Implement the rotation of the stiffness tensor from the local frame to the global frame (+ other fixes) by jeanmichelscherer · Pull Request #3 · bourdin/mef90

jeanmichelscherer · 2022-06-22T22:11:37Z

No description provided.

…yaml input) to the global frame. The rotation matrix follows the Bunge (passive) convention (ZXZ) and can be defined from 3 angles (phi1, Phi, phi2) or three vectors (V1,V2,V3).

…ton-like evolution equation). Single slip and Body-Centered Cubic (BCC) behaviours are considered.

… True or False)

…y/june 2020 (ATmodel and split operator)

…injon regularization (L-inf norm regularized by L-p norm): see for instance Manik, Asadkandi and Holmedal (CMAME, 2022)

bourdin · 2022-07-26T18:28:48Z

Would you mind splitting this MR into smaller ones:

changes to python scripts
memory leak
double contraction of tensor and bug fix in addition of 4th order tensor

Then, I am not sure to understand why you need both the rotation matrix , the Euler angles and the vectors.

jeanmichelscherer · 2022-07-26T19:35:32Z

It seems that my newest commits have automatically been added to the MR, which is not what I wanted.
I will do new/smaller MRs for each part.

It is convenient to be able to give a rotation matrix either from vectors or angles.
The first option is particularly useful when you want to test a precise orientation, say the [111] axis of a crystal.
The second option is more useful for randomly generated or measured orientations.
From my experience, rotation matrices/angles are very error-prone. So, it seems to be a good idea to have all those transformations hard-coded and validated in vDef instead of letting each user make its own pre-processor.
Let me know if you would prefer to keep only one of the two options.

bourdin · 2022-07-26T20:54:53Z

It is convenient to be able to give a rotation matrix either from vectors or angles.
The first option is particularly useful when you want to test a precise orientation, say the [111] axis of a crystal.
The second option is more useful for randomly generated or measured orientations.
From my experience, rotation matrices/angles are very error-prone. So, it seems to be a good idea to have all those transformations hard-coded and validated in vDef instead of letting each user make its own pre-processor.
Let me know if you would prefer to keep only one of the two options.

Very good point. That is actually why I implemented isotropic Hooke's laws on top of full tensor

bourdin · 2022-08-15T19:20:53Z

Was everything in here included in the 3 PR I just merged?

jeanmichelscherer · 2022-08-17T00:38:49Z

There are two commits I did not include in the smaller PRs:

Commit 9b44402 : In the Makefile.include, I have to add -I${SNLP_DIR}/obj in the FC_FLAGS and C_FLAGS.
Commit d64abaf : It implements the double contraction of two 4th order tensors. It might be useful for the Hydro-Dev split of a cubic elasticity behavior, which I've not really been using as of now. I have not checked if the double contraction implementation gives valid results yet.

bourdin · 2022-08-17T01:07:23Z

There are two commits I did not include in the smaller PRs:

Commit 9b44402 : In the Makefile.include, I have to add -I${SNLP_DIR}/obj in the FC_FLAGS and C_FLAGS.

THese should not be needed since we build a shared library

Commit d64abaf : It implements the double contraction of two 4th order tensors. It might be useful for the Hydro-Dev split of a cubic elasticity behavior, which I've not really been using as of now. I have not checked if the double contraction implementation gives valid results yet.

We can either merge these and print a message stating that this has not been tested, or build a test. Can we compare (AB)e and A(Be) for random A, B, and e, for instance?

jeanmichelscherer · 2022-08-17T17:01:20Z

There are two commits I did not include in the smaller PRs:

Commit 9b44402 : In the Makefile.include, I have to add -I${SNLP_DIR}/obj in the FC_FLAGS and C_FLAGS.

THese should not be needed since we build a shared library

If not added, I get the following compilation error:

/home/scherer/Softwares/MEF90/mef90dev/m_DefMech/m_MEF90_DefMechPlasticity.F90:1026:10:

 1026 |       use SNLPF90
      |          1
Fatal Error: Cannot open module file ‘snlpf90.mod’ for reading at (1): No such file or directory
compilation terminated.

Commit d64abaf : It implements the double contraction of two 4th order tensors. It might be useful for the Hydro-Dev split of a cubic elasticity behavior, which I've not really been using as of now. I have not checked if the double contraction implementation gives valid results yet.

We can either merge these and print a message stating that this has not been tested, or build a test. Can we compare (AB)e and A(Be) for random A, B, and e, for instance?

I will work on building a test and add it to a new PR.

In addition, I forgot to mention that I get the following compilation error from MEF90/m_MEF90_EXO.F90:

/home/scherer/Softwares/MEF90/mef90dev/MEF90/m_MEF90_EXO.F90:350:51:

  350 |          Call MPI_Bcast(EXOelemType,MXSTLN,MPI_CHAR,0,MEF90Ctx%comm,ierr)
      |                                                   1
Error: Symbol ‘mpi_char’ at (1) has no IMPLICIT type

That error can be solved by replacing MPI_CHAR with MPI_CHARACTER. This was also included in my original (big) PR but not in the smaller ones.

bourdin · 2022-08-17T19:39:20Z

There are two commits I did not include in the smaller PRs:

Commit 9b44402 : In the Makefile.include, I have to add -I${SNLP_DIR}/obj in the FC_FLAGS and C_FLAGS.

THese should not be needed since we build a shared library

If not added, I get the following compilation error:
/home/scherer/Softwares/MEF90/mef90dev/m_DefMech/m_MEF90_DefMechPlasticity.F90:1026:10:

 1026 |       use SNLPF90
      |          1
Fatal Error: Cannot open module file ‘snlpf90.mod’ for reading at (1): No such file or directory
compilation terminated.

This makes no sense what is $SNLP_DIR and what's in this directory?
in $MEF90_DIR/Makefile.include:

ifneq (${SNLP_DIR},)
	FC_FLAGS+= -DMEF90_HAVE_SNLP -I${SNLP_DIR}/include
	C_FLAGS+= -I${SNLP_DIR}/include
	SNLP_LIB:= -L${SNLP_DIR}/lib -lsnlp -lsnlpF90 -Wl,-rpath,${SNLP_DIR}/lib
endif

also, what gives make -C $MEF90_DIR/Tests debug?

Commit d64abaf : It implements the double contraction of two 4th order tensors. It might be useful for the Hydro-Dev split of a cubic elasticity behavior, which I've not really been using as of now. I have not checked if the double contraction implementation gives valid results yet.

We can either merge these and print a message stating that this has not been tested, or build a test. Can we compare (AB)e and A(Be) for random A, B, and e, for instance?

I will work on building a test and add it to a new PR.

In addition, I forgot to mention that I get the following compilation error from MEF90/m_MEF90_EXO.F90:
/home/scherer/Softwares/MEF90/mef90dev/MEF90/m_MEF90_EXO.F90:350:51:

  350 |          Call MPI_Bcast(EXOelemType,MXSTLN,MPI_CHAR,0,MEF90Ctx%comm,ierr)
      |                                                   1
Error: Symbol ‘mpi_char’ at (1) has no IMPLICIT type
That error can be solved by replacing MPI_CHAR with MPI_CHARACTER. This was also included in my original (big) PR but not in the smaller ones.

done

jeanmichelscherer · 2022-08-17T20:08:15Z

There are two commits I did not include in the smaller PRs:

Commit 9b44402 : In the Makefile.include, I have to add -I${SNLP_DIR}/obj in the FC_FLAGS and C_FLAGS.

THese should not be needed since we build a shared library

If not added, I get the following compilation error:
/home/scherer/Softwares/MEF90/mef90dev/m_DefMech/m_MEF90_DefMechPlasticity.F90:1026:10:

 1026 |       use SNLPF90
      |          1
Fatal Error: Cannot open module file ‘snlpf90.mod’ for reading at (1): No such file or directory
compilation terminated.
This makes no sense what is $SNLP_DIR and what's in this directory?
in $MEF90_DIR/Makefile.include:
ifneq (${SNLP_DIR},)
	FC_FLAGS+= -DMEF90_HAVE_SNLP -I${SNLP_DIR}/include
	C_FLAGS+= -I${SNLP_DIR}/include
	SNLP_LIB:= -L${SNLP_DIR}/lib -lsnlp -lsnlpF90 -Wl,-rpath,${SNLP_DIR}/lib
endif

$SNLP_DIR points to where I installed SNLP: /home/scherer/Softwares/MEF90/snlp. It containes obj, lib, include,src, etc for SNLP.

also, what gives make -C $MEF90_DIR/Tests debug?

This line returns the following:

scherer@pasadena:~/Softwares/MEF90/mef90dev$ make -C $MEF90_DIR/Tests debug
make: Entering directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'
FFLAGS
FC_FLAGS -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g -DMEF90_HAVE_SNLP
FCFLAGS
FCPPFLAGS -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include/sieve
FCPP_FLAGS
LDFLAGS
CLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpic++ -Wall -Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -O3 -march=cascadelake -march=westmere -g
FLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpif90 -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g
PETSC_LIB -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -L/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -lpetsc -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -ltriangle -lX11 -lyaml -lpthread -lparmetis -lmetis -lexoIIv2for32 -lexodus -lml -lHYPRE -lchaco -lf2clapack -lf2cblas -lm -lnetcdf -lm -Wl,-rpath,/usr/lib/gcc/x86_64-linux-gnu/9 -L/usr/lib/gcc/x86_64-linux-gnu/9 -Wl,-rpath,/usr/lib/x86_64-linux-gnu -L/usr/lib/x86_64-linux-gnu -Wl,-rpath,/lib/x86_64-linux-gnu -L/lib/x86_64-linux-gnu -lmpi_f90 -lmpi_f77 -lgfortran -lm -lgfortran -lm -lgfortran -lm -lm -lquadmath -lm -lmpi_cxx -lstdc++ -ldl -lmpi -lopen-rte -lopen-pal -lnsl -lutil -lgcc_s -lpthread -ldl
make: Leaving directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'

bourdin · 2022-08-17T20:11:51Z

This line returns the following:

scherer@pasadena:~/Softwares/MEF90/mef90dev$ make -C $MEF90_DIR/Tests debug
make: Entering directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'
FFLAGS
FC_FLAGS -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g -DMEF90_HAVE_SNLP
FCFLAGS
FCPPFLAGS -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include/sieve
FCPP_FLAGS
LDFLAGS
CLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpic++ -Wall -Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -O3 -march=cascadelake -march=westmere -g
FLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpif90 -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g
PETSC_LIB -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -L/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -lpetsc -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -ltriangle -lX11 -lyaml -lpthread -lparmetis -lmetis -lexoIIv2for32 -lexodus -lml -lHYPRE -lchaco -lf2clapack -lf2cblas -lm -lnetcdf -lm -Wl,-rpath,/usr/lib/gcc/x86_64-linux-gnu/9 -L/usr/lib/gcc/x86_64-linux-gnu/9 -Wl,-rpath,/usr/lib/x86_64-linux-gnu -L/usr/lib/x86_64-linux-gnu -Wl,-rpath,/lib/x86_64-linux-gnu -L/lib/x86_64-linux-gnu -lmpi_f90 -lmpi_f77 -lgfortran -lm -lgfortran -lm -lgfortran -lm -lm -lquadmath -lm -lmpi_cxx -lstdc++ -ldl -lmpi -lopen-rte -lopen-pal -lnsl -lutil -lgcc_s -lpthread -ldl
make: Leaving directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'

That is inconsistent with Makefile.include Are you sure that you did not modify any of the makefiles? I don't see how you could get -DMEF90_HAVE_SNLP without -I${SNLP_DIR}/include

jeanmichelscherer · 2022-08-17T20:30:01Z

That is inconsistent with Makefile.include Are you sure that you did not modify any of the makefiles? I don't see how you could get -DMEF90_HAVE_SNLP without -I${SNLP_DIR}/include

Yes my bad, I removed both -I${SNLP_DIR}/include and -I${SNLP_DIR}/obj from the flags instead of only the latter.
Running make -C $MEF90_DIR/Tests debug outputs the following:

scherer@pasadena:~/Softwares/MEF90/mef90dev$ make -C $MEF90_DIR/Tests debug
make: Entering directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'
FFLAGS
FC_FLAGS -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g -DMEF90_HAVE_SNLP -I/home/scherer/Softwares/MEF90/snlp/include
FCFLAGS
FCPPFLAGS -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/include -I/home/scherer/Softwares/MEF90/petsc-3.3-mef90/include/sieve
FCPP_FLAGS
LDFLAGS
CLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpic++ -Wall -Wwrite-strings -Wno-strict-aliasing -Wno-unknown-pragmas -O3 -march=cascadelake -march=westmere -g
FLINKER /home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/bin/mpif90 -ffree-line-length-none -fPIC -O3 -march=cascadelake -march=westmere -g
PETSC_LIB -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -L/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -lpetsc -Wl,-rpath,/home/scherer/Softwares/MEF90/petsc-3.3-mef90/arch-linux2-cxx-opt/lib -ltriangle -lX11 -lyaml -lpthread -lparmetis -lmetis -lexoIIv2for32 -lexodus -lml -lHYPRE -lchaco -lf2clapack -lf2cblas -lm -lnetcdf -lm -Wl,-rpath,/usr/lib/gcc/x86_64-linux-gnu/9 -L/usr/lib/gcc/x86_64-linux-gnu/9 -Wl,-rpath,/usr/lib/x86_64-linux-gnu -L/usr/lib/x86_64-linux-gnu -Wl,-rpath,/lib/x86_64-linux-gnu -L/lib/x86_64-linux-gnu -lmpi_f90 -lmpi_f77 -lgfortran -lm -lgfortran -lm -lgfortran -lm -lm -lquadmath -lm -lmpi_cxx -lstdc++ -ldl -lmpi -lopen-rte -lopen-pal -lnsl -lutil -lgcc_s -lpthread -ldl
make: Leaving directory '/home/scherer/Softwares/MEF90/mef90dev/Tests'

Jean-Michel Scherer added 15 commits June 22, 2022 11:22

Implement the rotation of the stiffness tensor from the local frame (…

a3e1b71

…yaml input) to the global frame. The rotation matrix follows the Bunge (passive) convention (ZXZ) and can be defined from 3 angles (phi1, Phi, phi2) or three vectors (V1,V2,V3).

fix MPI types

a658907

make prints compatible with python3

613beef

add operator for double contraction of 4th order tensors

d64abaf

fix incompatibility issue in Hookeslaw2DSum

eb4dfb7

fix link with snlp

9b44402

Implement crystal plasticity in a rate-dependent setting (viscous Nor…

641f06a

…ton-like evolution equation). Single slip and Body-Centered Cubic (BCC) behaviours are considered.

make surfing BCs implementation match the notation used in the papers

8e22bfe

remove debugging comments

76bb487

launch visit without window when computing J-integral

99043c1

fix bug on boolean isNoPlCoupling which was not initialized (randomly…

0a57dbe

… True or False)

fix implementation of coupled damage-plasticity after commits from ma…

d5da9ed

…y/june 2020 (ATmodel and split operator)

implement rate-independent crystal plasticity based on the Gambin-Arm…

ac46331

…injon regularization (L-inf norm regularized by L-p norm): see for instance Manik, Asadkandi and Holmedal (CMAME, 2022)

fix memory leak

4e4754d

the trace of the 3D total plastic strain tensor must be zero

7472672

compute dissipation with plastic strain only

988d3a1

merge from upstream

39a5da4

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implement the rotation of the stiffness tensor from the local frame to the global frame (+ other fixes)#3

Implement the rotation of the stiffness tensor from the local frame to the global frame (+ other fixes)#3
jeanmichelscherer wants to merge 17 commits intobourdin:masterfrom
jeanmichelscherer:master

jeanmichelscherer commented Jun 22, 2022

Uh oh!

bourdin commented Jul 26, 2022

Uh oh!

jeanmichelscherer commented Jul 26, 2022

Uh oh!

bourdin commented Jul 26, 2022

Uh oh!

bourdin commented Aug 15, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

jeanmichelscherer commented Jun 22, 2022

Uh oh!

bourdin commented Jul 26, 2022

Uh oh!

jeanmichelscherer commented Jul 26, 2022

Uh oh!

bourdin commented Jul 26, 2022

Uh oh!

bourdin commented Aug 15, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

bourdin commented Aug 17, 2022

Uh oh!

jeanmichelscherer commented Aug 17, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants