Changes

Summary

1. src/pde/NS/ (details)
2. Finally commiting some artificial viscosity options (details)
3. Merge conflict (details)
4. tweak sandbox cases (details)
5. WHOLEPATCH and INNERPATCH flags actually work now. They're fairly janky, but we can use them for comparisons (details)
6. added some asserts to avoid that bug again (details)
7. eurgh typo (details)
8. VMSD DeltaWing case (details)
9. experimenting with source stabilization (details)
10. perturbedSource functions (details)
11. Added unit tests for 2D RANSSA perturbedSource. Added 3D perturbedSource (needs test) (details)
12. the janky test (details)
13. the janky test (details)
14. working on speedups (details)
15. Failing unit test for ProjectSolnCell_Continuous (details)
16. various spot fixes, also working towards L shaped domain Poisson case (details)
17. minor changes. Getting them to coincide so I can get off this branch (details)
18. merging the 6 group backwards step onto my branch (details)
19. added a graceful restart capability. It searches for the last mesh from EPIC and restarts there, then appends to the adapthist file accordingly. Should also solve on the final grid, without appending (details)
20. make the 10% bump on beta canon (details)
21. L shaped domain case actually works pretty well now (details)
22. lower the linear solver tolerance (details)
23. 25% bump on the ball (details)
24. turns out perturbedDiffusion is not worth it.. (details)
25. bumping up beta. Also actually got the graceful restart working (details)
26. updates to pdes for VMSD (details)
27. adding command line args (details)
28. bug fix (details)
29. typo (details)
30. bumping beta up to 1.5*sqrt(d) for vertex and 2sqrt(d) for elemental. Also switching the two cases to ReverseCuthillMcKee, and various other PETSc tweaks (details)
31. bumping beta up to 1.5*sqrt(d) for vertex and 2sqrt(d) for elemental. Also switching the two cases to ReverseCuthillMcKee, and various other PETSc tweaks (details)
32. added checks based on number of processors, to start with smallest grids possible (details)
33. change back to QDM (details)
34. added checks based on number of processors, to start with smallest grids possible (details)
35. change world size per grid (details)
36. missing ndconverts.. new pyrite files (details)
37. fix to allow 0 element in field (details)
38. duplicate functions (details)
39. cases BL update: MMS sad (details)
40. bug fix for perturbedFluxViscous (details)
41. unit test fixes (details)
42. fix uninitialized data (details)
43. switched to mms sinusoidal decay fcn (details)
44. fixes to tests. Now there's a weird one... (details)
45. avoid uninitialized memory (details)
46. this should be passing. VMSD should be able to replace the lack of a residual field with a full field evaluation, by setting local_cell_groups to {0,1}. HOWEVER, it can't (details)
47. this should be passing. VMSD should be able to replace the lack of a residual field with a full field evaluation, by setting local_cell_groups to {0,1}. HOWEVER, it can't (details)
48. print the offending error1. stop print nElem to screen. A notebook with Corner singularity exact solution in it (details)
49. new exact solution AD function. (details)
50. moving to Arthur (details)
51. read in from big Grids (details)
52. forgot to commit (details)
53. switched to quasi-BL exact solution case. (details)
54. vmsd local solve unit test working (details)
55. switching to using projectTo (details)
56. Chain adapt case. Use the lower cost levels to precondition the upper levels. (details)
57. vera fix (details)
58. added powerL, powerH back to specify width of the chain (details)
59. missing include (details)
60. removed SANS parallel from test case. (details)
61. petsc memory consumption now approx correct (details)
62. tweaks (details)
63. 2.0 multiplier on beta for elemental model is much too large, gives twitch for P1 Corner sing L2. Also made the chain adapt look for a candidate starting mesh in the folder below it (details)
64. Actually append to adapthist. Sigh (details)
65. merge con (details)
66. RANS MMS for PETSc comparisons (details)
67. added an exception to dump local fields if theres a nan generated in the local solve estimation, it's virtual so a different disc can dump more fields. Also some comments about nan checking. (details)
68. switching to using the projectTo field capability. Removing the explicit projector usage (details)
69. add to BC list (details)
70. restarting tweak (details)
71. fix for jenkins and a tweak to a notebook (details)
72. MPI blocker (details)
73. forgot to turn off the new logic (details)
74. another missing include blocker (details)
75. Fix InterpolateFunctionCell_Continuous for Hierarchical (details)
76. added option to read in much finer intial grid (details)
77. A chain adapt for DGBR2 (details)
78. tweaks (details)
79. pre merge (details)
80. if line search cuts back all the way and is unphysical, don't take the update (details)
81. new triple BL with temporal forcing (details)
82. fixed dumb mistakes (details)
83. fix + case adjustment. (details)
84. vera fix (details)
85. avro-proofing for jenkins. (details)
86. Consistent hRefinement with new error model (details)
87. Fix compile error (details)
88. Fix compile warning (details)
89. Relaxed a tolerance (details)
90. added scaling for RANS equations and unit tests (details)
91. Have SANS nlopt optimizer use optimizer base. Add trace pentalty term to SANS optimizer. (details)
92. Line search wasn't doing final validity check (details)
93. add 3D RANS ojeda test cases (details)
94. Fix link error (details)
95. forgot some files (details)
96. Vera fixes (details)
97. mod output fcn (details)
98. cases realigned. (details)
99. removed conflicting includes (details)
100. tweaks to linesearch (details)
101. fixed warning (details)
102. Vera fixes (details)
103. hack to libmeshb, added hooks for eventual restart (details)
104. put in crude restart.. seems to work? (details)
105. sandbox edits (details)
106. added new residnorm to localsolves (details)
107. forgot a place to add the residNorm (details)
108. lines too long (details)
109. bug in serial version (details)
110. instantiate some things (details)
111. missing includes (details)
112. forgot to add unit test (details)
113. whoops forgot to add the instantiations (details)
114. nonlinear speedups for vmsd (details)
115. fixed Roe btest, added more instantiations (details)
116. more instantiations for restart.. undef'ing the nonlinear stuff (details)
117. added validity check on interior traces, fix for bug in local solve (details)
118. Jenkins let me through... (details)
119. one more (details)
120. moving 0-0 trace thing over to solverinterface_vmsd, fixing some unit tests (details)
121. edit cmake, fix up some unit testing (details)
122. relax tol for intel (details)
123. moving linear VMSD to a different algEqSet (details)

src/pde/NS/

Finally commiting some artificial viscosity options

Merge conflict

tweak sandbox cases

WHOLEPATCH and INNERPATCH flags actually work now. They're fairly janky, but we can use them for comparisons

added some asserts to avoid that bug again

eurgh typo

VMSD DeltaWing case

experimenting with source stabilization

perturbedSource functions

Added unit tests for 2D RANSSA perturbedSource. Added 3D perturbedSource (needs test)

the janky test

the janky test

working on speedups

Failing unit test for ProjectSolnCell_Continuous

various spot fixes, also working towards L shaped domain Poisson case

minor changes. Getting them to coincide so I can get off this branch

merging the 6 group backwards step onto my branch

added a graceful restart capability. It searches for the last mesh from EPIC and restarts there, then appends to the adapthist file accordingly. Should also solve on the final grid, without appending

make the 10% bump on beta canon

L shaped domain case actually works pretty well now

lower the linear solver tolerance

25% bump on the ball

turns out perturbedDiffusion is not worth it..

bumping up beta. Also actually got the graceful restart working

updates to pdes for VMSD

adding command line args

bug fix

typo

bumping beta up to 1.5*sqrt(d) for vertex and 2sqrt(d) for elemental. Also switching the two cases to ReverseCuthillMcKee, and various other PETSc tweaks

bumping beta up to 1.5*sqrt(d) for vertex and 2sqrt(d) for elemental. Also switching the two cases to ReverseCuthillMcKee, and various other PETSc tweaks

added checks based on number of processors, to start with smallest grids possible

change back to QDM

added checks based on number of processors, to start with smallest grids possible

change world size per grid

missing ndconverts.. new pyrite files

fix to allow 0 element in field

duplicate functions

cases BL update: MMS sad

bug fix for perturbedFluxViscous

unit test fixes

fix uninitialized data

switched to mms sinusoidal decay fcn

fixes to tests. Now there's a weird one...

avoid uninitialized memory

this should be passing. VMSD should be able to replace the lack of a residual field with a full field evaluation, by setting local_cell_groups to {0,1}. HOWEVER, it can't

this should be passing. VMSD should be able to replace the lack of a residual field with a full field evaluation, by setting local_cell_groups to {0,1}. HOWEVER, it can't

print the offending error1. stop print nElem to screen. A notebook with Corner singularity exact solution in it

new exact solution AD function.

moving to Arthur

read in from big Grids

forgot to commit

switched to quasi-BL exact solution case.

vmsd local solve unit test working

switching to using projectTo

Chain adapt case. Use the lower cost levels to precondition the upper levels.

vera fix

added powerL, powerH back to specify width of the chain

missing include

removed SANS parallel from test case.

petsc memory consumption now approx correct

tweaks

2.0 multiplier on beta for elemental model is much too large, gives twitch for P1 Corner sing L2. Also made the chain adapt look for a candidate starting mesh in the folder below it

Actually append to adapthist. Sigh

merge con

RANS MMS for PETSc comparisons

added an exception to dump local fields if theres a nan generated in the local solve estimation, it's virtual so a different disc can dump more fields. Also some comments about nan checking.

switching to using the projectTo field capability. Removing the explicit projector usage

add to BC list

restarting tweak

fix for jenkins and a tweak to a notebook

MPI blocker

forgot to turn off the new logic

another missing include blocker

Fix InterpolateFunctionCell_Continuous for Hierarchical

added option to read in much finer intial grid

A chain adapt for DGBR2

tweaks

pre merge

if line search cuts back all the way and is unphysical, don't take the update

new triple BL with temporal forcing

fixed dumb mistakes

fix + case adjustment.

vera fix

avro-proofing for jenkins.

Consistent hRefinement with new error model

Fix compile error

Fix compile warning

Relaxed a tolerance

added scaling for RANS equations and unit tests

Have SANS nlopt optimizer use optimizer base. Add trace pentalty term to SANS optimizer.

Line search wasn't doing final validity check

add 3D RANS ojeda test cases

Fix link error

forgot some files

Vera fixes

mod output fcn

cases realigned.

removed conflicting includes

tweaks to linesearch

fixed warning

Vera fixes

hack to libmeshb, added hooks for eventual restart

put in crude restart.. seems to work?

sandbox edits

added new residnorm to localsolves

forgot a place to add the residNorm

lines too long

bug in serial version

instantiate some things

missing includes

forgot to add unit test

whoops forgot to add the instantiations

nonlinear speedups for vmsd

fixed Roe btest, added more instantiations

more instantiations for restart.. undef'ing the nonlinear stuff

added validity check on interior traces, fix for bug in local solve

Jenkins let me through...

one more

moving 0-0 trace thing over to solverinterface_vmsd, fixing some unit tests

edit cmake, fix up some unit testing

relax tol for intel

moving linear VMSD to a different algEqSet