>> -------- Forwarded Message -------- >> From: Yori Zwols <address@hidden> >> To: address@hidden >> Subject: Model causes glpk to hang >> Date: Sun, 17 May 2015 22:43:10 +0100 >> >> The following model causes glpk to hang at the preprocessing stage. To >> reproduce this behavior, save this to a file, say, model.mod, and run: >> glpsol --math model.mod >> >> >> var x1 >= 0, integer; >> var x2 >= 0, integer; >> >> >> subject to c1: x1 - x2 >= 0.9; >> subject to c2: x1 - x2 <= 0.8; >> >> >> I have verified that this is still a bug in GLPK 4.55. >> >> Yori >> >> >>
var x1 >= 0, integer; var x2 >= 0, integer; subject to c1: x1 - x2 >= 0.9; subject to c2: x1 - x2 <= 0.8; Obviously the problem is infeasable. Preprocessing in npp_process_row() does the following: c1 results in an implied lower bound of .9 for x1. which is rounded to 1. Row c2 is activated. c2 results in an implied lower bound of .8 for x2, which is rounded to 1. Row c1 is activated. c1 results in an implied lower bound of 1.9 for x1. which is rounded to 2. Row c2 is activated. c2 results in an implied lower bound of 1.8 for x2, which is rounded to 2. Row c1 is activated. ... Due to rounding c1 and c2 never reaches DBL_MAX so even after 1.7976931348623157e+308 iterations the loop will not exit. I guess the only reasonable possibility to improve the coding here would be observing some time limit. Best regards Heinrich Schuchardt _______________________________________________ Bug-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/bug-glpk
