July 2009 - PPL-devel - BUGSENG Mailman

Re: [PPL-devel] integer versus rational solutions
by Tobias Grosser 09 Jul '09

09 Jul '09

On Thu, 2009-07-09 at 14:22 +0200, Roberto Bagnara wrote: > Konrad Trifunovic wrote: > > I attach a patch that uses 'integer existence test'. > > I have forced the use of dependence check. > > Hi there. > > I am sorry if the answer is obvious, but I am wondering > why you are using NNC polyhedra (NNC_Polyhedron) instead > of ordinary polyhedra (C_Polyhedron). Do you need to > characterize rational solutions and use strict constraints? No, we do not need this. I introduced NNC_Polyhedron as I did not understand the difference correct. We should switch back to C_Polyhedron or even better to integer(Z) polyhedron. Tobias

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by Michael Classen 09 Jul '09

09 Jul '09

make check failed with one test: tests failed: test06 On Thu, Jul 9, 2009 at 11:52 AM, Michael Classen<michael.classen(a)uni-passau.de> wrote: > Thank you! now it seems to work... :-) > > On Thu, Jul 9, 2009 at 11:24 AM, P M Hill<hill(a)comp.leeds.ac.uk> wrote: >> 2.61 should be ok. You should be running autoreconf and not autoconf. Does >> that work? >> >> On Thu, 9 Jul 2009, Michael Classen wrote: >> >>> Sorry to ask again, but now that I think I have checked out the >>> products branch, I get all sorts of autoconf errors when trying to >>> generate the "configure" script: >>> >>> configure.ac:42: error: possibly undefined macro: AM_INIT_AUTOMAKE >>> If this token and others are legitimate, please use m4_pattern_allow. >>> See the Autoconf documentation. >>> configure.ac:143: error: possibly undefined macro: AM_CONDITIONAL >>> configure.ac:657: error: possibly undefined macro: AC_CHECK_FPU_CONTROL >>> configure.ac:660: error: possibly undefined macro: >>> AC_CXX_PROVIDES_PROPER_LONG_DOUBLE >>> configure.ac:663: error: possibly undefined macro: >>> AC_CXX_FLOAT_BINARY_FORMAT >>> configure.ac:666: error: possibly undefined macro: >>> AC_CXX_DOUBLE_BINARY_FORMAT >>> configure.ac:669: error: possibly undefined macro: >>> AC_CXX_LONG_DOUBLE_BINARY_FORMAT >>> [...] >>> >>> We use autoconf 2.61 here, could that be out of date? >>> >>> Michael >>> >>> On Thu, Jul 9, 2009 at 10:31 AM, P M Hill<hill(a)comp.leeds.ac.uk> wrote: >>>> >>>> I also have the same problem. We'll fix the problem asap but, in the >>>> meantime, the git anonymous access should let you grab a copy. >>>> >>>> Pat >>>> >>>> On Thu, 9 Jul 2009, Michael Classen wrote: >>>> >>>>> Hello Patricia, >>>>> >>>>> thanks for the effort! I just have a simple technical question: I just >>>>> tried to check out the latest snapshot of the "products" branch from >>>>> the GIT web access: >>>>> >>>>> >>>>> http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=shortlog;h=refs/heads… >>>>> >>>>> But it seems that the resulting .tar.gz file is somehow not in any >>>>> format that gzip can handle: >>>>> >>>>> 2677> tar xvzf ppl-a648638a1c5f2b288e03625b5dc07a2bc69d9bd8.tar.gz >>>>> gzip: stdin: not in gzip format >>>>> tar: Child returned status 1 >>>>> tar: Error exit delayed from previous errors >>>>> >>>>> Is there anything obvious wrong with what I'm doing here? I also tried >>>>> a few other snapshots and also some graphical archive program, but to >>>>> the same result... >>>>> >>>>> greetings, >>>>> Michael >>>>> >>>>> On Thu, Jul 9, 2009 at 10:09 AM, P M Hill<hill(a)comp.leeds.ac.uk> wrote: >>>>>> >>>>>> Resending to ppl-devel. >>>>>> >>>>>> ---------- Forwarded message ---------- >>>>>> Date: Wed, 8 Jul 2009 14:24:15 +0100 (BST) >>>>>> From: P M Hill <hill(a)comp.leeds.ac.uk> >>>>>> To: Tobias Grosser <grosser(a)fim.uni-passau.de> >>>>>> Cc: "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com>, >>>>>> ppl-devel <ppl-devel(a)cs.unipr.it> >>>>>> Subject: Re: [PPL-devel] integer versus rational solutions >>>>>> >>>>>> On Wed, 8 Jul 2009, Tobias Grosser wrote: >>>>>> >>>>>>> Hi, >>>>>>> >>>>>>> I just tried to use the ppl to work with integer polyhedron, however I >>>>>>> am stuck with the examples in the documentation for the >>>>>>> Partial_Reduced_Product: >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> http://www.cs.unipr.it/ppl/Documentation/user/ppl-user-0.10.2-html/classPar… >>>>>>> >>>>>>> --------------------------------------------------------------------- >>>>>>> In all the examples it is assumed that the template R is the >>>>>>> No_Reduction<D1, D2> class and that variables x and y are defined >>>>>>> (where >>>>>>> they are used) as follows: >>>>>>> Variable x(0); >>>>>>> Variable y(1); >>>>>>> >>>>>>> Example 1 >>>>>>> The following code builds a direct product of a Grid and NNC >>>>>>> Polyhedron, corresponding to the positive even integer pairs in >>>>>>> R^2, given as a system of congruences: >>>>>>> >>>>>>> Congruence_System cgs; >>>>>>> cgs.insert((x %= 0) / 2); >>>>>>> cgs.insert((y %= 0) / 2); >>>>>>> Partially_Reduced_Product<Grid, NNC_Polyhedron, >>>>>>> No_Reduction<D1, >>>>>>> D2> > >>>>>>> dp(cgs); >>>>>>> dp.add_constraint(x >= 0); >>>>>>> dp.add_constraint(y >= 0); >>>>>>> >>>>>>> Example 2 >>>>>>> The following code builds the same product in R^2: >>>>>>> Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, >>>>>>> D2> >>>>>>>> >>>>>>>> dp(2); >>>>>>> >>>>>>> dp.add_constraint(x >= 0); >>>>>>> dp.add_constraint(y >= 0); >>>>>>> dp.add_congruence((x %= 0) / 2); >>>>>>> dp.add_congruence((y %= 0) / 2); >>>>>>> --------------------------------------------------------------------- >>>>>> >>>>>> Hi Tobias, >>>>>> >>>>>>> >>>>>>> I would like to create the set of even positive integer pairs, so I >>>>>>> tried Example 1 and 2. With both I have the same problems. >>>>>>> >>>>>>> 1. D1 and D2 are not defined: >>>>>>> >>>>>>> test.cpp:9: error: 'D1' was not declared in this scope >>>>>>> test.cpp:9: error: 'D2' was not declared in this scope >>>>>>> >>>>>>> So I replace D1 with "Grid" and D2 with "NNC_Polyhedron" >>>>>> >>>>>> That is correct. They have to be specified. >>>>>> >>>>>>> >>>>>>> 2. An exception when adding constrains: >>>>>> >>>>>> Only equalities can be added to a product of a grid and polyhedron, >>>>>> best to use refine_with-...() methods here. >>>>>> >>>>>>> >>>>>>> terminate called after throwing an instance of 'std::invalid_argument' >>>>>>> what(): PPL::Grid::add_constraint(c): >>>>>>> c is not an equality constraint. >>>>>>> >>>>>>> So I removed the positive constraints. >>>>>>> >>>>>>> 3. An exception when adding congruences: >>>>>>> >>>>>>> terminate called after throwing an instance of 'std::invalid_argument' >>>>>>> what(): PPL::NNC_Polyhedron::add_congruence(cg): >>>>>>> cg is a non-trivial, proper congruence. >>>>>>> >>>>>>> Do you have any idea what I did wrong? >>>>>> >>>>>> Yes, you have to use refiine_with_constraint() >>>>>> and refine_with_congruence(); >>>>>> >>>>>> The add methods cannot be used here unless the constraint/congruence is >>>>>> an >>>>>> equality. >>>>>> >>>>>> There is also another problem in your code in that you cannot build the >>>>>> product directly from the congruences as they cannot be add'ed to the >>>>>> NNC_Polyhedron component. Instead, first build a grid from the >>>>>> congruences >>>>>> and then build the product using the grid. >>>>>> >>>>>> Here is my code, based on your examples, that works here: >>>>>> >>>>>> ------------------------------------------------- >>>>>> >>>>>> /* >>>>>> Example 1 >>>>>> */ >>>>>> >>>>>> Congruence_System cgs; >>>>>> cgs.insert((x %= 0) / 2); >>>>>> cgs.insert((y %= 0) / 2); >>>>>> Grid gr(cgs); >>>>>> Partially_Reduced_Product<Grid, NNC_Polyhedron, >>>>>> No_Reduction<Grid, NNC_Polyhedron> > >>>>>> dp(gr); >>>>>> dp.refine_with_constraint(x >= 0); >>>>>> dp.refine_with_constraint(y >= 0); >>>>>> >>>>>> -------------------------------------------------- >>>>>> >>>>>> /* >>>>>> Example 2 >>>>>> */ >>>>>> >>>>>> Partially_Reduced_Product<Grid, NNC_Polyhedron, >>>>>> No_Reduction<Grid, NNC_Polyhedron> > >>>>>> dp(2); >>>>>> >>>>>> dp.refine_with_constraint(x >= 0); >>>>>> dp.refine_with_constraint(y >= 0); >>>>>> dp.refine_with_congruence((x %= 0) / 2); >>>>>> dp.refine_with_congruence((y %= 0) / 2); >>>>>> >>>>>> -------------------------------------------------- >>>>>> >>>>>> I used here the master branch of the git repository head version. The >>>>>> latest >>>>>> developments for the PPL domain for partially reduced products are in >>>>>> the >>>>>> products branch. >>>>>> >>>>>> HTH. Let me know if you have further queries wrt this domain, we will >>>>>> glad >>>>>> to help. >>>>>> >>>>>> Pat >>>>>> _______________________________________________ >>>>>> PPL-devel mailing list >>>>>> PPL-devel(a)cs.unipr.it >>>>>> http://www.cs.unipr.it/mailman/listinfo/ppl-devel >>>>>> >>>> >> >

2 1

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

I also have the same problem. We'll fix the problem asap but, in the meantime, the git anonymous access should let you grab a copy. Pat On Thu, 9 Jul 2009, Michael Classen wrote: > Hello Patricia, > > thanks for the effort! I just have a simple technical question: I just > tried to check out the latest snapshot of the "products" branch from > the GIT web access: > http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=shortlog;h=refs/heads… > > But it seems that the resulting .tar.gz file is somehow not in any > format that gzip can handle: > > 2677> tar xvzf ppl-a648638a1c5f2b288e03625b5dc07a2bc69d9bd8.tar.gz > gzip: stdin: not in gzip format > tar: Child returned status 1 > tar: Error exit delayed from previous errors > > Is there anything obvious wrong with what I'm doing here? I also tried > a few other snapshots and also some graphical archive program, but to > the same result... > > greetings, > Michael > > On Thu, Jul 9, 2009 at 10:09 AM, P M Hill<hill(a)comp.leeds.ac.uk> wrote: >> Resending to ppl-devel. >> >> ---------- Forwarded message ---------- >> Date: Wed, 8 Jul 2009 14:24:15 +0100 (BST) >> From: P M Hill <hill(a)comp.leeds.ac.uk> >> To: Tobias Grosser <grosser(a)fim.uni-passau.de> >> Cc: "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com>, >> ppl-devel <ppl-devel(a)cs.unipr.it> >> Subject: Re: [PPL-devel] integer versus rational solutions >> >> On Wed, 8 Jul 2009, Tobias Grosser wrote: >> >>> Hi, >>> >>> I just tried to use the ppl to work with integer polyhedron, however I >>> am stuck with the examples in the documentation for the >>> Partial_Reduced_Product: >>> >>> >>> http://www.cs.unipr.it/ppl/Documentation/user/ppl-user-0.10.2-html/classPar… >>> >>> --------------------------------------------------------------------- >>> In all the examples it is assumed that the template R is the >>> No_Reduction<D1, D2> class and that variables x and y are defined (where >>> they are used) as follows: >>> Variable x(0); >>> Variable y(1); >>> >>> Example 1 >>> The following code builds a direct product of a Grid and NNC >>> Polyhedron, corresponding to the positive even integer pairs in >>> R^2, given as a system of congruences: >>> >>> Congruence_System cgs; >>> cgs.insert((x %= 0) / 2); >>> cgs.insert((y %= 0) / 2); >>> Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, >>> D2> > >>> dp(cgs); >>> dp.add_constraint(x >= 0); >>> dp.add_constraint(y >= 0); >>> >>> Example 2 >>> The following code builds the same product in R^2: >>> Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> >>>> dp(2); >>> dp.add_constraint(x >= 0); >>> dp.add_constraint(y >= 0); >>> dp.add_congruence((x %= 0) / 2); >>> dp.add_congruence((y %= 0) / 2); >>> --------------------------------------------------------------------- >> >> Hi Tobias, >> >>> >>> I would like to create the set of even positive integer pairs, so I >>> tried Example 1 and 2. With both I have the same problems. >>> >>> 1. D1 and D2 are not defined: >>> >>> test.cpp:9: error: 'D1' was not declared in this scope >>> test.cpp:9: error: 'D2' was not declared in this scope >>> >>> So I replace D1 with "Grid" and D2 with "NNC_Polyhedron" >> >> That is correct. They have to be specified. >> >>> >>> 2. An exception when adding constrains: >> >> Only equalities can be added to a product of a grid and polyhedron, >> best to use refine_with-...() methods here. >> >>> >>> terminate called after throwing an instance of 'std::invalid_argument' >>> what(): PPL::Grid::add_constraint(c): >>> c is not an equality constraint. >>> >>> So I removed the positive constraints. >>> >>> 3. An exception when adding congruences: >>> >>> terminate called after throwing an instance of 'std::invalid_argument' >>> what(): PPL::NNC_Polyhedron::add_congruence(cg): >>> cg is a non-trivial, proper congruence. >>> >>> Do you have any idea what I did wrong? >> >> Yes, you have to use refiine_with_constraint() >> and refine_with_congruence(); >> >> The add methods cannot be used here unless the constraint/congruence is an >> equality. >> >> There is also another problem in your code in that you cannot build the >> product directly from the congruences as they cannot be add'ed to the >> NNC_Polyhedron component. Instead, first build a grid from the congruences >> and then build the product using the grid. >> >> Here is my code, based on your examples, that works here: >> >> ------------------------------------------------- >> >> /* >> Example 1 >> */ >> >> Congruence_System cgs; >> cgs.insert((x %= 0) / 2); >> cgs.insert((y %= 0) / 2); >> Grid gr(cgs); >> Partially_Reduced_Product<Grid, NNC_Polyhedron, >> No_Reduction<Grid, NNC_Polyhedron> > >> dp(gr); >> dp.refine_with_constraint(x >= 0); >> dp.refine_with_constraint(y >= 0); >> >> -------------------------------------------------- >> >> /* >> Example 2 >> */ >> >> Partially_Reduced_Product<Grid, NNC_Polyhedron, >> No_Reduction<Grid, NNC_Polyhedron> > >> dp(2); >> >> dp.refine_with_constraint(x >= 0); >> dp.refine_with_constraint(y >= 0); >> dp.refine_with_congruence((x %= 0) / 2); >> dp.refine_with_congruence((y %= 0) / 2); >> >> -------------------------------------------------- >> >> I used here the master branch of the git repository head version. The latest >> developments for the PPL domain for partially reduced products are in the >> products branch. >> >> HTH. Let me know if you have further queries wrt this domain, we will glad >> to help. >> >> Pat >> _______________________________________________ >> PPL-devel mailing list >> PPL-devel(a)cs.unipr.it >> http://www.cs.unipr.it/mailman/listinfo/ppl-devel >> >

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

---------- Forwarded message ---------- Date: Thu, 9 Jul 2009 09:17:25 +0100 From: Michael Classen <michael.classen(a)uni-passau.de> To: P M Hill <hill(a)comp.leeds.ac.uk> Subject: Re: [PPL-devel] integer versus rational solutions (fwd) Hello Patricia, thanks for the effort! I just have a simple technical question: I just tried to check out the latest snapshot of the "products" branch from the GIT web access: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=shortlog;h=refs/heads… But it seems that the resulting .tar.gz file is somehow not in any format that gzip can handle: 2677> tar xvzf ppl-a648638a1c5f2b288e03625b5dc07a2bc69d9bd8.tar.gz gzip: stdin: not in gzip format tar: Child returned status 1 tar: Error exit delayed from previous errors Is there anything obvious wrong with what I'm doing here? I also tried a few other snapshots and also some graphical archive program, but to the same result... greetings, Michael On Thu, Jul 9, 2009 at 10:09 AM, P M Hill<hill(a)comp.leeds.ac.uk> wrote: > Resending to ppl-devel. > > ---------- Forwarded message ---------- > Date: Wed, 8 Jul 2009 14:24:15 +0100 (BST) > From: P M Hill <hill(a)comp.leeds.ac.uk> > To: Tobias Grosser <grosser(a)fim.uni-passau.de> > Cc: "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com>, > ppl-devel <ppl-devel(a)cs.unipr.it> > Subject: Re: [PPL-devel] integer versus rational solutions > > On Wed, 8 Jul 2009, Tobias Grosser wrote: > >> Hi, >> >> I just tried to use the ppl to work with integer polyhedron, however I >> am stuck with the examples in the documentation for the >> Partial_Reduced_Product: >> >> >> http://www.cs.unipr.it/ppl/Documentation/user/ppl-user-0.10.2-html/classPar… >> >> --------------------------------------------------------------------- >> In all the examples it is assumed that the template R is the >> No_Reduction<D1, D2> class and that variables x and y are defined (where >> they are used) as follows: >> Variable x(0); >> Variable y(1); >> >> Example 1 >> The following code builds a direct product of a Grid and NNC >> Polyhedron, corresponding to the positive even integer pairs in >> R^2, given as a system of congruences: >> >> Congruence_System cgs; >> cgs.insert((x %= 0) / 2); >> cgs.insert((y %= 0) / 2); >> Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, >> D2> > >> dp(cgs); >> dp.add_constraint(x >= 0); >> dp.add_constraint(y >= 0); >> >> Example 2 >> The following code builds the same product in R^2: >> Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> >>> dp(2); >> dp.add_constraint(x >= 0); >> dp.add_constraint(y >= 0); >> dp.add_congruence((x %= 0) / 2); >> dp.add_congruence((y %= 0) / 2); >> --------------------------------------------------------------------- > > Hi Tobias, > >> >> I would like to create the set of even positive integer pairs, so I >> tried Example 1 and 2. With both I have the same problems. >> >> 1. D1 and D2 are not defined: >> >> test.cpp:9: error: 'D1' was not declared in this scope >> test.cpp:9: error: 'D2' was not declared in this scope >> >> So I replace D1 with "Grid" and D2 with "NNC_Polyhedron" > > That is correct. They have to be specified. > >> >> 2. An exception when adding constrains: > > Only equalities can be added to a product of a grid and polyhedron, > best to use refine_with-...() methods here. > >> >> terminate called after throwing an instance of 'std::invalid_argument' >> what(): PPL::Grid::add_constraint(c): >> c is not an equality constraint. >> >> So I removed the positive constraints. >> >> 3. An exception when adding congruences: >> >> terminate called after throwing an instance of 'std::invalid_argument' >> what(): PPL::NNC_Polyhedron::add_congruence(cg): >> cg is a non-trivial, proper congruence. >> >> Do you have any idea what I did wrong? > > Yes, you have to use refiine_with_constraint() > and refine_with_congruence(); > > The add methods cannot be used here unless the constraint/congruence is an > equality. > > There is also another problem in your code in that you cannot build the > product directly from the congruences as they cannot be add'ed to the > NNC_Polyhedron component. Instead, first build a grid from the congruences > and then build the product using the grid. > > Here is my code, based on your examples, that works here: > > ------------------------------------------------- > > /* > Example 1 > */ > > Congruence_System cgs; > cgs.insert((x %= 0) / 2); > cgs.insert((y %= 0) / 2); > Grid gr(cgs); > Partially_Reduced_Product<Grid, NNC_Polyhedron, > No_Reduction<Grid, NNC_Polyhedron> > > dp(gr); > dp.refine_with_constraint(x >= 0); > dp.refine_with_constraint(y >= 0); > > -------------------------------------------------- > > /* > Example 2 > */ > > Partially_Reduced_Product<Grid, NNC_Polyhedron, > No_Reduction<Grid, NNC_Polyhedron> > > dp(2); > > dp.refine_with_constraint(x >= 0); > dp.refine_with_constraint(y >= 0); > dp.refine_with_congruence((x %= 0) / 2); > dp.refine_with_congruence((y %= 0) / 2); > > -------------------------------------------------- > > I used here the master branch of the git repository head version. The latest > developments for the PPL domain for partially reduced products are in the > products branch. > > HTH. Let me know if you have further queries wrt this domain, we will glad > to help. > > Pat > _______________________________________________ > PPL-devel mailing list > PPL-devel(a)cs.unipr.it > http://www.cs.unipr.it/mailman/listinfo/ppl-devel >

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

Resending to ppl-devel. ---------- Forwarded message ---------- Date: Wed, 8 Jul 2009 14:24:15 +0100 (BST) From: P M Hill <hill(a)comp.leeds.ac.uk> To: Tobias Grosser <grosser(a)fim.uni-passau.de> Cc: "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com>, ppl-devel <ppl-devel(a)cs.unipr.it> Subject: Re: [PPL-devel] integer versus rational solutions On Wed, 8 Jul 2009, Tobias Grosser wrote: > Hi, > > I just tried to use the ppl to work with integer polyhedron, however I > am stuck with the examples in the documentation for the > Partial_Reduced_Product: > > http://www.cs.unipr.it/ppl/Documentation/user/ppl-user-0.10.2-html/classPar… > > --------------------------------------------------------------------- > In all the examples it is assumed that the template R is the > No_Reduction<D1, D2> class and that variables x and y are defined (where > they are used) as follows: > Variable x(0); > Variable y(1); > > Example 1 > The following code builds a direct product of a Grid and NNC > Polyhedron, corresponding to the positive even integer pairs in > R^2, given as a system of congruences: > > Congruence_System cgs; > cgs.insert((x %= 0) / 2); > cgs.insert((y %= 0) / 2); > Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> > > > dp(cgs); > dp.add_constraint(x >= 0); > dp.add_constraint(y >= 0); > > Example 2 > The following code builds the same product in R^2: > Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> > > dp(2); > dp.add_constraint(x >= 0); > dp.add_constraint(y >= 0); > dp.add_congruence((x %= 0) / 2); > dp.add_congruence((y %= 0) / 2); > --------------------------------------------------------------------- Hi Tobias, > > I would like to create the set of even positive integer pairs, so I > tried Example 1 and 2. With both I have the same problems. > > 1. D1 and D2 are not defined: > > test.cpp:9: error: 'D1' was not declared in this scope > test.cpp:9: error: 'D2' was not declared in this scope > > So I replace D1 with "Grid" and D2 with "NNC_Polyhedron" That is correct. They have to be specified. > > 2. An exception when adding constrains: Only equalities can be added to a product of a grid and polyhedron, best to use refine_with-...() methods here. > > terminate called after throwing an instance of 'std::invalid_argument' > what(): PPL::Grid::add_constraint(c): > c is not an equality constraint. > > So I removed the positive constraints. > > 3. An exception when adding congruences: > > terminate called after throwing an instance of 'std::invalid_argument' > what(): PPL::NNC_Polyhedron::add_congruence(cg): > cg is a non-trivial, proper congruence. > > Do you have any idea what I did wrong? Yes, you have to use refiine_with_constraint() and refine_with_congruence(); The add methods cannot be used here unless the constraint/congruence is an equality. There is also another problem in your code in that you cannot build the product directly from the congruences as they cannot be add'ed to the NNC_Polyhedron component. Instead, first build a grid from the congruences and then build the product using the grid. Here is my code, based on your examples, that works here: ------------------------------------------------- /* Example 1 */ Congruence_System cgs; cgs.insert((x %= 0) / 2); cgs.insert((y %= 0) / 2); Grid gr(cgs); Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<Grid, NNC_Polyhedron> > dp(gr); dp.refine_with_constraint(x >= 0); dp.refine_with_constraint(y >= 0); -------------------------------------------------- /* Example 2 */ Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<Grid, NNC_Polyhedron> > dp(2); dp.refine_with_constraint(x >= 0); dp.refine_with_constraint(y >= 0); dp.refine_with_congruence((x %= 0) / 2); dp.refine_with_congruence((y %= 0) / 2); -------------------------------------------------- I used here the master branch of the git repository head version. The latest developments for the PPL domain for partially reduced products are in the products branch. HTH. Let me know if you have further queries wrt this domain, we will glad to help. Pat

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

Resending to ppl-devel. ---------- Forwarded message ---------- Date: Wed, 8 Jul 2009 20:52:13 +0100 (BST) From: P M Hill <hill(a)comp.leeds.ac.uk> To: Michael Classen <michael.classen(a)uni-passau.de> Cc: ppl-devel(a)cs.unipr.it, "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com> Subject: Re: [PPL-devel] integer versus rational solutions On Wed, 8 Jul 2009, Michael Classen wrote: >> HTH. Let me know if you have further queries wrt this domain, we will glad >> to help. >> >> Pat > > Hi Pat, > > I basically just want to know if I can get correct integer results > when using this combination of datatypes. I want to use operations > like union, intersection, projection, basically most standard > operations you want to use on polytopes. Yes. These are already available in the latest release, but the best version of the product domain is in the products branch of the GIT repository. I would recommend you to use that if possible. Sorry, I did not reply sooner to your previous email. I did try and see what could be done to help solve the problem you described. (i.e., transforming a grid x polyhedron product to one where the grid is the integer lattice). In fact, I have a proposal for adding a method to the product domain that I hope would be sufficient for what you need while, from the point of view of the PPL, fits into the existing structures. In particular, I am believe that the implementation work would be small! That is: In the product domain, (assuming for this explanation that the first domain is a grid and the second a C polyhedron) there would be a method such as: bool Partially_Reduced_Product<Grid, C_Polyhedron, R> ::affine_lattice_transform(const Grid& gr1) that assigns to *this = <gr, ph> the product <gr1, ph1> such that there is an affine function (ie a sequence of affine image mappings) T, T(gr) = gr1 and T(ph) = ph1. If there is no invertible affine function T such that T(gr) = gr1, then the method could return false and otherwise true. If you call the method with gr1 = integral lattice, and gr as a full dimensional discrete grid you will get the polyhedron transformed to one where the integral points correspond to the grid points and you can use whatever tools you like to count the number of integer points. By calling the same method with this* = <gr1, ph1> with the argument gr, you will be able to invert the operation. I can try and implement something like this next week (while away at a conference) as I think this would be useful anyway for other applications. Best wishes, Pat

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

Resending to ppl-devel. ---------- Forwarded message ---------- Date: Wed, 8 Jul 2009 14:24:15 +0100 (BST) From: P M Hill <hill(a)comp.leeds.ac.uk> To: Tobias Grosser <grosser(a)fim.uni-passau.de> Cc: "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com>, ppl-devel <ppl-devel(a)cs.unipr.it> Subject: Re: [PPL-devel] integer versus rational solutions On Wed, 8 Jul 2009, Tobias Grosser wrote: > Hi, > > I just tried to use the ppl to work with integer polyhedron, however I > am stuck with the examples in the documentation for the > Partial_Reduced_Product: > > http://www.cs.unipr.it/ppl/Documentation/user/ppl-user-0.10.2-html/classPar… > > --------------------------------------------------------------------- > In all the examples it is assumed that the template R is the > No_Reduction<D1, D2> class and that variables x and y are defined (where > they are used) as follows: > Variable x(0); > Variable y(1); > > Example 1 > The following code builds a direct product of a Grid and NNC > Polyhedron, corresponding to the positive even integer pairs in > R^2, given as a system of congruences: > > Congruence_System cgs; > cgs.insert((x %= 0) / 2); > cgs.insert((y %= 0) / 2); > Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> > > > dp(cgs); > dp.add_constraint(x >= 0); > dp.add_constraint(y >= 0); > > Example 2 > The following code builds the same product in R^2: > Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<D1, D2> > > dp(2); > dp.add_constraint(x >= 0); > dp.add_constraint(y >= 0); > dp.add_congruence((x %= 0) / 2); > dp.add_congruence((y %= 0) / 2); > --------------------------------------------------------------------- Hi Tobias, > > I would like to create the set of even positive integer pairs, so I > tried Example 1 and 2. With both I have the same problems. > > 1. D1 and D2 are not defined: > > test.cpp:9: error: 'D1' was not declared in this scope > test.cpp:9: error: 'D2' was not declared in this scope > > So I replace D1 with "Grid" and D2 with "NNC_Polyhedron" That is correct. They have to be specified. > > 2. An exception when adding constrains: Only equalities can be added to a product of a grid and polyhedron, best to use refine_with-...() methods here. > > terminate called after throwing an instance of 'std::invalid_argument' > what(): PPL::Grid::add_constraint(c): > c is not an equality constraint. > > So I removed the positive constraints. > > 3. An exception when adding congruences: > > terminate called after throwing an instance of 'std::invalid_argument' > what(): PPL::NNC_Polyhedron::add_congruence(cg): > cg is a non-trivial, proper congruence. > > Do you have any idea what I did wrong? Yes, you have to use refiine_with_constraint() and refine_with_congruence(); The add methods cannot be used here unless the constraint/congruence is an equality. There is also another problem in your code in that you cannot build the product directly from the congruences as they cannot be add'ed to the NNC_Polyhedron component. Instead, first build a grid from the congruences and then build the product using the grid. Here is my code, based on your examples, that works here: ------------------------------------------------- /* Example 1 */ Congruence_System cgs; cgs.insert((x %= 0) / 2); cgs.insert((y %= 0) / 2); Grid gr(cgs); Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<Grid, NNC_Polyhedron> > dp(gr); dp.refine_with_constraint(x >= 0); dp.refine_with_constraint(y >= 0); -------------------------------------------------- /* Example 2 */ Partially_Reduced_Product<Grid, NNC_Polyhedron, No_Reduction<Grid, NNC_Polyhedron> > dp(2); dp.refine_with_constraint(x >= 0); dp.refine_with_constraint(y >= 0); dp.refine_with_congruence((x %= 0) / 2); dp.refine_with_congruence((y %= 0) / 2); -------------------------------------------------- I used here the master branch of the git repository head version. The latest developments for the PPL domain for partially reduced products are in the products branch. HTH. Let me know if you have further queries wrt this domain, we will glad to help. Pat

1 0

Re: [PPL-devel] integer versus rational solutions (fwd)
by P M Hill 09 Jul '09

09 Jul '09

Resending to ppl-devel. ---------- Forwarded message ---------- Date: Wed, 8 Jul 2009 20:52:13 +0100 (BST) From: P M Hill <hill(a)comp.leeds.ac.uk> To: Michael Classen <michael.classen(a)uni-passau.de> Cc: ppl-devel(a)cs.unipr.it, "gcc-graphite(a)googlegroups.com" <gcc-graphite(a)googlegroups.com> Subject: Re: [PPL-devel] integer versus rational solutions On Wed, 8 Jul 2009, Michael Classen wrote: >> HTH. Let me know if you have further queries wrt this domain, we will glad >> to help. >> >> Pat > > Hi Pat, > > I basically just want to know if I can get correct integer results > when using this combination of datatypes. I want to use operations > like union, intersection, projection, basically most standard > operations you want to use on polytopes. Yes. These are already available in the latest release, but the best version of the product domain is in the products branch of the GIT repository. I would recommend you to use that if possible. Sorry, I did not reply sooner to your previous email. I did try and see what could be done to help solve the problem you described. (i.e., transforming a grid x polyhedron product to one where the grid is the integer lattice). In fact, I have a proposal for adding a method to the product domain that I hope would be sufficient for what you need while, from the point of view of the PPL, fits into the existing structures. In particular, I am believe that the implementation work would be small! That is: In the product domain, (assuming for this explanation that the first domain is a grid and the second a C polyhedron) there would be a method such as: bool Partially_Reduced_Product<Grid, C_Polyhedron, R> ::affine_lattice_transform(const Grid& gr1) that assigns to *this = <gr, ph> the product <gr1, ph1> such that there is an affine function (ie a sequence of affine image mappings) T, T(gr) = gr1 and T(ph) = ph1. If there is no invertible affine function T such that T(gr) = gr1, then the method could return false and otherwise true. If you call the method with gr1 = integral lattice, and gr as a full dimensional discrete grid you will get the polyhedron transformed to one where the integral points correspond to the grid points and you can use whatever tools you like to count the number of integer points. By calling the same method with this* = <gr1, ph1> with the argument gr, you will be able to invert the operation. I can try and implement something like this next week (while away at a conference) as I think this would be useful anyway for other applications. Best wishes, Pat

1 0

Re: [PPL-devel] integer versus rational solutions
by P M Hill 08 Jul '09

08 Jul '09

On Wed, 8 Jul 2009, Michael Classen wrote: >> HTH. Let me know if you have further queries wrt this domain, we will glad >> to help. >> >> Pat > > Hi Pat, > > I basically just want to know if I can get correct integer results > when using this combination of datatypes. I want to use operations > like union, intersection, projection, basically most standard > operations you want to use on polytopes. Yes. These are already available in the latest release, but the best version of the product domain is in the products branch of the GIT repository. I would recommend you to use that if possible. Sorry, I did not reply sooner to your previous email. I did try and see what could be done to help solve the problem you described. (i.e., transforming a grid x polyhedron product to one where the grid is the integer lattice). In fact, I have a proposal for adding a method to the product domain that I hope would be sufficient for what you need while, from the point of view of the PPL, fits into the existing structures. In particular, I am believe that the implementation work would be small! That is: In the product domain, (assuming for this explanation that the first domain is a grid and the second a C polyhedron) there would be a method such as: bool Partially_Reduced_Product<Grid, C_Polyhedron, R> ::affine_lattice_transform(const Grid& gr1) that assigns to *this = <gr, ph> the product <gr1, ph1> such that there is an affine function (ie a sequence of affine image mappings) T, T(gr) = gr1 and T(ph) = ph1. If there is no invertible affine function T such that T(gr) = gr1, then the method could return false and otherwise true. If you call the method with gr1 = integral lattice, and gr as a full dimensional discrete grid you will get the polyhedron transformed to one where the integral points correspond to the grid points and you can use whatever tools you like to count the number of integer points. By calling the same method with this* = <gr1, ph1> with the argument gr, you will be able to invert the operation. I can try and implement something like this next week (while away at a conference) as I think this would be useful anyway for other applications. Best wishes, Pat

1 0

Re: [PPL-devel] Graphite build fails if PPL configured with --disable-shared
by Roberto Bagnara 08 Jul '09

08 Jul '09

Janis Johnson wrote: > On Mon, 2009-05-11 at 13:07 -0700, Ian Lance Taylor wrote: >> Another Graphite build issue: it appears that I must not use >> --disable-shared when I configure PPL. If I do use --disable-shared, I >> get this: >> >> /home/iant/gnu/ppl-0.10.2-install/lib/libppl_c.a(ppl_c_implementation_common.o): In function `finalize': >> /home/iant/gnu/ppl-0.10.2/interfaces/C/../../src/ppl.hh:1842: undefined reference to `operator delete(void*)' >> >> followed by thousands of similar errors. This is unfortunate, as it >> means that I must manually set LD_LIBRARY_PATH to the directory where >> the PPL library is installed. This also makes it harder for anybody >> else to run the compiler that I build. This needs to be fixed. > > I get around this by setting LDFLAGS for the ppl configure: > > LDFLAGS="-static" \ > ./configure \ > --prefix=$PREFIX \ > --build=powerpc-linux \ > --with-gnu-ld \ > --with-libgmp-prefix=$PREFIX \ > --with-libgmpxx-prefix=$PREFIX \ > --disable-shared I am not sure I understand: we trust that Libtool, which provides us with the --disable-shared option, will do the right thing. And it seems it does here: the static library is built and passes its checks. Perhaps you want something different from what --disable-shared promises, that is, not to build any shared libraries? > I copy libstdc++.a into the directory with the other GCC host > libraries (gmp/mpfr/ppl/cloog/mpc). > > Building these libraries is indeed quite painful. Any suggestion about how to improve the PPL is welcome. This, of course, applies also to the build machinery. All the best, Roberto -- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

6 8