1 This is a (partial) reimplementation of latte
2 (http://www.math.ucdavis.edu/~latte/)
3 using polylib (http://icps.u-strasbg.fr/polylib/ or
4 http://www.kotnet.org/~skimo/polylib/).
6 Next to polylib, it uses NTL (http://shoup.net/ntl/)
7 Mac users may want to apply NTL_5_3_2.patch
9 Optionally, PIP (http://www.piplib.org/)
10 and Omega (http://www.cs.umd.edu/projects/omega/ or
11 http://www.liacs.nl/~sverdool/gitweb.cgi?p=omega.git;a=summary)
12 can be used during enumeration of integer projections.
14 Both NTL and polylib need to have been compiled with gmp
16 For NTL this means you have to specify
19 Suppose you want to put everyting in /opt,
20 Then you configure/compile polylib using
22 ./configure --with-libgmp=/path/to/gmp --prefix=/opt
29 ./configure NTL_GMP_LIP=on PREFIX=/opt GMP_PREFIX=/path/to/gmp
35 vi -c '/DESTDIR=/|s/=.*$/=\/opt/|wq' Makefile.config
36 make libomega.a libcode_gen.a
41 ./configure --with-gmp=/path/to/gmp --prefix=/opt
45 Configure this library using
47 ./configure --prefix=/opt --with-polylib=/opt --with-ntl=/opt
51 ./configure --prefix=/opt --with-polylib=/opt --with-ntl=/opt --enable-fractional
59 If you want to put any one of these packages in the default location
60 (typically /usr/local), then the prefix argument can be omitted.
61 For more information on other arguments,
65 Please use a version of the PolyLib library that is not
66 older than August 13th 2004.
68 The input format of barvinok_enumerate is the same as that
69 of testehrhart from the PolyLib distribution.
70 See the PolyLib manual, barvinok_enumerate.c or the
71 examples in tests/ehrhart/ .
73 Note that the fractional representation is subject to change.
75 The library supports three specialization algorithms.
76 The first performs a direct specialization using a random vector.
77 The second performs a "depth-first" incremental specialization.
78 The third performs a "breadth-first" incremental specialization.
80 The first is usually the fastest, but can fail if the random
81 vector happens to be chosen incorrectly.
82 The incremental algorithms should never fail, but can be slower,
83 in some cases by quite a bit. The breadth-first version should
84 be faster than the depth-first version, especially on polytopes
85 of moderate dimension with a lot of "structure", but can require
88 Sven Verdoolaege <skimo@kotnet.org>