api.c, glplp/extract_lp.c, store_lpsol.c Old API routines were deleted from the package. * include/glpk.h, include/glpapi.h, include/glplp.h Specifications of old API routines and data structures were removed from the headers. * sample/glpsol.c New version of the stand-alone solver GLPSOL that now uses new API routines was implemented. * glpapi/glp_set_row_fctr.c, glpapi/glp_set_col_fctr.c, * glpapi/glp_get_row_fctr.c, glpapi/glp_get_col_fctr.c, * glpapi/glp_scale_prob.c Scaling routines were added. * glpapi/glp_write_mps.c The routine for writing problem data in MPS format was added. * glpapi/glp_simplex1.c Comprehensive driver to the simplex method was added. * glpapi/glp_pivoting.c The routines glp_pivot_in() and glp_pivot_out() intended for basis maintaining were added. * glprsm/create_rsm.c, glprsm/delete_rsm.c, glprsm/scale_rsm.c, * glprsm/build_basis.c Additional low level routines related to the simplex method were added. * glpk.h, glpapi.h, glprsm.h Additional specifications for new routines and data structures were added. * sample/lpglpk30.c A non-trivial example was added. It allows using GLPK as a base LP solver for Concorde, a program for solving Traveling Salesman Problem (TSP). For details see comments in 'lpglpk30.c'. * doc/newapi.txt The document "Additional GLPK API Routines" that describes some new API routines was included. Thu Jul 19 12:00:00 2001 Andrew Makhorin * GLPK 3.0 has been released. Now GLPK is provided with new API, which is intended for using the package in more complex algorithmic schemes. * glpapi/old_api.c All routines related to old API were gathered in one file named 'old_api.c'. * glpapi/*.c These routines that implement new API were added to the package. * include/glpk.h, include/glpapi.h Specifications of new API routines and data structures were added to these headers. Specifications of old API routines and data structures were locked by #ifdef GLP_OLD_API directive. * doc/guide.texinfo New edition of the document "GLPK User's Guide" that correspond to new API was included. Thu Jun 14 12:00:00 2001 Andrew Makhorin * GLPK 2.4.1 has been released. * doc/glpk_ml.texinfo The new document "Modeling Language GLPK/L" was included. * doc/glpk_ug.texinfo New edition of the document "GLPK User's Guide" was included. * doc/language.txt The preliminary document "GLPK/L Modeling Language: A Brief description" was removed from the distribution, because it has been replaced by the new document "Modeling Language GLPK/L". * glplang/l_spar.c The routine comparison() was re-programmed in order to implement the relation operation as specified in the language description. * glpmip.h, glpmip/*.c The partition 'glpmip' was renamed to 'glpbbm'. Thu May 10 12:00:00 2001 Andrew Makhorin * GLPK 2.4 has been released. Now GLPK includes an implementation of a preliminary version of the GLPK/L modeling language. * glplang.h, glplang/*.c The header 'glplang.h' and a set of routines that implements the GLPK/L language processor (the partition 'glplang') were added to the package. * doc/language.txt The document "GLPK/L Modeling Language: A Brief Description (Supplement to GLPK User's Guide)" in plain text format was included in the package (see the file 'language.txt' in the subdirectory 'doc' of the distribution). * ex/model1.lpm, ex/model2.lpm Two examples of model descriptions written in GLPK/L were added to the package. * sample/glpsol.c This program was modified in order: a) to allow processing model description written in GLPK/L; b) to allow solving pure LP problem using the interior point method. * sample/glpipm.c This program was removed from the package, because its function was passed to the GLPSOL solver. * Makefile.in This file was changed in order to install the GLPSOL solver executable. Mon Apr 09 12:00:00 2001 Andrew Makhorin * GLPK 2.3 has been released. * glpmip.h, glpmip/*.c These routines (that implement the branch-and-bound method) were re-programmed in order to improve robustness of implementation. In particular, heuristic routines were carried out from the main driver routine. Additional GLPK API routines were documented. New edition of the document "GLPK User's Guide" was included in the package. The preliminary document "Mixed Integer Programming Using GLPK Version 2.2 (Supplement to GLPK User's Guide)" was removed from the package, because this material was included in GLPK User's Guide. Thu Mar 15 12:00:00 2001 Andrew Makhorin * GLPK 2.2 has been released. Now GLPK includes a tentative implementation of the branch-and-bound procedure based on the dual simplex method for mixed integer linear programming (MIP). The preliminary document "Mixed Integer Programming Using GLPK Version 2.2 (Supplement to GLPK User's Guide)" was included into the package in plain text format (see the file 'mip.txt' in the subdirectory 'doc' of the distribution). * glpmip.h, glpmip/*.c, glpapi/glp_integer.c These routines (that implement the branch-and-bound method) were added to the package. * sample/glpsol.c This program was modified in order to allow solving LP and MIP problems. * glprsm/rsm_primal.c, glprsm/rsm_dual.c, glprsm/rsm_feas.c, * glprsm/rsm1_driver.c These routines (which are drivers to basic components of the revised simplex method) were added to the package. Mon Feb 19 12:00:00 2001 Andrew Makhorin * GLPK 2.1 has been released. * glprsm.h, glprsm/*.c These routines (that implement components of the revised simplex method) were re-programmed and documented. The document "GLPK Implementation of the Revised Simplex Method" was included into the package. Thu Jan 25 12:00:00 2001 Andrew Makhorin * GLPK 2.0 has been released. Now GLPK includes a tentative implementation of the primal-dual interior point method for large-scale linear programming (for more details see the file `NEWS' in the distribution). A number of routines related to the interior point method were added to the package. * insist.c The routine `insist' and the macro of the same name were introduced into the package in order to replace the standard macro `assert'. Some routines require the expression specified in the `assert' macro to be evaluated, but compiling the package with NDEBUG option prevents from that. This bug was fixed due to bug report provided by Peter A. Huegler . * Makefile.in Minor bug was fixed due to a patch provided by Alexandre Oliva . Wed Jan 10 12:00:00 2001 Andrew Makhorin * GLPK 1.1.2 has been released. * umalloc.c, ufree.c, create_pool.c, get_atom.c, get_atomv.c These routines were changed in order to fix a bug due to report provided by Andrew Hood . Because of this bug data alignment error occured on the Sparc computer. Tue Dec 14 12:00:00 2000 Andrew Makhorin * GLPK 1.1.1 has been released. Minor bug was fixed in `Makefile.in'. GLPK Library Reference was included. Mon Nov 27 12:00:00 2000 Andrew Makhorin * GLPK 1.1 has been released. Minor changes were made in order to co-ordinate GLPK routines and their descriptions. GLPK User's Guide was included. Fri Oct 20 12:00:00 2000 Andrew Makhorin * GLPK 1.0 has been released.