Changes

Summary

1. new lagrange basis functions coming together, shouldnt pass jenkins yet. (details)
2. Important Comment (details)
3. fixed the dang lagrange basis functions. and also tested them. (details)
4. touches. (details)
5. better tests (details)
6. documentation update (details)
7. added matlab script. (details)
8. seeking cg issue, not for upload (details)
9. upgraded tests somewhat confident p= 2, p= 3 CG fields should work. not sure why nanning. (details)
10. l2 working ad not... (details)
11. added projection test. (details)
12. sprinkling holy water on hugh's comment vandalism. (details)
13. documentation fix (details)
14. experimenting with VMS with mixed form (details)
15. ad4d working, cg fields 4d working up to p=3, new tests. (details)
16. jenkins fixes (details)
17. flip switches for commit. (details)
18. tweaks to AGLS, EDG stab; VMSD adaption case running (details)
19. added some RANS 2D unit tests (details)
20. Refactoring of LocalPatch. Next we need to remove the others. (details)
21. Fixed compile error (details)
22. refactoring edge local. no parallel tests yet, but serial seems to be the same (details)
23. Warning fix (details)
24. forgot to turn DG back to non-edge mode (details)
25. ElementDual is no more. Moving towards LocalPatch only (details)
26. elementxfieldspacetime fix. stabilized seems to be working. needs unit tests still. (details)
27. jenkins fix. (details)
28. unstab/gls fix (details)
29. Added unsplit LocalPatch constructor and universal LocalGlobal consistency check (details)
30. Don't check interior trace local global equivalence (details)
31. Added LocalGlobal equiv check to all LocalPatch tests. Added USE_MPI_XTERM cmake option. (details)
32. Add some Line tests for LocalPatch (details)
33. jEnKiNSSSS (details)
34. working on parallelizing edge (details)
35. and FINALLY, edge models and patches have a serial parallel equivalency test that passes! (details)
36. parenthesis misplacement (details)
37. Nodal Mode only supports LocalPatch now. Also added getEField() method to MDG (details)
38. left full nodal mode on... (details)
39. it was actually a struct (details)
40. made the nodal switch a run time one so that at least jenkins has to compile the nodal bits, even if they're not tested much yet (details)
41. removing ElementDual amongst other things (details)
42. apparently needed to include array.hpp from boost? (details)
43. speedup for MDG (details)
44. added boost version dependent serialization for array (details)
45. revert a bunch of changes (details)
46. edges always use global node numbering. Fixed bug with sub-cell index and reversed edges (details)
47. serialization is in the MPI folder correctly (details)
48. playing w pseudotime (details)
49. Various LocalPatch fixes (details)
50. fixes to update local2globalBoundaryMap issues (details)
51. removed constructEdgeMap() (details)
52. forgot to actually send the serialization (details)
53. f2d localpatch DG test move, refactor in progress. (details)
54. converted the 1D edge tests to LocalPatch (details)
55. the 2D local patch test now works for LocalPatch. Also minor typos in the other two tests (details)
56. split out test function into separate testing header. Also include blocker for make_unique.h (details)
57. bloody typo (details)
58. misnamed (details)
59. 3D edge local split and test. (details)
60. a 4D test. It can't actually call yet as Cory has left some developer exceptions in place, but this should work in the future (details)
61. tidy up (details)
62. a bunch of non-passing tests. Also testing the exception throwing from the edge local constructor (details)
63. assertion tests working (details)
64. More local patch updates (details)
65. refactor, debugging field2d_local_split_DG (details)
66. broke the checkLocalField test function out of the grid checker. Can now pass a local and a global field and check the transfer for all elements within. Doesn't check interior traces yet. (details)
67. asserted can only use the test with XField_LocalPatch (details)
68. Added 2D Edge Split LocalPatch tests (details)
69. Move more XFeid2D local edge tests (details)
70. Created XField3D_LocalPatch and removed old tests (details)
71. Remove overloded nDOF and DOF functions in XField_LocalPatch (details)
72. Alow high-order hierarchical meshes with LocalPatch (details)
73. added a test to compare the serial edge model construction between the old 'at node' technique and the new parallel friendly 'edge map' version. Seems to have been done correctly, so unsure what is causing the bug now... (details)
74. Updated 2D XField Element Dual LocalPatch tests (details)
75. work in progress. working tool functions, working on unit tests. (details)
76. renamed field_localpatch_edge_cg_btest.h to ..._toolkit.h (details)
77. More XField LocalPatch unit test conversions (details)
78. name fix (details)
79. name fix pt. ii (details)
80. moving towards block petsc solves (details)
81. EG cell stuff (details)
82. Parallelized some LocalPatch tests (details)
83. Checking that the error1_list and step_matrix_list match the old serial implementation. Still passing. wth (details)
84. asserted assumption of error0>=0 and adding exception catch tests (details)
85. the new synthesize_edge works in serial! Using a pifld(), a pointer to the indicator field instead of to the estimate field (details)
86. And with those bug fixes, we now have parallel edge solves (details)
87. Fix so local boundary patches are always consistent with hashing (details)
88. forgot to change NField to Field. Also a grid for the ErrorModel test (details)
89. forgot to turn off FULL_NODAL. Also removed the local function call that will be eroneous in guture (details)
90. forgot another NField() (details)
91. Merge from hcarson/develop (details)
92. field2d_localpatch_dg_btest done. (details)
93. pull fix (details)
94. Merge conflicts (details)
95. xfield2d_localpatch_btest done, working, added. (details)
96. traceorientation signed sealed delivered also tested. (details)
97. Set unit normal sign for 1D XField_Lagrange. Rename a bunch of LocalPatch tests. (details)
98. Updated LocalPatch Element DG tests (details)
99. ??? (details)
100. Lots of Local updates (details)
101. removed solution transfer from xfield test. (details)
102. xfield4d_localpatch_element_dg_btest started up. (details)
103. almost there... (details)
104. added 24ptope to xfield4d_localpatch_element_dg_btest. (details)
105. refactored XField4D_library to actually say it's a 6 pentatope case (details)
106. Lots of Local updates (details)
107. Remove the local patch #if switch (details)
108. Removed moew XField_Local (details)
109. Relaxed some tolerances (details)
110. added 6ptope test (details)
111. unit suite naming fix (details)
112. Removed the last use of old XField local (details)
113. Depricated old local XFields (details)
114. tryna adapt galerkin first shot. (details)
115. passing working state up to norm. (details)
116. added NewLinearDOFs accessors. The allocation of the local error indicator in the local patches should work for higher order local patches now (details)
117. set the FULL_NODAL flag to 1 permanently. Also added an exception on the CG and element branch of the local solves. Directs everyone to use the EDGE mode. (details)
118. forgot some files (details)
119. small changes (details)
120. Added hessq to diffusionViscousGradient. Finished QCR terms for 2D. (details)
121. added parallel to the AD test (details)
122. Fixes for g++-9 (details)
123. Allow parallel MMA option with SANSparallel optimizer (details)
124. realized we need ghosts.. lots of work to do (details)
125. Rename DOF_rank to DOFghost_rank for clarity (details)
126. test case mods to run on laptop... (details)
127. change MDG to VMSD (details)
128. working on EG field (details)
129. implementation of EG field (details)
130. tweaks (details)
131. fieldcopy tags (details)
132. fieldcopy fixed? (details)
133. diff fieldtag (details)
134. eg field compiling (details)
135. Made Field.deallocate protected (details)
136. unit tests for EG field (details)
137. bug fix for curved grids (details)
138. Initial attempt at 2x2 block PETSc (details)
139. Adding more parallel testing (details)
140. seeking bug in changes that's frustrating 4d adapts... (details)
141. Working on tests for new 2x2 block parallel solves (details)
142. Finished up additional parallel tests (details)
143. Working on parallel EDG (details)
144. progress toward static condensation (details)
145. missing instantiation (details)
146. add static condense to linear solver options (details)
147. Need Embedded DG field for EDG to work in parallel (details)
148. serial SC working (details)
149. Fixed compile errors (details)
150. added l2 sandbox distinct from philips since hes working on paper (details)
151. unfinished if fix (details)
152. VMSD seems to function in parallel (details)
153. One more compile error (details)
154. looking for bugs (details)
155. More fixes (details)
156. vmsd adapt almostttt working (details)
157. Relaxed tolerance (details)
158. fixed up error estimate so that VMSD adaption should work (details)
159. One more tolerance relaxation (details)
160. Another relaxation for intel (details)
161. Yet another tolerance relaxation (details)
162. Remove accidental paste (details)
163. don't use unnecessary field for error estimate`:wq (details)
164. update error indicator (details)
165. got rid of eSfld (details)
166. moving towards ghost residuals for VMSD (details)
167. commit fixes (details)
168. petsc static condense refactor (details)
169. cg embedded ghosts (details)
170. added isStaticCondensed to algEqSet base class (details)
171. jenkins fixins (details)
172. trying to get unit tests to build (details)
173. Something is still wrong with VMSD in parallel (details)
174. unit tests compiling (details)
175. SA unit tests for static condensed VMSD (details)
176. using BiCGstab for unit test (details)
177. some fixes (details)
178. added new 4d function, errorestimate tests for 4d (details)
179. push up to show weird results. (details)
180. errorestimate_l2 tests ready for 4d, failing, probably due to bug. (details)
181. thing that broke earlker commit. (details)
182. turn off failing test. (details)
183. affline typo ffs (details)
184. jenkins is a real chowderhead. (details)
185. fixed petsc btest, reverted some edg things (details)
186. Guard against committing SA with QCR (details)
187. updating error estimate test (details)
188. fixed up VMSD unit tests, PTC static condensation (details)
189. vmsd local solves working...? (details)
190. hunting some parallel bugs for Arthur and the curved grids (details)
191. added an adapt to implied metric method. Useful for getting avro out of a hole on the corner singularity adapt (details)
192. fixed unit test problem (details)
193. got rid of warning, added PTC unit test (details)
194. fixed mpi reduce div by zero bug (details)
195. errorestimate 4d test is working... (details)
196. cg field constructor became ambiguous... fixing this (details)
197. finished cg constructor stuff.. added accessor for bubble dofs (details)
198. vmsd working with p>D (extra equations zeroed out) (details)
199. fixed up unit tests (details)
200. missed one fix (details)
201. changed a slightly misnamed template (details)
202. Refactored synthesize_edge. Didn't find anything fishy. Other small fixes. (details)
203. PETSc debug matrix dumps for MATLAB (details)
204. case adjustment. (details)
205. static condensed parallel VMSD is working!!! (details)
206. formatting (details)
207. fixed cg field constructor (details)
208. forgot a file (details)
209. last CG constructor change? (details)
210. fixed warning, added PETSc memory monitor flag (details)
211. Added edge-element ErrorModel. Various small fixes. (details)
212. push to hypersonic (details)
213. revert some things (details)
214. add petsc stuff to cmake file (details)
215. Fixes for 4D L2 projection adaptation. SANS optimizer has no lower bound on costTarget. FrobNorm does not use coarsening factor for FrobNorm. (details)
216. speedup for vmsd jacobians (details)
217. Vera fixes (details)
218. more speedup (details)
219. switched to using some unique_ptrs so that fields only get created when they're actually needed (details)
220. Fixed warning (details)
221. skip some unnecessary jacobians during local solves (details)
222. some random tests and 4d updates (details)
223. Trying to reduce compile time (details)
224. Missed one (details)
225. added some more timers. Also undid that merge conflict horror (details)
226. check for NaNs in residual norm (details)
227. fix (details)
228. annoying bit to change. (details)
229. ambiguous max (details)
230. static condensation for local solves (details)
231. revert to usual pseudotime logic (details)
232. trying out nonlinear version of vmsd (details)
233. another std::max fix (details)
234. case mods for hypersonic. (details)
235. parallel fix for 2D+T AD cases. (details)
236. bug fix (details)
237. another jacobian fix (details)
238. last bugs hopefully (details)
239. fixing stabilization (details)
240. flip to edge localsolve on CG adapt3d case to pass. (details)
241. fixed funkiness from paralell outputs. (details)
242. Just-in-time cell quadrature caching (details)
243. added validity check for q+q' (details)
244. bug fix on error estimate (details)
245. fix parallel mistake. (details)
246. Fix divide by zero and link error (details)
247. Just-in-time basis quadrature trace caching (details)
248. Longer timeout for AlgebraicEquationSet_Galerkin_Stabilized_Triangle_SA_btest (details)
249. Caching of basis hessian at quadrature points (details)
250. Compile fix (details)
251. Add Hex Legendre hessians (details)
252. Fixes in tests (details)
253. Hopefully last minor fixes (details)
254. parallel output fix, hopefully nightly fix. (details)
255. Added SA QCR for 3D. Other small cleanups. (details)
256. Fix compile error (details)
257. fixed stabilized output issue. (details)
258. hacky new local patch (works on interiors) (details)
259. Trace normal checking only works for P1. Added option for valgrind massif flags. (details)
260. forgot some tests (details)
261. Use prolongated adjoint as initial guess for P+1 adjoint with VMSD (details)
262. fixed true output (details)
263. Missed one (details)
264. Missed another (details)
265. Added MeshNet option for EPIC (details)
266. One more fix (details)
267. added a function for constructing maps across broken field traces. (details)
268. Created safe_at with better error message for std::container 'at' method (details)
269. Deal with const for safe_at (details)
270. Add missing include (details)
271. Actual correct const treatment for safe_at (details)
272. more fields are now pointers. Should reduce the desctructors cost (details)
273. a 3D flat plate case (details)
274. missed some bits (details)
275. missed some more.... (details)
276. local solves on broken patches working (details)
277. vmsd working with new brokenfield local solves (details)
278. why isn't this working jenkins? (details)
279. fixed SIP stuff w paramflds (details)
280. add some instantiations (details)
281. fixed errormodel_impl weird bug, scripts to SANSparallel optimizer (details)
282. put in #def to switch between local solve modes (details)
283. missed one (details)
284. changed some of the unique ptrs to use Fields (details)
285. change test name (details)
286. Add static condensation to UMFPACK (details)
287. updating unit tests (details)
288. final unit tests, fixed functional bug, added 3D adapt (details)
289. fixing up more unit tests (details)
290. more unit test fixes (details)
291. lowered tolerance (details)
292. whoops wrong way (details)
293. caught bug in vmsd volume fcn jacobian, added tests (details)
294. added DG to CG dof map function. Also added pointers for the residual field to the field bundles. The local fields should populate directly from the global residual field. There still needs to be a function populated in solverInterface_Galerkin to fill the globalDG residual field (details)
295. eliminated duplicate residual evaluations for local solves (details)
296. vmsd speedup residuals working (details)
297. Nodal estimation uses an internal DG field, which can be accessed through a get. Need to figure out how to get this data out and into the local problem (details)
298. added switch to use wholepatch (details)
299. added RAE2822 sandbox case and grids. (details)
300. local estimation specialized for Galerkin now works. Maybe? (details)
301. add 3d NS solution fcn (details)
302. beginning to add new asymptotic test but not done yet. (details)
303. working on speeding up the estimates (details)
304. error estimate transfer working! (details)
305. revert a change (details)
306. arrayQ to real for systems (details)
307. vera fix (details)
308. fixed the virtual function, removed isLocal from the estimation. Remains to fix the boundary maps and interior trace groups in the basic LocalProblem (details)
309. bug fix, adding some toy problems (details)
310. fixed DGAdvective adapts should use DGAdvective solver interface (details)
311. removed isLocal and non-interior trace group constructors for CG local. The local solves now work by setting cell, interior and boundary groups to limit the solves. (details)
312. pushing to hypersonic (details)
313. add another 3d ojeda (details)
314. fix typo (details)
315. move the output inside the loop (details)
316. everything compiles... (details)
317. mod soln fcn for 3d ojeda (details)
318. L2 now uses the new patch (details)
319. Fixes for SANSparallel metric optimization (details)
320. update unit test (details)
321. need to initialize fields.. (details)
322. sandbox cases for thesis (details)
323. more unit tests fixes.. cmon Jenkins! (details)
324. lines too long (details)
325. added jump penalization for L2 projection. NOWEVERYTHINGWORKSOHYEAH. (details)
326. fixed bug in the test (details)
327. modified the Triple BL cases to match up. Bunch of things: volume functional is solution, they take command line arguments, they actually work with avro, they can do chain adaptation if we want. (details)
328. unused variabels (details)
329. Added a JumpPenalty integrand for interior traces in L2. Doesn't work as well as hoped. Also made the WHOLEPATCH flag work again for AD (details)
330. take the cell groups off of the functional used in the estimation. Avoids double counting when switching between inner and whole patch (details)
331. Potential Adjoint updates (details)
332. sandbox with timings (details)
333. functionalcell galerkin fix on cory (details)
334. oops missed one. (details)
335. Fix kutta frame for potential adjoint/drela (details)
336. Caching unit test fix (details)
337. DeltaWing cases (details)
338. more tweaks (details)
339. more tweaks (details)
340. Relax PETSc divergence tolerance. Use P adjoint for P+1 for VMSD. Add PolarGauss function for 2D L2 project. (details)
341. Fix avro interface (details)
342. Fix segfault for parallel metric dump (details)
343. Updated PETSc repository URL (details)
344. Nightly fixes (details)
345. working on error order tests, making sure everything is the same across machines. (details)
346. all that time down the drain? ffs. (details)
347. Revived avro fill function for 3D (details)
348. Report min/max edge count on processors for edge local solves (details)
349. Add enum to track local timing. Split out communication time, and report parallel efficiency (details)
350. Fix typo (details)
351. Various small fixes (details)
352. adding a bunch of verbosity and turning the SANSparallel optimizer back off (details)
353. More robust SANS-MMA parallle optimization (details)
354. added a safe_at to XField_impl.h, also working towards a re-entrant Corner Poisson case (details)
355. Use fluxViscousLinearInGradient trick for VMSD (details)
356. clang fix. (details)
357. jenkins fixes (details)
358. ejnkins fix 2.0 (details)
359. Check the boundary count matches between xfld and avro (details)
360. Updated XField2D_BackwardsStep so it works with avro (details)
361. prep for 4d BL problem, no MMS. (details)
362. Comment out egads dump (details)
363. quick fix (details)
364. avroXField patched. (details)
365. Longer allowed time for ErrorOrder 4D (details)
366. Update avro build to use gitlab (details)
367. exact solution fix. (details)
368. Relaxed timing for refine (details)
369. Fixed issues with Hierarchical basis. Refactored Element projections. Some implied metric experiments. (details)
370. Fixed compile errors (details)
371. Fix assert (details)
372. Declare PyDict methods extern to reduce compile time (details)
373. Relaxed some tolerances (details)
374. Warning fix (details)

new lagrange basis functions coming together, shouldnt pass jenkins yet.

Important Comment

fixed the dang lagrange basis functions. and also tested them.

touches.

better tests

documentation update

added matlab script.

seeking cg issue, not for upload

upgraded tests somewhat confident p= 2, p= 3 CG fields should work. not sure why nanning.

l2 working ad not...

added projection test.

sprinkling holy water on hugh's comment vandalism.

documentation fix

experimenting with VMS with mixed form

ad4d working, cg fields 4d working up to p=3, new tests.

jenkins fixes

flip switches for commit.

tweaks to AGLS, EDG stab; VMSD adaption case running

added some RANS 2D unit tests

Refactoring of LocalPatch. Next we need to remove the others.

Fixed compile error

refactoring edge local. no parallel tests yet, but serial seems to be the same

Warning fix

forgot to turn DG back to non-edge mode

ElementDual is no more. Moving towards LocalPatch only

elementxfieldspacetime fix. stabilized seems to be working. needs unit tests still.

jenkins fix.

unstab/gls fix

Added unsplit LocalPatch constructor and universal LocalGlobal consistency check

Don't check interior trace local global equivalence

Added LocalGlobal equiv check to all LocalPatch tests. Added USE_MPI_XTERM cmake option.

Add some Line tests for LocalPatch

jEnKiNSSSS

working on parallelizing edge

and FINALLY, edge models and patches have a serial parallel equivalency test that passes!

parenthesis misplacement

Nodal Mode only supports LocalPatch now. Also added getEField() method to MDG

left full nodal mode on...

it was actually a struct

made the nodal switch a run time one so that at least jenkins has to compile the nodal bits, even if they're not tested much yet

removing ElementDual amongst other things

apparently needed to include array.hpp from boost?

speedup for MDG

added boost version dependent serialization for array

revert a bunch of changes

edges always use global node numbering. Fixed bug with sub-cell index and reversed edges

serialization is in the MPI folder correctly

playing w pseudotime

Various LocalPatch fixes

fixes to update local2globalBoundaryMap issues

removed constructEdgeMap()

forgot to actually send the serialization

f2d localpatch DG test move, refactor in progress.

converted the 1D edge tests to LocalPatch

the 2D local patch test now works for LocalPatch. Also minor typos in the other two tests

split out test function into separate testing header. Also include blocker for make_unique.h

bloody typo

misnamed

3D edge local split and test.

a 4D test. It can't actually call yet as Cory has left some developer exceptions in place, but this should work in the future

tidy up

a bunch of non-passing tests. Also testing the exception throwing from the edge local constructor

assertion tests working

More local patch updates

refactor, debugging field2d_local_split_DG

broke the checkLocalField test function out of the grid checker. Can now pass a local and a global field and check the transfer for all elements within. Doesn't check interior traces yet.

asserted can only use the test with XField_LocalPatch

Added 2D Edge Split LocalPatch tests

Move more XFeid2D local edge tests

Created XField3D_LocalPatch and removed old tests

Remove overloded nDOF and DOF functions in XField_LocalPatch

Alow high-order hierarchical meshes with LocalPatch

added a test to compare the serial edge model construction between the old 'at node' technique and the new parallel friendly 'edge map' version. Seems to have been done correctly, so unsure what is causing the bug now...

Updated 2D XField Element Dual LocalPatch tests

work in progress. working tool functions, working on unit tests.

renamed field_localpatch_edge_cg_btest.h to ..._toolkit.h

More XField LocalPatch unit test conversions

name fix

name fix pt. ii

moving towards block petsc solves

EG cell stuff

Parallelized some LocalPatch tests

Checking that the error1_list and step_matrix_list match the old serial implementation. Still passing. wth

asserted assumption of error0>=0 and adding exception catch tests

the new synthesize_edge works in serial! Using a pifld(), a pointer to the indicator field instead of to the estimate field

And with those bug fixes, we now have parallel edge solves

Fix so local boundary patches are always consistent with hashing

forgot to change NField to Field. Also a grid for the ErrorModel test

forgot to turn off FULL_NODAL. Also removed the local function call that will be eroneous in guture

forgot another NField()

Merge from hcarson/develop

field2d_localpatch_dg_btest done.

pull fix

Merge conflicts

xfield2d_localpatch_btest done, working, added.

traceorientation signed sealed delivered also tested.

Set unit normal sign for 1D XField_Lagrange. Rename a bunch of LocalPatch tests.

Updated LocalPatch Element DG tests

???

Lots of Local updates

removed solution transfer from xfield test.

xfield4d_localpatch_element_dg_btest started up.

almost there...

added 24ptope to xfield4d_localpatch_element_dg_btest.

refactored XField4D_library to actually say it's a 6 pentatope case

Lots of Local updates