tpmt broken by OpticksCSG enum move¶

shape/operator enum unification to use sysrap/OpticksCSG.{h,py} is incomplete
tpmt broken due to mis-interpretation of part buffer

Vague recollections¶

GCSG not used for raytrace ? It exists just to allow Geant4 CSG re-creation of geometry ?

2017-10-20 tpmt– broken (again) : QUALITATIVELY FIXED¶

Following some effort to clean up GGeoTest::makePmtInBox still finding failure symptom that the raytrace of the PMT is missing.
Also wierd container box distortion, possible bad bbox ? This turns out to be too tight a far causing clipping.
Found the default apmtidx of 0 loads analytic PMT with old enum codes, created a new one in slot 2.

TODO: commit fixed analytic PMT into opticksdata slot 0 once this issue is resolved
Found that the GParts from the loaded PMT was not getting combined with the container. After fixing that get OptiX hard crash at launch, forcing reboot. To just do init and avoid crash do:
```
tpmt-;tpmt-- -PV   ## --nopropagate --noviz
```
Cause of crash was omitting GParts::setPartList for the combi GParts, so a PartList was being interpreted as a NodeTree.
Now get a raytrace, but with a z-slipped dynode : two relative positions wrong. Fixed : posXYZ applied at physvol level were not getting linked thru to the primitives.

Slice testing to isolate z-offset part¶

Formation 4-4-2-1-1-(1) slicing prim by prim

tpmt-;tpmt-- --apmtslice 0:4   (4)
tpmt-;tpmt-- --apmtslice 4:8   (4)
    # expected PMT shape
    # near scan cutting into cylinder : not dark on inside ?

tpmt-;tpmt-- --apmtslice 3:4
    # just the tubs : open shape (!) bright on inside ?
    # TODO: use CSG tubs with partlist ?

tpmt-;tpmt-- --apmtslice 8:10   (2)
    # just cathode
    # GPmtTest --apmtidx 2 --apmtslice 8:10

tpmt-;tpmt-- --apmtslice 10:11  (1)
    # z-shifted
    # hemisphere missing transform
    # GPmtTest --apmtidx 2 --apmtslice 10:11

tpmt-;tpmt-- --apmtslice 11:12  (1)
    # z-shifted
    # tubs missing transform : its looks symetrical in z
    # GPmtTest --apmtidx 2 --apmtslice 11:12

GPmtTest --apmtidx 2 --apmtslice 10:11

2017-10-21 18:23:51.168 INFO  [64542] [GParts::dump@1380] GParts::dump ni 1 lim 20 ulim 1
     0.0000      0.0000      0.0000     98.0000
     0.0000      0.0000     123 <-bnd        0 <-INDEX    bn OpaqueVacuum///Vacuum
   -97.1514    -97.1514    -98.0000           5 (sphere) TYPECODE
    97.1514     97.1514    -12.8687           3 (nodeIndex)

GPmtTest --apmtidx 2 --apmtslice 11:12
...

2017-10-21 18:18:34.213 INFO  [63388] [GParts::dump@1380] GParts::dump ni 1 lim 20 ulim 1
     0.0000      0.0000      0.0000     27.5000
   166.0000      0.0000      30 <-bnd        0 <-INDEX    bn Vacuum///OpaqueVacuum
   -27.5000    -27.5000    -83.0000          11 (tubs) TYPECODE
    27.5000     27.5000     83.0000           4 (nodeIndex)

FIXED : two physvol offsets not being honoured¶

105
106     <physvol name="pvPmtHemiBottom"
107          logvol="/dd/Geometry/PMT/lvPmtHemiBottom">
108       <posXYZ z="PmtHemiFaceOff+PmtHemiBellyOff"/>
109     </physvol>
110
111     <physvol name="pvPmtHemiDynode"
112          logvol="/dd/Geometry/PMT/lvPmtHemiDynode">
113       <posXYZ z="-0.5*PmtHemiGlassBaseLength+PmtHemiGlassThickness"/>
114     </physvol>

GPU dumping : note parallelised shuffle¶

suspect the part to prim assoc for primIdx 0 wrong...

## intersect_analytic.cu:bounds buffer sizes pts:  13 pln:   0 trs:   0
// primIdx: 0  p: 0 bnd: 27 typ:  5 pt.q2.f (  -101.1682  -101.1682   -23.8382     0.0000 ) pt.q3.f (   101.1682   101.1682    56.0000     0.0000 )
// primIdx: 1  p: 0 bnd:124 typ:  5 pt.q2.f (   -98.1428   -98.1428    56.0000     0.0000 ) pt.q3.f (    98.1428    98.1428    98.0465     0.0000 )
// primIdx: 2  p: 0 bnd:125 typ:  5 pt.q2.f (   -98.0932   -98.0932    55.9934     0.0000 ) pt.q3.f (    98.0932    98.0932    98.0128     0.0000 )
// primIdx: 3  p: 0 bnd: 30 typ: 11 pt.q2.f (   -27.5000   -27.5000   -83.0000     0.0000 ) pt.q3.f (    27.5000    27.5000    83.0000     0.0000 )
// primIdx: 4  p: 0 bnd:127 typ:  6 pt.q2.f (  -400.0100  -400.0100  -400.0100     0.0000 ) pt.q3.f (   400.0100   400.0100   400.0100     0.0000 )
// primIdx: 0  p: 1 bnd: 27 typ:  5 pt.q2.f (  -101.1682  -101.1682    56.0000     0.0000 ) pt.q3.f (   101.1682   101.1682   100.0698     0.0000 )
// primIdx: 1  p: 1 bnd:124 typ:  5 pt.q2.f (   -82.2854   -82.2854    98.0465     0.0000 ) pt.q3.f (    82.2854    82.2854   128.0000     0.0000 )
// primIdx: 2  p: 1 bnd:126 typ:  5 pt.q2.f (   -97.1514   -97.1514   -98.0000     0.0000 ) pt.q3.f (    97.1514    97.1514   -12.8687     0.0000 )
// primIdx: 0  p: 2 bnd: 27 typ:  5 pt.q2.f (   -84.5402   -84.5402   100.0698     0.0000 ) pt.q3.f (    84.5402    84.5402   131.0000     0.0000 )
// primIdx: 1  p: 2 bnd: 28 typ: 11 pt.q2.f (   -39.2500   -39.2500  -164.5000     0.0000 ) pt.q3.f (    39.2500    39.2500   -21.8869     0.0000 )
// primIdx: 0  p: 3 bnd: 27 typ: 11 pt.q2.f (   -42.2500   -42.2500  -169.0000     0.0000 ) pt.q3.f (    42.2500    42.2500   -23.8382     0.0000 )
// primIdx: 1  p: 3 bnd:125 typ:  5 pt.q2.f (   -82.2478   -82.2478    98.0128     0.0000 ) pt.q3.f (    82.2478    82.2478   127.9500     0.0000 )
// primIdx: 0  p: 4 bnd:123 typ:  5 pt.q2.f (   -98.1428   -98.1428   -21.8869     0.0000 ) pt.q3.f (    98.1428    98.1428    56.0000     0.0000 )


// manually unshuffled

## intersect_analytic.cu:bounds buffer sizes pts:  13 pln:   0 trs:   0
// primIdx: 0  p: 0 bnd: 27 typ:  5 pt.q2.f (  -101.1682  -101.1682   -23.8382     0.0000 ) pt.q3.f (   101.1682   101.1682    56.0000     0.0000 )
// primIdx: 0  p: 1 bnd: 27 typ:  5 pt.q2.f (  -101.1682  -101.1682    56.0000     0.0000 ) pt.q3.f (   101.1682   101.1682   100.0698     0.0000 )
// primIdx: 0  p: 2 bnd: 27 typ:  5 pt.q2.f (   -84.5402   -84.5402   100.0698     0.0000 ) pt.q3.f (    84.5402    84.5402   131.0000     0.0000 )
// primIdx: 0  p: 3 bnd: 27 typ: 11 pt.q2.f (   -42.2500   -42.2500  -169.0000     0.0000 ) pt.q3.f (    42.2500    42.2500   -23.8382     0.0000 )

// primIdx: 0  p: 4 bnd:123 typ:  5 pt.q2.f (   -98.1428   -98.1428   -21.8869     0.0000 ) pt.q3.f (    98.1428    98.1428    56.0000     0.0000 )
^^^^^^^^^^

// primIdx: 1  p: 0 bnd:124 typ:  5 pt.q2.f (   -98.1428   -98.1428    56.0000     0.0000 ) pt.q3.f (    98.1428    98.1428    98.0465     0.0000 )
// primIdx: 1  p: 1 bnd:124 typ:  5 pt.q2.f (   -82.2854   -82.2854    98.0465     0.0000 ) pt.q3.f (    82.2854    82.2854   128.0000     0.0000 )
// primIdx: 1  p: 2 bnd: 28 typ: 11 pt.q2.f (   -39.2500   -39.2500  -164.5000     0.0000 ) pt.q3.f (    39.2500    39.2500   -21.8869     0.0000 )
// primIdx: 1  p: 3 bnd:125 typ:  5 pt.q2.f (   -82.2478   -82.2478    98.0128     0.0000 ) pt.q3.f (    82.2478    82.2478   127.9500     0.0000 )

// primIdx: 2  p: 0 bnd:125 typ:  5 pt.q2.f (   -98.0932   -98.0932    55.9934     0.0000 ) pt.q3.f (    98.0932    98.0932    98.0128     0.0000 )
// primIdx: 2  p: 1 bnd:126 typ:  5 pt.q2.f (   -97.1514   -97.1514   -98.0000     0.0000 ) pt.q3.f (    97.1514    97.1514   -12.8687     0.0000 )

// primIdx: 3  p: 0 bnd: 30 typ: 11 pt.q2.f (   -27.5000   -27.5000   -83.0000     0.0000 ) pt.q3.f (    27.5000    27.5000    83.0000     0.0000 )

// primIdx: 4  p: 0 bnd:127 typ:  6 pt.q2.f (  -400.0100  -400.0100  -400.0100     0.0000 ) pt.q3.f (   400.0100   400.0100   400.0100     0.0000 )



// intersect_analytic.cu:bounds primIdx 0 primFlag 100 partOffset   0 numParts   5  min  -101.1682  -101.1682  -169.0000 max   101.1682   101.1682   131.0000
// intersect_analytic.cu:bounds primIdx 1 primFlag 100 partOffset   5 numParts   4  min   -98.1428   -98.1428  -164.5000 max    98.1428    98.1428   128.0000
// intersect_analytic.cu:bounds primIdx 2 primFlag 100 partOffset   9 numParts   2  min   -98.0932   -98.0932   -98.0000 max    98.0932    98.0932    98.0128
// intersect_analytic.cu:bounds primIdx 3 primFlag 100 partOffset  11 numParts   1  min   -27.5000   -27.5000   -83.0000 max    27.5000    27.5000    83.0000
// intersect_analytic.cu:bounds primIdx 4 primFlag 100 partOffset  12 numParts   1  min  -400.0100  -400.0100  -400.0100 max   400.0100   400.0100   400.0100

Expecting 4 not 5 parts for 1st:

2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1135] OGeo::makeAnalyticGeometry --dbganalytic (part_offset, parts_for_prim, tran_offset, plan_offset)  numPrim: 5 ulim: 5
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1146]  (   0   -5    0    0)
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1146]  (   5   -4    0    0)
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1146]  (   9   -2    0    0)
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1146]  (  11   -1    0    0)
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dumpPrimInfo@1146]  (  12   -1    0    0)
2017-10-21 16:39:52.022 INFO  [34449] [GParts::dump@1380] GParts::dump ni 13 lim 10 ulim 10

GPmtTest --apmtidx 2   //  4-4-2-1-1 ... so combination is incorrectly adding to 1st prim

Yep fixed by setNodeIndex for the container GParts.

old overview¶

DONE: old PMT serialization needs to be rebuilt with new unified enum
rebuilt analytic PMT and stored into opticksdata with non-default apmtidx slot 1 (not committed)

2017-10-20 issue : boundaries not getting into the GParts ?¶

seems the bndspec is OK, but this is not being treated as the input ? Instead the bnd in the .npy which are all zero is the input.
- just need to GParts::close in order to registerBoundaries

simon:opticks blyth$ cat /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/GPmt_boundaries.txt
CONTAINING_MATERIAL///Pyrex
CONTAINING_MATERIAL///Pyrex
CONTAINING_MATERIAL///Pyrex
CONTAINING_MATERIAL///Pyrex
Pyrex///OpaqueVacuum
Pyrex/SENSOR_SURFACE//Bialkali
Pyrex/SENSOR_SURFACE//Bialkali
Pyrex///Vacuum
Bialkali///Vacuum
Bialkali///Vacuum
OpaqueVacuum///Vacuum
Vacuum///OpaqueVacuum

tpmt-;tpmt-- -PV    # just init for debug

2017-10-20 18:22:21.843 INFO  [627696] [GGeoTest::createPmtInBox@293] GGeoTest::createPmtInBox  spec Rock/NONE/perfectAbsorbSurface/MineralOil container_inner_material MineralOil
2017-10-20 18:22:21.845 INFO  [627696] [GPmt::dump@167] GGeoTest::loadPmt (GPmt)pmt --dbganalytic  m_index 0 m_path /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0 m_parts 0x7f98c5ccc180 m_csg 0x7f98c5ccb990 m_bndlib 0x7f98c3e049d0
2017-10-20 18:22:21.845 INFO  [627696] [GParts::Summary@1120] GGeoTest::loadPmt (GParts)pts --dbganalytic  num_parts 12 num_prim 0
 part  0 : node  0 type  1 boundary [  0] Vacuum///Vacuum
 part  1 : node  0 type  1 boundary [  0] Vacuum///Vacuum
 part  2 : node  0 type  1 boundary [  0] Vacuum///Vacuum
 part  3 : node  0 type  2 boundary [  0] Vacuum///Vacuum
 part  4 : node  1 type  1 boundary [  0] Vacuum///Vacuum
 part  5 : node  1 type  1 boundary [  0] Vacuum///Vacuum
 part  6 : node  1 type  1 boundary [  0] Vacuum///Vacuum
 part  7 : node  1 type  2 boundary [  0] Vacuum///Vacuum
 part  8 : node  2 type  1 boundary [  0] Vacuum///Vacuum
 part  9 : node  2 type  1 boundary [  0] Vacuum///Vacuum
 part 10 : node  3 type  1 boundary [  0] Vacuum///Vacuum
 part 11 : node  4 type  2 boundary [  0] Vacuum///Vacuum
2017-10-20 18:22:21.845 INFO  [627696] [*GMergedMesh::combine@138] GMergedMesh::combine making new mesh  index 0 solids 1 verbosity 3
2017-10-20 18:22:21.845 INFO  [627696] [GSolid::Dump@204] GMergedMesh::combine (source solids) numSolid 1


GPmtTest   # shows same issue ...

Review NCSG::Deserialize boundary handling¶

In tboolean- the boundary strings are planted in the python, which get serialized into the csg.txt

cat /tmp/blyth/opticks/tboolean-torus--/csg.txt
Rock//perfectAbsorbSurface/Vacuum
Vacuum///GlassSchottF2

each NCSG tree has only a single boundary spec string which gets set in NCSG::Deserialize

1153 int NCSG::Deserialize(const char* basedir, std::vector<NCSG*>& trees, int verbosity )
1154 {
....
1157     std::string txtpath = BFile::FormPath(basedir, FILENAME) ;
....
1166     NTxt bnd(txtpath.c_str());
1167     bnd.read();
1169
1170     unsigned nbnd = bnd.getNumLines();
....
1181     // order is reversed so that a tree with the "container" meta data tag at tree slot 0
1182     // is handled last, so container_bb will then have been adjusted to hold all the others...
1183     // allowing the auto-bbox setting of the container
1184
1185     for(unsigned j=0 ; j < nbnd ; j++)
1186     {
1187         unsigned i = nbnd - 1 - j ;
1188         std::string treedir = BFile::FormPath(basedir, BStr::itoa(i));
1189
1190         NCSG* tree = new NCSG(treedir.c_str());
1191         tree->setIndex(i);
1192         tree->setVerbosity( verbosity );
1193         tree->setBoundary( bnd.getLine(i) );

165 GParts* GParts::make( NCSG* tree, const char* spec, unsigned verbosity )
166 {
167     assert(spec);
168
...
238     // GParts originally intended to handle lists of parts each of which
239     // must have an associated boundary spec. When holding CSG trees there
240     // is really only a need for a single common boundary, but for
241     // now enable reuse of the old GParts by duplicating the spec
242     // for every node of the tree
243
244     const char* reldir = "" ;  // empty reldir avoids defaulting to GItemList
245
246     GItemList* lspec = GItemList::Repeat("GParts", spec, ni, reldir) ;
247
248     GParts* pts = new GParts(nodebuf, tranbuf, planbuf, lspec) ;
249
250     //pts->setTypeCode(0u, root->type);   //no need, slot 0 is the root node where the type came from
251     return pts ;
252 }

hmm does GParts::close translate the spec into boundary int and write into partBuffer ? YEP : void GParts::registerBoundaries() // convert boundary spec names into integer codes using bndlib

200 RT_PROGRAM void intersect(int primIdx)
201 {
202     const Prim& prim    = primBuffer[primIdx];
203
204     unsigned partOffset  = prim.partOffset() ;
205     unsigned numParts    = prim.numParts() ;
206     unsigned primFlag    = prim.primFlag() ;
207
208     uint4 identity = identityBuffer[instance_index] ;
209
210
211     if(primFlag == CSG_FLAGNODETREE)
212     {
213         Part pt0 = partBuffer[partOffset + 0] ;
214
215         identity.z = pt0.boundary() ;        // replace placeholder zero with test analytic geometry root node boundary
216
217         evaluative_csg( prim, identity );
218         //intersect_csg( prim, identity );
219
220     }
221     else if(primFlag == CSG_FLAGINVISIBLE)
222     {
223         // do nothing : report no intersections for primitives marked with primFlag CSG_FLAGINVISIBLE
224     }
225 #ifdef WITH_PARTLIST
226     else if(primFlag == CSG_FLAGPARTLIST)
227     {
228         for(unsigned int p=0 ; p < numParts ; p++)
229         {
230             Part pt = partBuffer[partOffset + p] ;
231
232             identity.z = pt.boundary() ;
233

revisit tpmt–¶

The –apmtidx 1 option results in loading:

2017-04-10 15:02:46.231 FATAL [50057] [GGeo::loadAnalyticPmt@733] GGeo::loadAnalyticPmt AnalyticPMTIndex 1 AnalyticPMTSlice ALL Path /usr/local/opticks/opticksdata/export/DayaBay/GPmt/1

155 tpmt--(){
...
176
177     local apmtidx=1
178     # non-default AnalyticPMTIndex currently required for updated enums
...
181    op.sh \
182        --anakey $anakey \
183        --save \
184        --test --testconfig "$(tpmt-testconfig)" \
185        --torch --torchconfig "$(tpmt-torchconfig)" \
186        --cat $(tpmt-det) \
187        --tag $tag \
188        --timemax 10 \
189        --animtimemax 10 \
190        --eye 0.0,-0.5,0.0 \
191        --geocenter \
192        --apmtidx $apmtidx \
193        $*
194
195 }

root cause of difficulty¶

kludgy association of an old triangulated PMT with the analytic CSG one, actually it looks like there is one extra node in the triangulated ?
best solution would be to find a way to triangulate the CSG, so there would then be no solid/node matching problem
developing CSG to triangulation will take a while, so meanwhile just construct meshes using CSG bboxen ? See ggeo/test/GPmtTest.cc for start of this

symptom3 : surface attachement failure¶

see SurLib with Test Geometry

2017-03-16 17:49:08.898 INFO  [980504] [CTraverser::Traverse@128] CTraverser::Traverse DONE
2017-03-16 17:49:08.898 INFO  [980504] [CTraverser::Summary@104] CDetector::traverse numMaterials 5 numMaterialsWithoutMPT 0
2017-03-16 17:49:08.898 INFO  [980504] [CDetector::attachSurfaces@240] CDetector::attachSurfaces
2017-03-16 17:49:08.898 INFO  [980504] [GSurLib::examineSolidBndSurfaces@115] GSurLib::examineSolidBndSurfaces numSolids 7
Assertion failed: (node == i), function examineSolidBndSurfaces, file /Users/blyth/opticks/ggeo/GSurLib.cc, line 124.
Process 79145 stopped
* thread #1: tid = 0xef618, 0x00007fff96f1a866 libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
    frame #0: 0x00007fff96f1a866 libsystem_kernel.dylib`__pthread_kill + 10
libsystem_kernel.dylib`__pthread_kill + 10:
-> 0x7fff96f1a866:  jae    0x7fff96f1a870            ; __pthread_kill + 20
   0x7fff96f1a868:  movq   %rax, %rdi
   0x7fff96f1a86b:  jmp    0x7fff96f17175            ; cerror_nocancel
   0x7fff96f1a870:  retq
(lldb) bt
* thread #1: tid = 0xef618, 0x00007fff96f1a866 libsystem_kernel.dylib`__pthread_kill + 10, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
  * frame #0: 0x00007fff96f1a866 libsystem_kernel.dylib`__pthread_kill + 10
    frame #1: 0x00007fff8e5b735c libsystem_pthread.dylib`pthread_kill + 92
    frame #2: 0x00007fff95307b1a libsystem_c.dylib`abort + 125
    frame #3: 0x00007fff952d19bf libsystem_c.dylib`__assert_rtn + 321
    frame #4: 0x0000000101ce0ac9 libGGeo.dylib`GSurLib::examineSolidBndSurfaces(this=0x000000010e21e4a0) + 521 at GSurLib.cc:124
    frame #5: 0x0000000101ce08ad libGGeo.dylib`GSurLib::close(this=0x000000010e21e4a0) + 29 at GSurLib.cc:93
    frame #6: 0x0000000103ee0497 libcfg4.dylib`CDetector::attachSurfaces(this=0x000000010e21e1c0) + 247 at CDetector.cc:244
    frame #7: 0x0000000103e5ad26 libcfg4.dylib`CGeometry::init(this=0x000000010e21dc30) + 1446 at CGeometry.cc:73
    frame #8: 0x0000000103e5a770 libcfg4.dylib`CGeometry::CGeometry(this=0x000000010e21dc30, hub=0x000000010980c7a0) + 112 at CGeometry.cc:39
    frame #9: 0x0000000103e5ad8d libcfg4.dylib`CGeometry::CGeometry(this=0x000000010e21dc30, hub=0x000000010980c7a0) + 29 at CGeometry.cc:40
    frame #10: 0x0000000103f01286 libcfg4.dylib`CG4::CG4(this=0x000000010cadeab0, hub=0x000000010980c7a0) + 214 at CG4.cc:122
    frame #11: 0x0000000103f017bd libcfg4.dylib`CG4::CG4(this=0x000000010cadeab0, hub=0x000000010980c7a0) + 29 at CG4.cc:144
    frame #12: 0x0000000103ff1da3 libokg4.dylib`OKG4Mgr::OKG4Mgr(this=0x00007fff5fbfe6b0, argc=23, argv=0x00007fff5fbfe790) + 547 at OKG4Mgr.cc:35
    frame #13: 0x0000000103ff1ff3 libokg4.dylib`OKG4Mgr::OKG4Mgr(this=0x00007fff5fbfe6b0, argc=23, argv=0x00007fff5fbfe790) + 35 at OKG4Mgr.cc:41
    frame #14: 0x00000001000139be OKG4Test`main(argc=23, argv=0x00007fff5fbfe790) + 1486 at OKG4Test.cc:56
    frame #15: 0x00007fff9238d5fd libdyld.dylib`start + 1
(lldb)

(lldb) f 7
frame #7: 0x0000000103e5ad26 libcfg4.dylib`CGeometry::init(this=0x000000010e21dc30) + 1446 at CGeometry.cc:73
   70           detector  = static_cast<CDetector*>(new CGDMLDetector(m_hub, query)) ;
   71       }
   72
-> 73       detector->attachSurfaces();
   74       //m_csurlib->convert(detector);
   75
   76       m_detector = detector ;
(lldb)

symptom 2 : CPU/G4 cfg4/CTestDetector misunderstanding primordial CSG buffer ?¶

actually the PmtInBox code appears to be unaware of GCSG

tpmt-- --okg4

2017-03-16 13:51:10.046 INFO  [889146] [OpticksGen::targetGenstep@125] OpticksGen::targetGenstep setting frame 1 1.0000,0.0000,0.0000,0.0000 0.0000,1.0000,0.0000,0.0000 0.0000,0.0000,1.0000,0.0000 0.0000,0.0000,0.0000,1.0000
2017-03-16 13:51:10.047 FATAL [889146] [GenstepNPY::setPolarization@212] GenstepNPY::setPolarization pol 0.0000,0.0000,0.0000,0.0000 npol nan,nan,nan,nan m_polw nan,nan,nan,380.0000
2017-03-16 13:51:10.047 INFO  [889146] [SLog::operator@15] OpticksHub::OpticksHub DONE

*************************************************************
 Geant4 version Name: geant4-10-02-patch-01    (26-February-2016)
                      Copyright : Geant4 Collaboration
                      Reference : NIM A 506 (2003), 250-303
                            WWW : http://cern.ch/geant4
*************************************************************

2017-03-16 13:51:10.122 FATAL [889146] [CGeometry::init@59] CGeometry::init G4 simple test geometry
2017-03-16 13:51:10.122 INFO  [889146] [GGeo::createSurLib@656] deferred creation of GSurLib
2017-03-16 13:51:10.122 INFO  [889146] [GSurLib::collectSur@79]  nsur 48
2017-03-16 13:51:10.122 INFO  [889146] [CPropLib::init@68] CPropLib::init
2017-03-16 13:51:10.122 INFO  [889146] [CPropLib::initCheckConstants@120] CPropLib::initCheckConstants mm 1 MeV 1 nanosecond 1 ns 1 nm 1e-06 GC::nanometer 1e-06 h_Planck 4.13567e-12 GC::h_Planck 4.13567e-12 c_light 299.792 GC::c_light 299.792 dscale 0.00123984
2017-03-16 13:51:10.122 INFO  [889146] [*CTestDetector::makeDetector@118] CTestDetector::makeDetector PmtInBox 1 BoxInBox 0 numSolids (from mesh0) 7 numSolids (from config) 1
Assertion failed: (numSolids == numSolidsConfig), function makeDetector, file /Users/blyth/opticks/cfg4/CTestDetector.cc, line 127.
/Users/blyth/opticks/bin/op.sh: line 580: 41465 Abort trap: 6           /usr/local/opticks/lib/OKG4Test --anakey tpmt --save --test --testconfig mode=PmtInBox_pmtpath=/usr/local/opticks/opticksdata/export/dpib/GMergedMesh/0_control=1,0,0,0_analytic=1_apmtidx=1_node=box_parameters=0,0,0,300_boundary=Rock/NONE/perfectAbsorbSurface/MineralOil --torch --torchconfig type=disc_photons=500000_wavelength=380_frame=1_source=0,0,300_target=0,0,0_radius=100_zenithazimuth=0,1,0,1_material=Vacuum_mode=_polarization= --cat PmtInBox --tag 10 --timemax 10 --animtimemax 10 --eye 0.0,-0.5,0.0 --geocenter --okg4
/Users/blyth/opticks/bin/op.sh RC 134
simon:opticks blyth$


2017-03-16 14:17:21.209 INFO  [901864] [CPropLib::initCheckConstants@120] CPropLib::initCheckConstants mm 1 MeV 1 nanosecond 1 ns 1 nm 1e-06 GC::nanometer 1e-06 h_Planck 4.13567e-12 GC::h_Planck 4.13567e-12 c_light 299.792 GC::c_light 299.792 dscale 0.00123984
2017-03-16 14:17:21.209 INFO  [901864] [*CTestDetector::makeDetector@118] CTestDetector::makeDetector PmtInBox 1 BoxInBox 0 numSolidsMesh 7 numSolidsConfig 1
2017-03-16 14:17:21.209 INFO  [901864] [GMergedMesh::dumpSolids@617] CTestDetector::makeDetector (solid count inconsistent)
    0 ce             gfloat4      0.000      0.000      0.000    300.000  bb bb min   -300.000   -300.000   -300.000  max    300.000    300.000    300.000  ni(         0,         0,         0,4294967295) id(         0,         5,         0,         0)
    1 ce             gfloat4      0.000      0.000    -18.997    149.997  bb bb min   -100.288   -100.288   -168.995  max    100.288    100.288    131.000  ni(       720,       362,         1,         0) id(         1,         4,         1,         0)
    2 ce             gfloat4      0.000      0.000    -18.247    146.247  bb bb min    -97.288    -97.288   -164.495  max     97.288     97.288    128.000  ni(       720,       362,         2,         1) id(         2,         3,         2,         0)
    3 ce             gfloat4      0.005      0.004     91.998     98.143  bb bb min    -98.138    -98.139     55.996  max     98.148     98.147    128.000  ni(       960,       482,         3,         2) id(         3,         0,         3,         0)
    4 ce             gfloat4      0.000      0.000     13.066     98.143  bb bb min    -98.143    -98.143    -30.000  max     98.143     98.143     56.131  ni(       576,       288,         4,         2) id(         4,         1,         4,         0)
    5 ce             gfloat4      0.000      0.000    -81.500     83.000  bb bb min    -27.500    -27.500   -164.500  max     27.500     27.500      1.500  ni(        96,        50,         5,         2) id(         5,         2,         4,         0)
    6 ce             gfloat4      0.000      0.000      0.000    300.000  bb bb min   -300.000   -300.000   -300.000  max    300.000    300.000    300.000  ni(        12,        24,         0,4294967295) id(         0,      1000,         0,         0)
Assertion failed: (numSolidsMesh == numSolidsConfig), function makeDetector, file /Users/blyth/opticks/cfg4/CTestDetector.cc, line 133.

looks like okg4 not updated since primordial GCSG¶

Approach

make connection between the analytic GCSG volumes that CTestDetector::makePMT is going to use and the triangulated GMergedMesh solid count, then can update the assert
avoid duplicity regards the analytic PMT and honour the apmtidx version, by eliminating CPropLib::getPmtCSG

simon:opticks blyth$ opticks-find getPmtCSG
./cfg4/CPropLib.cc:GCSG* CPropLib::getPmtCSG(NSlice* slice)
./cfg4/CPropLib.cc:        LOG(error) << "CPropLib::getPmtCSG failed to load PMT" ;
./cfg4/CPropLib.cc:        LOG(error) << "CPropLib::getPmtCSG failed to getCSG from GPmt" ;
./cfg4/CTestDetector.cc:    GCSG* csg = m_mlib->getPmtCSG(slice);
./cfg4/CPropLib.hh:       GCSG*       getPmtCSG(NSlice* slice);


162 GCSG* CPropLib::getPmtCSG(NSlice* slice)
163 {
164    // hmm this is probably already loaded ???
165
166     GPmt* pmt = GPmt::load( m_ok, m_bndlib, 0, slice );    // pmtIndex:0
167
168     if(pmt == NULL)
169     {
170         LOG(error) << "CPropLib::getPmtCSG failed to load PMT" ;
171         return NULL ;
172     }
173
174     GCSG* csg = pmt->getCSG();
175
176     if(csg == NULL)
177     {
178         LOG(error) << "CPropLib::getPmtCSG failed to getCSG from GPmt" ;
179         return NULL ;
180     }
181     return csg ;
182 }

FIXED : symptom 1, GPU side mis-interpreting parts buffer after enum change¶

tpmt--

2017-03-15 20:48:44.712 INFO  [829428] [OContext::close@219] OContext::close numEntryPoint 2
##hemi-pmt.cu:bounds primIdx 0 is_partlist:0 min  -101.1682  -101.1682   -23.8382 max   101.1682   101.1682    56.0000
##hemi-pmt.cu:bounds primIdx 1 is_partlist:0 min   -98.1428   -98.1428    56.0000 max    98.1428    98.1428    98.0465
##hemi-pmt.cu:bounds primIdx 2 is_partlist:0 min   -98.0932   -98.0932    55.9934 max    98.0932    98.0932    98.0128
##hemi-pmt.cu:bounds primIdx 3 is_partlist:0 min   -27.5000   -27.5000  -164.5000 max    27.5000    27.5000     1.5000
##hemi-pmt.cu:bounds primIdx 4 is_partlist:0 min  -300.0100  -300.0100  -300.0100 max   300.0100   300.0100   300.0100
2017-03-15 20:48:45.342 INFO  [829428] [OPropagator::prelaunch@149] 1 : (0;500000,1) prelaunch_times vali,comp,prel,lnch  0.0000 0.2694 0.2364 0.0000
evaluative_csg primIdx_ 1 numParts 4 perfect tree fullHeight 4294967295 exceeds current limit
evaluative_csg primIdx_ 1 numParts 4 perfect tree fullHeight 4294967295 exceeds current limit
evaluative_csg primIdx_ 1 numParts 4 perfect tree fullHeight 4294967295 exceeds current limit
evaluative_csg primIdx_ 1 numParts 4 perfect tree fullHeight 4294967295 exceeds current limit

review of analytic PMT serialization¶

ana/pmt/analytic.py

Recreate the analytic PMT from detdecs parse with

pmt-analytic-tmp   # writing to $TMP/GPmt/0/GPmt.npy
pmt-analytic       # writing to $IDPATH/GPmt/0/GPmt.npy

Actual one in use is from opticksdata repo $OPTICKS_DATA/export/DayaBay/GPmt/0/

Comparing existing serializations¶

All three look effectively the same, with no influence from new enum so far:

simon:pmt blyth$ l /usr/local/opticks/opticksdata/export/DayaBay_VGDX_20140414-1300/g4_00.96ff965744a2f6b78c24e33c80d3a4cd.dae/GPmt/0/
total 48
-rw-r--r--  1 blyth  staff   848 Mar 15 16:27 GPmt.npy
-rw-r--r--  1 blyth  staff   289 Mar 15 16:27 GPmt_boundaries.txt
-rw-r--r--  1 blyth  staff  1168 Mar 15 16:27 GPmt_csg.npy
-rw-r--r--  1 blyth  staff    74 Mar 15 16:27 GPmt_lvnames.txt
-rw-r--r--  1 blyth  staff    47 Mar 15 16:27 GPmt_materials.txt
-rw-r--r--  1 blyth  staff    74 Mar 15 16:27 GPmt_pvnames.txt
simon:pmt blyth$
simon:pmt blyth$
simon:pmt blyth$ l $TMP/GPmt/0/
total 48
-rw-r--r--  1 blyth  wheel   848 Mar 15 17:31 GPmt.npy
-rw-r--r--  1 blyth  wheel   289 Mar 15 17:31 GPmt_boundaries.txt
-rw-r--r--  1 blyth  wheel  1168 Mar 15 17:31 GPmt_csg.npy
-rw-r--r--  1 blyth  wheel    74 Mar 15 17:31 GPmt_lvnames.txt
-rw-r--r--  1 blyth  wheel    47 Mar 15 17:31 GPmt_materials.txt
-rw-r--r--  1 blyth  wheel    74 Mar 15 17:31 GPmt_pvnames.txt
simon:pmt blyth$ diff -r --brief $IDPATH/GPmt/0 $TMP/GPmt/0
simon:pmt blyth$
simon:pmt blyth$
simon:pmt blyth$ l /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/
total 80
-rw-r--r--  1 blyth  staff   848 Jul  5  2016 GPmt.npy
-rw-r--r--  1 blyth  staff   289 Jul  5  2016 GPmt.txt
-rw-r--r--  1 blyth  staff   289 Jul  5  2016 GPmt_boundaries.txt
-rw-r--r--  1 blyth  staff   848 Jul  5  2016 GPmt_check.npy
-rw-r--r--  1 blyth  staff   289 Jul  5  2016 GPmt_check.txt
-rw-r--r--  1 blyth  staff  1168 Jul  5  2016 GPmt_csg.npy
-rw-r--r--  1 blyth  staff    47 Jul  5  2016 GPmt_csg.txt
-rw-r--r--  1 blyth  staff    74 Jul  5  2016 GPmt_lvnames.txt
-rw-r--r--  1 blyth  staff    47 Jul  5  2016 GPmt_materials.txt
-rw-r--r--  1 blyth  staff    74 Jul  5  2016 GPmt_pvnames.txt

simon:pmt blyth$ echo $OPTICKS_DATA
/usr/local/opticks/opticksdata
simon:pmt blyth$
simon:pmt blyth$ diff -r --brief $OPTICKS_DATA/export/DayaBay/GPmt/0/ $TMP/GPmt/0/
Only in /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/: GPmt.txt
Only in /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/: GPmt_check.npy
Only in /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/: GPmt_check.txt
Only in /usr/local/opticks/opticksdata/export/DayaBay/GPmt/0/: GPmt_csg.txt

Table Of Contents

Previous topic

Next topic

This Page

tpmt broken by OpticksCSG enum move¶

Vague recollections¶

2017-10-20 tpmt– broken (again) : QUALITATIVELY FIXED¶

Slice testing to isolate z-offset part¶

FIXED : two physvol offsets not being honoured¶

GPU dumping : note parallelised shuffle¶

old overview¶

2017-10-20 issue : boundaries not getting into the GParts ?¶

Review NCSG::Deserialize boundary handling¶

revisit tpmt–¶

root cause of difficulty¶

symptom3 : surface attachement failure¶

symptom 2 : CPU/G4 cfg4/CTestDetector misunderstanding primordial CSG buffer ?¶

looks like okg4 not updated since primordial GCSG¶

FIXED : symptom 1, GPU side mis-interpreting parts buffer after enum change¶

review of analytic PMT serialization¶

Comparing existing serializations¶

Navigation

Table Of Contents

Previous topic

Next topic

This Page

Quick search

tpmt broken by OpticksCSG enum move¶

Vague recollections¶

2017-10-20 tpmt– broken (again) : QUALITATIVELY FIXED¶

Slice testing to isolate z-offset part¶

FIXED : two physvol offsets not being honoured¶

GPU dumping : note parallelised shuffle¶

old overview¶

2017-10-20 issue : boundaries not getting into the GParts ?¶

Review NCSG::Deserialize boundary handling¶

revisit tpmt–¶

root cause of difficulty¶

symptom3 : surface attachement failure¶

symptom 2 : CPU/G4 cfg4/CTestDetector misunderstanding primordial CSG buffer ?¶

looks like okg4 not updated since primordial GCSG¶

FIXED : symptom 1, GPU side mis-interpreting parts buffer after enum change¶

review of analytic PMT serialization¶

Comparing existing serializations¶

Navigation