7b1080fdb7
better and consistent variable name
2022-01-20 07:42:16 +01:00
1f86111f57
call SNESSetDM after DMDASNESSetFunctionLocal
...
following example ex5f.F90, seems to resolve segmentation fault
2022-01-19 22:57:22 +01:00
29530da579
use correct kind of constants for calls to MPI/PETSc
2022-01-13 13:50:30 +01:00
a7417a7ad7
default integer, PETSc integer, and MPI integer might be different
2022-01-13 12:02:33 +01:00
1ddf1e5694
support for PETSc with 64bit integers
...
compiles, but untested
2021-12-21 23:53:46 +01:00
da9fdf53d2
consistent indentation and line-spacings in reporting
2021-11-15 12:35:44 -05:00
4160c4fdb4
fix for parallel HDF5
...
if filters are applied, writing from one process does not work if the
file is opened for parallel write
2021-08-15 13:26:15 +02:00
85735605f8
more flexibility for the L in the load case
...
Note that mixed boundary conditions for L introduce an ambiguity.
Consider:
L = [[1.0, x, x],
[ 0, 0, 0],
[ 0, 0, 0]]
P = [[x, 0, 0],
[x, x, x],
[x, x, x]]
What we need is F^(n+1)=F_dot^(n+1) x Delta_t, where F_dot^(n+1) is
F_dot^(n+1)_ij = L_ik F^n_kj.
So component F_11 has contributions from L_12 and L_13. We first assume
L_12=L_13=0 and then choose F^(n+1)_12 and F^(n+1)_13 to get
P_12=P_13=0. This implicitly gives a solution for L_12 and L_13, which
is however only one out of infinitely many.
2021-07-20 07:10:28 +02:00
03b7532cc5
numpy.MaskedArray behavior
2021-07-19 23:27:10 +02:00
1c1dc9383e
symbolic names
2021-07-19 22:30:20 +02:00
f9edeb40a5
descriptive names
2021-07-17 11:50:21 +02:00
2a84aa7ae4
obvious, no need for comment
2021-07-16 20:32:21 +02:00
136a4b1377
PETSc defines are rather complicated
...
now mpi_f08 can be used on newer PETSc installations if old MPI modules
are not exposed
2021-07-09 18:48:25 +02:00
637f78bd52
old name (for PETSc < 3.15)
2021-07-09 14:50:29 +02:00
139f2c177a
use MPI_f08 if possible
...
most PETSc installations provide outdated MPI (f90 version)
MPI_COMM_WORLD is now of derived type (Fortran 08 style)
PETSC_COMM_WORLD is the plain integer (f90 style) alias.
Note that HDF5 is assumed to have f90 interfaces
2021-07-08 16:27:37 +02:00
58bc6e2ba6
avoid chained inclusions
2021-07-08 14:27:04 +02:00
5d0fc4fca3
more meaningful order
...
and intent(out) variables for read are at the front
2021-06-01 16:46:24 +02:00
0072ebfa64
polishing
2021-03-27 23:17:04 +01:00
7320120c5d
Merge branch 'development' into avoid_data_copy_restart_MPI
2021-03-26 08:58:03 +01:00
fc172921fb
unified citation style continued
2021-03-19 10:41:47 +01:00
4912342b1b
added missing arguments
2021-03-15 11:46:30 +01:00
a59af55f1a
read data by one process and broadcast it
2021-03-15 10:58:59 +01:00
adcb24d2e1
write data of average quantities non parallel
2021-03-10 16:33:02 +01:00
4dd99d4c39
solver is selected in load case, not numerics.yaml
2021-02-28 19:13:20 +01:00
d54e49e3bc
restore functionality to write non-parallel
...
not needed at the moment, but in general useful. If PETSc = parallel
should always hold, we can simplify much more
2021-02-22 13:37:21 +01:00
e855083964
systematic names
2021-02-11 14:19:04 +01:00
437d91495b
No FEsolving
2021-01-06 14:24:02 +01:00
3884549e19
store field variables as 1D array
...
first step of simplifying layout:
1) Solver translates from ip,el tuple (FEM) or cells(1),cells(2),cells(3) triple to list.
2) DAMASK iterates over all points
3) homogenization knows mapping (point,constituent) -> (instance,member)
2020-12-16 21:56:02 +01:00
afbcb82502
Merge remote-tracking branch 'origin/development' into misc-improvements
2020-12-15 10:32:59 +01:00
2dd520b4a2
P_aim should be independent from P_av
...
P_av is not defined after restart or cutback. Restart with change of
load case is probably still an issue
2020-12-15 08:06:50 +01:00
8fbadef524
print instead of write
2020-12-14 22:37:14 +01:00
97ee7e6ee5
not needed anymore
2020-12-07 17:26:50 +01:00
c16ad845bc
Merge branch 'modernize-testing' into 'development'
...
Modernize testing
See merge request damask/DAMASK!280
2020-11-16 20:10:38 +01:00
cf3a28952a
fixed tests
2020-11-16 16:38:17 +01:00
429c226008
IO is not inherited when using ifort 19
...
this is intented, public modules should only bring their own functions
and variables into global scope
2020-11-11 12:19:39 +01:00
b46a25a7d2
C capitalization
2020-11-11 11:47:23 +01:00
9119254210
materialpoint => homogenization
...
especially as prefix for global data (clear name spaces)
2020-10-24 17:26:42 +02:00
f25f7ad01e
[skip sc] consistent naming convention
2020-10-16 16:17:06 +02:00
3f3a88fdda
capitalization
2020-10-15 23:32:38 +02:00
6c79093a75
[skip sc] Fdot --> dotF
2020-10-15 23:22:37 +02:00
f8816a6e0c
Merge remote-tracking branch 'origin/development' into stress-ramp-loadcase
2020-09-24 16:57:07 +02:00
41fbc58c1b
standard conforming name
2020-09-22 13:09:12 +02:00
53ce4e07d2
forgotten renames in subfolders
2020-09-22 13:02:33 +02:00
329cc1c953
tighter tolerance for stress
2020-09-20 17:36:11 +02:00
0a7d4f61ac
Need only logical mask
...
'merge' substitutes multiplication with float mask
2020-09-20 16:54:58 +02:00
6367cb8fcb
consistent handling of boundary conditions
...
A stress boundary condition 'P' indicates the stress at the end of the
load case (same as for 'F') 'Pdot' for given increase of (technical)
strain is not implemented.
Does not change anything for the most common case of zero-stress
boundary conditions, but simplifies the specification of stress ramps
2020-09-20 15:48:41 +02:00
d584207e0a
same layout for easy diff
2020-09-20 13:01:56 +02:00
8dfb972ac1
private is already default (module wide)
2020-09-20 11:49:20 +02:00
593a40b392
Merge branch 'Fortran-simplifications' into stress-ramp-loadcase
2020-09-20 11:42:04 +02:00
053c3f39ea
solution completely relies on state defined by 'forward'
2020-09-19 23:40:17 +02:00
Martin Diehl