From ac621cd41ddb1cfc87bbc5078e1c1129eac93c01 Mon Sep 17 00:00:00 2001 From: Alinson Xavier Date: Wed, 26 Jul 2017 18:47:03 -0400 Subject: [PATCH] Update README --- lifting/README.md | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/lifting/README.md b/lifting/README.md index a146a63..ada0352 100644 --- a/lifting/README.md +++ b/lifting/README.md @@ -12,9 +12,9 @@ To produce the tables in the paper, the following tools and libraries were used. Different versions may produce slightly different outputs. - GNU Make, version 3.81 -- CMake, version 3.5.1 -- GCC, the GNU Compiler Collection, version 4.8.4 -- Ruby, version 1.9.3 +- CMake, version 3.7.2 +- GCC, the GNU Compiler Collection, version 6.3.0 +- Ruby, version 2.4.0 - IBM® ILOG® CPLEX®, version 12.6 Build instructions @@ -35,16 +35,18 @@ Running the experiments 3. Two CSV files will be generated inside the folder `lifting/benchmark/tables`, corresponding to the two tables that appear in the paper. +Modifying the instances +----------------------- + In order to run the experiments with a different set of instances, the file `lifting/benchmark/instances/filtered/all.txt` should be modified. Each line in this file describes the origin `f` and a lattice-free set `B`. -The set `B` is described by the coordinates of its vertices. Since the generator -only deals with maximal lattice-free sets, it is also necessary to specify the -lattice-point in the relative interior of each facet of `B`. The format of the file -is as follows: +The set `B` is described by the coordinates of its vertices. Since the +benchmark code only deals with maximal lattice-free sets, it is also necessary +to specify the lattice-points that belong to each facet of `B`. +If `n` is the number of facets, `v` is an n-by-2 matrix of doubles corresponding +to the vertices and l is an n-by-2 matrix of doubles corresponding to the +lattice-points, then each line of the file should be written as f[0] f[1] n v[0][0] v[0][1] ... v[n-1][0] v[n-1][1] n l[0][0] l[0][1] ... l[n-1][0] l[n-1][1] -where `n` is the number of facets, `v` is a n-by-2 matrix of doubles -corresponding to the vertices and l is an n-by-2 matrix of doubles -corresponding to the lattice-points.