December 2011 - moab-dev - mailman3.cels.anl.gov

r5283 - in MOAB/trunk: src src/moab test
by iulian＠mcs.anl.gov 29 Dec '11

29 Dec '11

Author: iulian Date: 2011-12-29 10:40:59 -0600 (Thu, 29 Dec 2011) New Revision: 5283 Modified: MOAB/trunk/src/GeomTopoTool.cpp MOAB/trunk/src/moab/GeomTopoTool.hpp MOAB/trunk/test/crop_vol_test.cpp Log: modify the "duplicate_model" method in GeomTopoTool. It can take as input a gentity (surface or volume, usually), to duplicate the dependents of this gentity in a new GeomTopoTool. The new model will have cleaner sense tags when exported to a new database. Before this, variable length edge-face tags would have some values of 0, corresponding to faces that were not extracted (trimmed off) Modified: MOAB/trunk/src/GeomTopoTool.cpp =================================================================== --- MOAB/trunk/src/GeomTopoTool.cpp 2011-12-23 18:56:36 UTC (rev 5282) +++ MOAB/trunk/src/GeomTopoTool.cpp 2011-12-29 16:40:59 UTC (rev 5283) @@ -1118,12 +1118,24 @@ return MB_SUCCESS; } -// this would be a deep copy, into a new geom topo tool - // sets will be duplicated, but entities not - // modelSet will be a new one - // if the original set was null (root), a new model set will be created for - // original model, and its entities will be the original g sets -ErrorCode GeomTopoTool::duplicate_model(GeomTopoTool *& duplicate) + +/* + * This would create a deep copy, into a new geom topo tool + * sets will be duplicated, but entities not + * modelSet will be a new one + * if the original set was null (root), a new model set will be created for + * original model, and its entities will be the original g sets + * Scenario: split a face along a ground line, then write only one surface + * the common line has 2 faces in var len tag for sense edge; if we write only one + * surface to a new database, the var len tag of the edge will be extracted with 2 values, but + * only one value will make sense, the other will be zero. + * + * There is no workaround; we need to create a duplicate model that has only that surface + * and its children (and grand-children). Then the var len sense edge-face tags will have + * the right size. + * + */ +ErrorCode GeomTopoTool::duplicate_model(GeomTopoTool *& duplicate, EntityHandle geomSet) { // will EntityHandle rootModelSet; @@ -1140,6 +1152,19 @@ if (MB_SUCCESS != rval) return rval; } + // extract from the geomSet the dimension, children, and grand-children + Range depSets;// dependents of the geomSet, including the geomSet + // add in this range all the dependents of this, to filter the ones that need to be deep copied + + if (geomSet) + { + rval = mdbImpl->get_child_meshsets(geomSet, depSets, 0); // 0 for numHops means that all + // dependents are returned, not only the direct children. + if (MB_SUCCESS != rval) + return rval; + depSets.insert(geomSet); + }

1 0

r5282 - in MOAB/trunk: . itaps
by tautges＠mcs.anl.gov 23 Dec '11

23 Dec '11

Author: tautges Date: 2011-12-23 12:56:36 -0600 (Fri, 23 Dec 2011) New Revision: 5282 Added: MOAB/trunk/itaps/iBase_f.h.in Modified: MOAB/trunk/configure.ac Log: Adding autoconf logic to set POINTER_SIZE to result of SIZEOF_VOID_P if more proper setting isn't found in CPP logic in iBase_f.h. Passes make check in serial and parallel. Modified: MOAB/trunk/configure.ac =================================================================== --- MOAB/trunk/configure.ac 2011-12-22 18:33:13 UTC (rev 5281) +++ MOAB/trunk/configure.ac 2011-12-23 18:56:36 UTC (rev 5282) @@ -117,6 +117,7 @@ SIZEOF_MBENTITYHANDLE=0 AC_CHECK_SIZEOF([void *]) SIZEOF_VOID_P=$ac_cv_sizeof_void_p +AC_SUBST(SIZEOF_VOID_P) AC_ARG_ENABLE([64bit-handles], [AC_HELP_STRING([--enable-64bit-handles],[Force EntityHandle to be 64 bits])], [AC_CHECK_TYPE([uint64_t], @@ -1144,6 +1145,7 @@ test/io/Makefile test/parallel/Makefile test/oldinc/Makefile + itaps/iBase_f.h itaps/Makefile itaps/imesh/Makefile itaps/imesh/iMesh-Defs.inc Added: MOAB/trunk/itaps/iBase_f.h.in =================================================================== --- MOAB/trunk/itaps/iBase_f.h.in (rev 0) +++ MOAB/trunk/itaps/iBase_f.h.in 2011-12-23 18:56:36 UTC (rev 5282) @@ -0,0 +1,142 @@ +#ifndef IBASE_F_H +#define IBASE_F_H + +#ifdef POINTER_SIZE +! user-defined override +# define IBASE_HANDLE_T INTEGER*POINTER_SIZE +#elif defined(__SIZEOF_SIZE_T__) +! gfortran +# define IBASE_HANDLE_T integer*__SIZEOF_SIZE_T__ +#elif defined(__GNUC__) && defined (_LANGUAGE_FORTRAN) +! gfortran 4.2 and earlier +# ifdef _LP64 +# define IBASE_HANDLE_T integer*8 +# else +! NOTE: not integer*4 (see meaning of _LP64 in GNU docs) +# define IBASE_HANDLE_T integer +# endif +#elif defined(__INTEL_COMPILER) +# ifdef __X86_64 +# define IBASE_HANDLE_T integer*8 +# else +# define IBASE_HANDLE_T integer*4 +# endif +#elif defined(__XLCPP__) || defined(__bgp__) || defined(__bgq__) +! __XLCPP__ doesn't work for IBM. AT least make things work for BGP and BGQ

1 0

Anybody else seeing test failure in mbfacet_test?
by Tim Tautges 23 Dec '11

23 Dec '11

Log is attached... - tim -- ================================================================ "You will keep in perfect peace him whose mind is steadfast, because he trusts in you." Isaiah 26:3 Tim Tautges Argonne National Laboratory (tautges(a)mcs.anl.gov) (telecommuting from UW-Madison) phone (gvoice): (608) 354-1459 1500 Engineering Dr. fax: (608) 263-4499 Madison, WI 53706

1 2

Error reading ABAQUS file
by Rajeev Jain 22 Dec '11

22 Dec '11

The debugger shows it to break at: File: ReadABAQUS.cpp line 177: status = read_assembly(file_set); Attachment: 73.abq This file was exported to ABAQUS file format using CUBIT 12.2. Rajeev

2 1

r5281 - MOAB/trunk/tools/mbzoltan
by hongjun＠mcs.anl.gov 22 Dec '11

22 Dec '11

Author: hongjun Date: 2011-12-22 12:33:13 -0600 (Thu, 22 Dec 2011) New Revision: 5281 Modified: MOAB/trunk/tools/mbzoltan/MBZoltan.cpp MOAB/trunk/tools/mbzoltan/MBZoltan.hpp Log: o Better geometry partition result o Do repartition after round-robin partition for geometry partition o Modified face mesh load estimation o Passes make check Modified: MOAB/trunk/tools/mbzoltan/MBZoltan.cpp =================================================================== --- MOAB/trunk/tools/mbzoltan/MBZoltan.cpp 2011-12-22 15:36:34 UTC (rev 5280) +++ MOAB/trunk/tools/mbzoltan/MBZoltan.cpp 2011-12-22 18:33:13 UTC (rev 5281) @@ -61,6 +61,7 @@ static int *NborProcs=NULL; static double *ObjWeights=NULL; static double *EdgeWeights=NULL; +static int *Parts=NULL; const bool debug = false; @@ -270,7 +271,7 @@ std::vector<double> pts; // x[0], y[0], z[0], ... from MOAB std::vector<int> ids; // point ids from MOAB - std::vector<int> adjs, length; + std::vector<int> adjs, length, parts; std::vector<double> obj_weights, edge_weights; Range elems; #ifdef CGM @@ -318,7 +319,8 @@ else { #ifdef CGM result = assemble_graph(part_dim, pts, ids, adjs, length, obj_weights, - edge_weights, entities, part_geom_mesh_size); RR; + edge_weights, parts, entities, + part_geom_mesh_size, nparts); RR; if (debug) { int n_ids = ids.size(); @@ -327,7 +329,8 @@ for (int i = 0; i < n_ids; i++) { std::cout << "graph_input_ids[" << i << "]=" << ids[i] << ",obj_weights=" << obj_weights[i] - << ",entity_id=" << entities.get_and_step()->id() << std::endl; + << ",entity_id=" << entities.get_and_step()->id() + << ",part=" << parts[i] << std::endl; for (int j = 0; j < length[i]; j++) { std::cout << "adjs[" << j << "]=" << adjs[i_leng] << ",edge_weights=" << edge_weights[i_leng]<< std::endl; @@ -344,7 +347,7 @@ if (edge_weights.size() > 0) e_wgt = &edge_weights[0]; myNumPts = mbInitializePoints((int)ids.size(), &pts[0], &ids[0], &adjs[0], - &length[0], o_wgt, e_wgt); + &length[0], o_wgt, e_wgt, &parts[0]); // Initialize Zoltan. This is a C call. The simple C++ code @@ -421,6 +424,7 @@

1 0

r5280 - in MOAB/trunk: examples/itaps itaps/imesh src src/io src/parallel test/h5file
by tautges＠mcs.anl.gov 22 Dec '11

22 Dec '11

Author: tautges Date: 2011-12-22 09:36:34 -0600 (Thu, 22 Dec 2011) New Revision: 5280 Added: MOAB/trunk/examples/itaps/ListSetsNTagsCXX.cpp MOAB/trunk/examples/itaps/ListSetsNTagsF90.F90 Modified: MOAB/trunk/examples/itaps/Makefile.am MOAB/trunk/itaps/imesh/iMeshP_MOAB.cpp MOAB/trunk/itaps/imesh/iMesh_MOAB.cpp MOAB/trunk/itaps/imesh/iMesh_MOAB.hpp MOAB/trunk/src/Core.cpp MOAB/trunk/src/FileOptions.hpp MOAB/trunk/src/io/ReadHDF5.cpp MOAB/trunk/src/io/ReadHDF5.hpp MOAB/trunk/src/parallel/SharedSetData.cpp MOAB/trunk/test/h5file/h5partial.cpp Log: Fixing bug in parallel reading, where if a set contains only other sets, and the contained sets are read locally (based on containing locally-represented entities), the containing set is also selected for reading. This comes up with NEUMANN sets read from cub files, where those sets contain only the surface geom_topo sets. Fixes problem where Nek wasn't running with READ_PART as the parallel read option. A few other assorted changes: - change to listing of set entities in Core::list_entities - better option processing in iMeshP - added a few more iMesh example codes Passes checks in serial and parallel. Added: MOAB/trunk/examples/itaps/ListSetsNTagsCXX.cpp =================================================================== --- MOAB/trunk/examples/itaps/ListSetsNTagsCXX.cpp (rev 0) +++ MOAB/trunk/examples/itaps/ListSetsNTagsCXX.cpp 2011-12-22 15:36:34 UTC (rev 5280) @@ -0,0 +1,116 @@ +/** ListSetsNTags: list sets & tags from a mesh + * + * This program shows how to read and list sets and tags from a mesh + * + * Usage: SetsNTags <mesh_file_name> + * + */ + +#include <iostream> +#include <cstdlib> +#include <cstring> +#include <vector> +#include "iMesh.h" + +#define ERRORR(a) {if (iBase_SUCCESS != err) {std::cout << a << std::endl; return err;}} + +int main( int argc, char *argv[] ) +{ + const char *filename; + bool read_par = false; + if (argc < 2 || !strcmp(argv[1], "-h")) { + std::cout << "Usage: " << argv[0] << " <filename> [-p]" << std::endl; + return 0; + } + else { + // Check command line arg + filename = argv[1]; + + if (argc > 2 && !strcmp(argv[2], "-p")) read_par = true; + } + + // create the Mesh instance + iMesh_Instance mesh; + int err; + iMesh_newMesh(NULL, &mesh, &err, 0); + ERRORR("Error creating new mesh.\n"); + + + iBase_EntitySetHandle root_set; + iMesh_getRootSet(mesh, &root_set, &err); + ERRORR("Couldn't get root set."); + + // load the mesh + if (read_par) { + const char *opt = " moab:PARALLEL=READ_PART moab:PARTITION=PARALLEL_PARTITION moab:PARTITION_DISTRIBUTE moab:PARALLEL_RESOLVE_SHARED_ENTS moab:DEBUG_PIO=2 moab:DEBUG_IO=2 moab:SETS=SETS "; + iMesh_load(mesh, root_set, filename, opt, &err, strlen(filename), strlen(opt));

1 0

Re: [MOAB-dev] Access to API documentation
by James Porter 21 Dec '11

21 Dec '11

Actually, the original link is right (though I'm not sure what links to it...). The permissions got fouled up somehow. I'm going to keep an eye on this, since it might be an issue with the Buildbot. - Jim On 12/21/11, Rajeev Jain wrote: > > > > Hi Roman, > > > Sorry, we moved this page, it's here: http://gnep.mcs.anl.gov:8010/moab-docs/ > I'll update the website. > > > Rajeev > > > From: Roman Putanowicz <putanowr(a)l5.pk.edu.pl> > > To: > > > moab-dev(a)mcs.anl.gov > > Sent: Wednesday, December 21, 2011 9:56 AM > > Subject: [MOAB-dev] Access to API documentation > > > > > > > Dear All, > > > > I got "Access forbiden" error (403) when trying to > > connecto to MOAB API documentation at > > http://ftp.mcs.anl.gov/pub/fathom/moab-docs/ > > > > Regards > > > > Roman > > > > > > > > > > > > > > >

1 0

r5277 - MOAB/trunk/itaps/imesh
by iulian＠mcs.anl.gov 21 Dec '11

21 Dec '11

Author: iulian Date: 2011-12-21 13:15:50 -0600 (Wed, 21 Dec 2011) New Revision: 5277 Modified: MOAB/trunk/itaps/imesh/testc_cbind.c Log: remove #include <malloc.h> from the test, so more compilers would be happy malloc should be defined in stdlib.h Modified: MOAB/trunk/itaps/imesh/testc_cbind.c =================================================================== --- MOAB/trunk/itaps/imesh/testc_cbind.c 2011-12-21 17:50:24 UTC (rev 5276) +++ MOAB/trunk/itaps/imesh/testc_cbind.c 2011-12-21 19:15:50 UTC (rev 5277) @@ -56,7 +56,6 @@ */ #include <stdio.h> -#include <malloc.h> #include <string.h> #include <stdlib.h> #include "iMesh.h"

1 0

Access to API documentation
by Roman Putanowicz 21 Dec '11

21 Dec '11

Dear All, I got "Access forbiden" error (403) when trying to connecto to MOAB API documentation at http://ftp.mcs.anl.gov/pub/fathom/moab-docs/ Regards Roman

2 1

r5267 - in MOAB/trunk: itaps/igeom src src/moab test
by iulian＠mcs.anl.gov 18 Dec '11

18 Dec '11

Author: iulian Date: 2011-12-18 14:57:17 -0600 (Sun, 18 Dec 2011) New Revision: 5267 Modified: MOAB/trunk/itaps/igeom/FBiGeom_MOAB.cpp MOAB/trunk/src/FBEngine.cpp MOAB/trunk/src/GeomTopoTool.cpp MOAB/trunk/src/moab/FBEngine.hpp MOAB/trunk/src/moab/GeomTopoTool.hpp MOAB/trunk/test/gttool_test.cpp Log: introduce the convention that the moab sets with geometry dimension tag 4 are the sets of other geometry entities. As a reminder, moab sets with geom dimension 0 thru 3 are representing geometry entities for geom topo tool and for igeom implementation in moab (FBiGeom). This change is required to make the method (FB)iGeom_getEntSets() more efficient. We will retrieve only moab sets with geom dimension 4, and filter based on input set. The convention is used already in MeshKit, so it should not come as a big surprise. Modified: MOAB/trunk/itaps/igeom/FBiGeom_MOAB.cpp =================================================================== --- MOAB/trunk/itaps/igeom/FBiGeom_MOAB.cpp 2011-12-16 05:33:22 UTC (rev 5266) +++ MOAB/trunk/itaps/igeom/FBiGeom_MOAB.cpp 2011-12-18 20:57:17 UTC (rev 5267) @@ -5,6 +5,7 @@ #include "moab/OrientedBoxTreeTool.hpp" #include "moab/CartVect.hpp" #include "FileOptions.hpp" +#include "MBTagConventions.hpp" #include <stdlib.h> #include <cstring> #include <map> @@ -1235,12 +1236,32 @@ } void FBiGeom_getNumEntSets(FBiGeom_Instance instance, - iBase_EntitySetHandle entity_set_handle, int num_hops, int *num_sets, + iBase_EntitySetHandle entity_set_handle, int /*num_hops*/, int *num_sets, int *err) { - // TODO here, we should really get only the entity sets that have gents as members - // we have the convention that entity sets of geom dimension 4 - iMesh_getNumEntSets(IMESH_INSTANCE(instance), entity_set_handle, num_hops, - num_sets, err); + // here, we get only the entity sets that have gents as members + // we have the convention that entity sets of geom dimension 4 are + // sets of geo entities; they should contain only gentities as elements (or other sets of gents) + // we should also consider the number of hops + // first, get all sets of geo dim 4 from the entity_set_handle; then intersect with + // the range from geom topo tool + Tag geomTag; + ErrorCode rval = MBI->tag_get_handle(GEOM_DIMENSION_TAG_NAME, 1, MB_TYPE_INTEGER, geomTag); + if (MB_SUCCESS != rval) + RETURN(iBase_FAILURE); + GeomTopoTool * gtt = GETGTT(instance); + const Range * gRange =gtt-> geoRanges(); + // get all sets of geom dimension 4 from the entity set + EntityHandle moabSet = (EntityHandle)entity_set_handle; + const int four = 4; + const void* const four_val[] = { &four }; + Range tmp; + rval = MBI->get_entities_by_type_and_tag(moabSet, MBENTITYSET, &geomTag, + four_val, 1, tmp); + CHKERR(rval,"can't get sets of geo dim 4 "); + tmp=intersect(tmp, gRange[4]); + *num_sets = tmp.size();// ignore, for the time being, number of hops + + RETURN(iBase_SUCCESS); } void FBiGeom_getEntSets(FBiGeom_Instance instance, @@ -1248,11 +1269,30 @@

1 0