83cf062318
better define only once
2019-12-21 11:19:33 +01:00
ac182ef536
less variable, same style
2019-12-21 11:13:05 +01:00
6678770c43
public variables not needed anymore
2019-12-21 10:43:36 +01:00
7baf4e7f53
cleaning
2019-12-21 10:37:18 +01:00
91ad5092f4
preparing removal of old output data
2019-12-21 10:31:19 +01:00
4dc5dac831
was only needed for postResults
2019-12-21 10:20:50 +01:00
af6973adf6
HDF5 out is always on
2019-12-21 10:13:52 +01:00
6425c37f2d
use default string length
...
256 characters are more than enough for string values and keys
2019-12-21 10:07:37 +01:00
19a45d9c2b
cleaning of source modules:
...
- offset and instance are only used locally
- output names are needed only temporarly
- HDF5 is always enabled
2019-12-21 07:55:42 +01:00
41650cdd11
use standard string length
2019-12-21 07:38:17 +01:00
b10e862be8
the compiler can do the counting
2019-12-21 07:34:40 +01:00
f0d3b29b82
Merge branch 'development' into MiscImprovements
2019-12-21 06:53:56 +01:00
befd4e2adf
avoid imports
2019-12-19 07:49:53 +01:00
f0ad075802
fixed indentation
2019-12-19 00:01:44 +01:00
98e606d6f0
correct type
2019-12-19 00:00:00 +01:00
4b6388fbb2
always use HDF5 output
2019-12-18 20:05:51 +01:00
63fc125398
include origin do DADF5
...
requires updat of file version
2019-12-13 09:12:07 +01:00
ee56b82c8a
avoid spaces
2019-12-12 23:22:37 +01:00
a0a99c526b
some remaining IntOut format strings
2019-12-11 19:28:31 +01:00
9e801928b0
Merge branch 'development' into noPostResults
2019-12-11 09:28:45 +01:00
9b67ead62f
removed postResults completely
2019-12-11 00:10:02 +01:00
8fb8e9be6e
write temperature to DADF5
2019-12-10 20:25:19 +01:00
db91803b80
cleaning
2019-12-10 17:52:37 +01:00
ac35759c2f
not a module subroutine
2019-12-10 17:25:51 +01:00
5681e661e2
DADF5-results replaces postResults
2019-12-10 07:31:09 +01:00
4be7aa990c
HDF5 results output for constitutive damage models
2019-12-10 06:45:00 +01:00
5abe27ab60
only locally used
2019-12-09 06:24:37 +01:00
1c18086402
not needed anymore
2019-12-09 05:58:27 +01:00
07ebd8d1b3
only damage/thermal 'homogenization' postResults is currently needed
2019-12-09 05:48:37 +01:00
acc252ea5b
thermal/damage constitutive (i.e. source) results are not tested
2019-12-09 05:38:15 +01:00
c2e91ae5e1
copy and paste error, wrong array was set
2019-12-08 20:52:05 +01:00
b0ce324213
Merge remote-tracking branch 'origin/development' into simple-int-formatting
2019-12-08 20:24:05 +01:00
fbe2228ade
avoid string length mismatch
2019-12-07 20:22:55 +01:00
3e269f0419
Merge branch 'only-use-rotation-class' into 'development'
...
Only use rotation class
See merge request damask/DAMASK!110
2019-12-07 17:33:34 +01:00
4eacc1d16d
polishing
2019-12-07 15:45:37 +01:00
23944de6de
less variables for better readability
2019-12-07 15:20:04 +01:00
5a6111ec69
Merge branch 'incs-no-leading-zero' into 'development'
...
Incs no leading zero
See merge request damask/DAMASK!108
2019-12-07 11:32:37 +01:00
7ecb0245ec
simplified (use novel i0 format specifier)
2019-12-07 11:12:46 +01:00
137b235bc1
i0 format specifier easier then hand-written IO_intOut
...
using variable pStringLen allows string assignment without length issues
2019-12-07 11:06:39 +01:00
04492eeb0a
Merge branch 'almost-no-postResults' into only-use-rotation-class
2019-12-05 19:42:50 +01:00
78e96993df
Merge branch 'almost-no-postResults' into only-use-rotation-class
2019-12-05 19:41:32 +01:00
f585e1a553
not used
2019-12-05 11:20:05 +01:00
ab1f0dc16b
submodules allow inter-module communication
2019-12-04 22:31:15 +01:00
7d6a57dc17
Merge branch 'only-use-rotation-class' into plasticity-submodule
2019-12-04 19:02:01 +01:00
cb0d39eee6
not needed anymore
2019-12-04 19:00:56 +01:00
58a190ede8
Merge branch 'development' into incs-no-leading-zero
2019-12-04 05:19:38 +01:00
93da6d3f63
was not used
2019-12-02 23:07:01 +01:00
c36a5bdfbb
bugfix: calculation of size of postResults was wrong
2019-12-02 23:06:28 +01:00
9882c3532a
avoid use of low-level HDF5 routines
2019-12-02 22:26:08 +01:00
4ee2e551b8
migrating to submodules
2019-12-02 22:15:34 +01:00
226b715c46
file names reflect hierarchical structure
2019-12-02 21:38:41 +01:00
e52a747774
submodules allow inter-module communication
2019-12-02 20:57:18 +01:00
f5292019e5
use rotation class
2019-12-02 20:23:50 +01:00
8a9d3f8d6d
avoid code duplication
2019-12-02 20:09:27 +01:00
83453d10ef
use rotation class for consistent handling of rotations
2019-12-02 16:37:22 +01:00
be099e38c2
might be of use
2019-12-02 16:22:27 +01:00
8189b50509
same functionality but tested
2019-12-02 12:58:23 +01:00
2969a53338
not used/needed anymore
2019-12-02 12:54:45 +01:00
f85ee7d7fb
plasticity is postResults-free
2019-12-02 12:39:44 +01:00
2b3b26087f
consistent names
2019-12-01 20:07:16 +01:00
05aa38cc71
using short names
2019-12-01 15:05:49 +01:00
bb152a171a
consistent naming among plasticity laws
2019-12-01 10:32:45 +01:00
1cf1da5bcc
consistent naming
2019-12-01 09:35:44 +01:00
33c090d2ac
pointers for better readability
2019-12-01 08:55:24 +01:00
afad2f2dce
short and consistent naming
2019-12-01 08:39:30 +01:00
df4b642400
polishing
2019-12-01 08:33:04 +01:00
07fb915a7e
postResults not needed anymore
2019-12-01 08:27:56 +01:00
c7d675a1e9
not needed anymore
2019-11-30 16:03:18 +01:00
d93fb12e8c
remove extra output
...
current implementation was not very general, no separation of different
dislocation types.
if output is needed, re-implemenation is straight forward
2019-11-30 14:36:54 +01:00
8f77843b0f
improved naming
...
- clearer variable names
- no '-' in attributes
- single source of truth
2019-11-30 09:01:37 +01:00
dc3fc8f70c
do not clutter with leading zeros
2019-11-29 17:00:48 +01:00
530f2d3c08
Merge branch 'no-crystallite' into almost-no-postResults
2019-11-25 21:51:15 +01:00
67c8d3899d
crystallite output fully removed
2019-11-25 08:44:44 +01:00
127a29ab1a
cleaning
2019-11-24 13:42:19 +01:00
97474e05a8
not needed anymore
2019-11-24 11:01:04 +01:00
e20477099e
no crystallite output
2019-11-24 09:46:46 +01:00
50b48b8bf7
IP neighbourhood deprecated
...
trivial for grid and, hence, not written out.
Test for marc/abaqus/mesh would make more sense
2019-11-24 07:54:34 +01:00
a6a6519678
more local data handling
2019-11-24 07:44:17 +01:00
bb73a08cf4
better encapsulation
...
for results operations, do not rely directly on HDF5 module
2019-11-24 07:27:05 +01:00
1c10459a5a
also use fast operation order
2019-10-30 21:09:17 +01:00
ffb112b0d8
Merge branch 'development' into DADF5-improvements
2019-10-30 17:16:54 +01:00
368d4debda
calculate average nodal position directly
...
no need to average and have second memory access
2019-10-30 14:49:08 +01:00
40507846c9
cell displacement are between nodes which start at (0,0,0)
2019-10-30 14:47:42 +01:00
1d2d7d1ea3
remove public variables
...
they make code unreadable and unmaintainable
2019-10-29 23:24:12 +01:00
0c6b7e1112
not nice, but working ...
2019-10-29 23:15:02 +01:00
402e681cf5
initialization of arrays for fourier transform more reasonable
...
- padding entries in real data array need to be zero
- all values of the fourier data arrays are explicitly set
2019-10-29 16:18:58 +01:00
1ae33cf215
avoid one loop and use filtered frequencies
2019-10-29 09:15:35 +01:00
f451d19c1e
forall is obsolete
...
some calculation simplified
2019-10-28 21:27:57 +01:00
c47f516769
do not correct fluctuations
2019-10-28 21:16:43 +01:00
ed8834e371
maybe we can do some guessing for CPFEM ..
2019-10-28 13:38:30 +01:00
d1f719d3cf
separation of "age" and "forward" not useful
2019-10-28 13:36:36 +01:00
8de6fcfc24
unifying names
2019-10-28 13:29:32 +01:00
ab4a98fb53
make use of HDF5 and structure data
2019-10-28 13:17:05 +01:00
10f2c53256
clearly distiguish
...
- DAMASK as a materialpoint model (CPFEM, not fitting name)
- solver
2019-10-25 08:30:12 +02:00
07d8379313
further untangling
2019-10-25 07:55:23 +02:00
cf8b16c066
better have one function per task
...
but separating ageing and forward for the solver is not trivial due to
guessing
2019-10-25 07:31:09 +02:00
eb4cad17b5
avoid public variables
2019-10-25 00:53:48 +02:00
378729f5c4
clearer structure
2019-10-25 00:42:59 +02:00
9db5cbb667
trying to get rid of a global "restartWrite" variable
2019-10-24 22:50:30 +02:00
3d44e7598e
works only for MSC.Marc
2019-10-24 13:57:14 +02:00
0bc4326618
better have one function per task
2019-10-24 13:46:36 +02:00
7c0bb2fdf8
consistent description
2019-10-24 13:37:40 +02:00
71def397d5
no need to age before writing restart data
2019-10-24 13:14:40 +02:00
176f6028cd
CPFE2 writes current variables (like grid solvers)
2019-10-24 13:06:42 +02:00
c1cf64bf00
was not used/tested
2019-10-24 12:15:22 +02:00
c0c0831c92
multiple keyword are more confusing then helpful
2019-10-24 06:37:28 +02:00
50769b52a5
don't continue if calculation did not converge
2019-10-24 06:32:46 +02:00
0483fc7b3f
small functions with one task are better
2019-10-24 06:16:42 +02:00
534073358e
arguments are better than public variables
2019-10-23 23:06:47 +02:00
67228b46a6
modularize
2019-10-23 22:50:01 +02:00
52190382ac
cleaning
2019-10-23 22:42:46 +02:00
dfe65e406d
restart for commercial FE solvers untested
...
better no code than unreliable code. If functionality is needed,
implement it without the use of publicly writeable variables
2019-10-23 22:13:31 +02:00
8a4a5a8bef
Merge remote-tracking branch 'origin/development' into age-separately-2
2019-10-23 22:05:30 +02:00
e1d820bff1
does not support restart
2019-10-23 22:04:50 +02:00
60a19c020d
cleaning
2019-10-23 21:17:56 +02:00
ac0149e55e
not needed (is initialized to 0)
2019-10-21 11:48:35 +02:00
0b4605361b
plotv only needed for postResults
2019-10-20 19:23:50 +02:00
872c85112f
bugfix, Lp not available anymore for crystallite
2019-10-20 14:02:12 +02:00
b96bd71b09
postResults not needed anymore
2019-10-20 13:24:47 +02:00
f93336b072
postResults for isotropic not needed anymore
2019-10-20 13:21:51 +02:00
11993a3ad1
phasing out crystallite output
...
some outputs are still needed for a few remaining tests.
2019-10-20 12:03:08 +02:00
fe94718bcd
avoid accessing null-pointer
2019-10-20 10:32:14 +02:00
65eb4e364f
phasing out postResults
2019-10-19 23:14:17 +02:00
603973e963
correct output for multipe integration points
2019-10-19 19:55:00 +02:00
3de605d782
better name
2019-10-19 16:06:01 +02:00
1253c6abf1
Merge remote-tracking branch 'origin/development' into labeled-systems
2019-10-18 19:31:08 +02:00
2c0264191e
typo!
2019-10-18 17:13:19 +02:00
de912c22b4
Merge branch 'development' into HDF5-spectral-displacements
2019-10-18 12:54:12 +02:00
fea6c268a8
Merge branch 'nonlocal-again-with-Marc' into 'development'
...
Nonlocal again with marc
See merge request damask/DAMASK!99
2019-10-18 11:30:15 +02:00
7ddb724861
[skip ci] corrected wrongly copied comment
2019-10-18 11:27:37 +02:00
cebbfc906d
Merge branch 'hex-reorder-2ndtry' into development
2019-10-17 14:58:43 +02:00
b9027e3257
checking size of unallocated array does not work
2019-10-17 08:40:05 +02:00
88df7f2957
store cell normal directions
2019-10-17 07:48:57 +02:00
b386dc73b2
enable to non-transposed tensor data
...
usually, we store per data per cell, i.e. len(shape(x)) == 3 means x is
a tensor.
Due to the use of transposed tensors (due to column-major storage order
in Fortran), we usually want to store the transpose of (3x3) tensors.
Now the default can be changed
2019-10-17 07:46:20 +02:00
008f717c08
avoid reading from file
2019-10-17 06:10:00 +02:00
2b68c108f0
read from memory, not from file
2019-10-17 05:54:08 +02:00
f33a99d125
polishing
2019-10-17 00:21:48 +02:00
3f481e1cea
corrected volume calculation and write to DADF5.
...
follows https://www.osti.gov/servlets/purl/632793/
2019-10-16 22:00:25 +02:00
9b5545229f
leftover comment ...
2019-10-16 21:13:43 +02:00
95ecc05cb0
better test ...
2019-10-16 21:13:26 +02:00
0ae0e23325
volume calculation was wrong
...
could result in negative volumes
2019-10-16 20:56:31 +02:00
dbe15f88f2
bugfix: forgot to read file
...
first draft of nonlocal functions for area normal, area, and volume
2019-10-16 17:19:19 +02:00
4e213514bd
cleaning
2019-10-16 13:04:20 +02:00
3a1c4f95c3
some polishing
2019-10-16 08:40:27 +02:00
3abb549eab
better encapsulation
...
inputRead fully parses the input file and gives back the required data
2019-10-16 04:45:38 +02:00
0f77d2efdb
consistent use of subroutines
...
clearly reveal input(out) arguments
2019-10-16 04:22:21 +02:00
ffb4b2a455
theMesh is not useful
...
maybe a shared mesh class becomes useful when Abaqus and PETSc mesh
require similar functionality
elem type can be reported directly
2019-10-16 03:57:51 +02:00
1d70d9b6ae
mesh type not very beneficial
2019-10-15 21:58:49 +02:00
c4db2841ab
further cleaning
2019-10-15 14:16:03 +02:00
aab19af131
consistent use of subroutines, not functions.
...
grouping file reading into meta function
2019-10-15 12:54:46 +02:00
91691744c6
Merge branch 'Marc-No-nonlocal' into 'development'
...
Marc no nonlocal
See merge request damask/DAMASK!98
2019-10-15 10:35:37 +02:00
9ee709d214
polishing
2019-10-14 15:09:16 +02:00
1726d95a56
wrong macro was defined
2019-10-14 14:39:53 +02:00
dcbd7624dd
array order was wrong
2019-10-14 11:25:48 +02:00
57af822396
better readable
2019-10-14 10:36:59 +02:00
57fef8fa57
consistent naming
2019-10-14 10:08:35 +02:00
fae4546cfd
polishing
2019-10-14 09:37:31 +02:00
513b1906f6
bugfixes:
...
cell definition was not stored correctly due to wrong indexing
2019-10-14 08:23:21 +02:00
c41d8eb257
use new data structure to calculate cell node coordinates
2019-10-13 23:34:03 +02:00
2b65c888c4
avoid public variables
2019-10-13 22:16:42 +02:00
33e639426a
polishing
2019-10-13 21:45:08 +02:00
d03efade06
better calculate nodes later
2019-10-13 21:09:49 +02:00
369cae5332
no public variables
2019-10-13 20:52:49 +02:00
6456f9891b
avoid public variables
2019-10-13 20:34:20 +02:00
c2a8701940
further removal of deprecated stuff
2019-10-13 19:55:25 +02:00
4999aa4e14
more cleaning
2019-10-13 19:41:55 +02:00
e0cb1a87cd
cleaning
2019-10-13 19:22:57 +02:00
e3b16639bf
native integer needs to match
...
otherwise, results are wrong. Therefore, we need to use our own HDF5
library since MSC provides one for 4 byte integers
2019-10-13 18:20:54 +02:00
004abc2d4e
cell node definition needs to be stored
2019-10-13 13:28:04 +02:00
ed1d06d6f1
make functions independent of file state
2019-10-13 11:53:35 +02:00
9ea91b84e8
easier to read and without file access
2019-10-13 11:48:27 +02:00
680ed535d7
avoid file operations and line labels
2019-10-13 10:42:34 +02:00
d6b6096007
avoid global variables
2019-10-12 22:40:00 +02:00
7611513bb8
strange indentation
2019-10-12 22:18:55 +02:00
4c9bc326f6
better read file only once
2019-10-12 21:58:26 +02:00
e6d25bfdab
almost no shared functionality
2019-10-12 19:24:03 +02:00
d8d99f3694
geometry folder needs to be created earlier
2019-10-12 16:39:44 +02:00
fabab08936
did not run without HDF5 support
2019-10-12 16:25:59 +02:00
8cebf8a10d
needed
2019-10-12 15:50:10 +02:00
734e6ef15f
writing initial coordinates to DADF5
2019-10-12 15:03:26 +02:00
e85e13b380
[skip ci] only corrections to comments made
2019-10-11 15:21:29 +02:00
bf35589d66
DOI was pointint to the correction
2019-10-11 12:31:46 +02:00
b502300ffc
get labels of slip and twin systems
...
this info is for HDF5 output
2019-10-10 13:11:02 +02:00
1d51443607
polishing/unifying names
2019-10-10 12:26:45 +02:00
02d7ead3ab
better readable
2019-10-10 12:10:13 +02:00
cf0e279d93
labels were not used
...
better use automated labeling of all systems
2019-10-10 10:48:55 +02:00
a2a05158f2
not existing anymore
2019-10-09 01:32:19 +02:00
df3bb7c559
cleaning
2019-10-08 23:53:16 +02:00
90c03d94d1
further separation for clearer structure
2019-10-08 19:08:29 +02:00
b647245e39
general polishing
2019-10-08 18:52:34 +02:00
7d438d3868
not needed
2019-10-08 17:37:30 +02:00
040cd3e35d
no public variables
2019-10-08 17:26:02 +02:00
16fd608da6
more reasonable name
2019-10-08 17:12:53 +02:00
098f2903ea
not used anymore
2019-10-08 17:04:27 +02:00
5e79c360e9
no need for double definition
2019-10-08 10:03:03 +02:00
04272f88d5
untangling
...
- avoid public variables
- openMP in initialization hardly useful
- structure of init should reflect tasks:
1) reading
2) discretization
3) nonlocal stuff
2019-10-08 09:30:32 +02:00
795939c989
tests for re-ordered systems + new PETSc 3.11
2019-10-07 20:23:17 +02:00
2d7293726c
more useful order
2019-10-07 18:04:29 +02:00
7f403ad50e
avoid global variables
2019-10-05 20:23:33 +02:00
18b8e71f69
polishing
2019-10-05 20:09:01 +02:00
1d35699884
ip volume is only needed by plastic nonlocal
2019-10-05 19:46:08 +02:00
076aa3f72b
not needed
2019-10-05 17:37:14 +02:00
019f0556e6
better have explicit arguments
2019-10-05 17:18:21 +02:00
0b29d08109
bump to DADF5 v0.3
...
contains displacements (for grid solver)
2019-10-03 19:02:17 +02:00
fd3f8e2cc7
fixed precision even if compiler flag is not set
2019-09-29 15:59:10 -07:00
d47dff9dd8
typo
2019-09-29 14:18:21 -07:00
ad75ebd973
polishing
2019-09-29 11:26:57 -07:00
56c0f30687
tried to send to much data
2019-09-29 10:34:55 -07:00
8e5fd7c5e8
nodal discplacements for MPI
...
needs improvement with respect to readability and placement of origin
2019-09-29 10:04:07 -07:00
e04b074f3c
easier to read
2019-09-28 18:32:17 -07:00
c71374a3f3
do not write nodes shared over several processes
...
- probably useful only for grid solver, FEM have connectivity matrix
anyway
- name could be improved (in fact, the index is the last not-shared
node)
2019-09-28 18:08:15 -07:00
3206ee04bc
nodal displacements for grid solver
...
currently for non-MPI only
2019-09-27 20:38:02 -07:00
5ba6b6c244
name reflects planned functionality:
...
should update node and cell center coordinates
2019-09-27 15:02:36 -07:00
cc54139dae
no need for being compatible
2019-09-27 14:54:02 -07:00
fc6b603e3d
names reflect functionality
2019-09-27 14:48:51 -07:00
e311c913b8
avoid public variable without write protection
2019-09-27 14:44:28 -07:00
13c62b890b
module name should reflect file name
2019-09-27 14:34:34 -07:00
032b65fe65
not needed
2019-09-27 14:33:56 -07:00
8fd9341e39
not needed
2019-09-27 14:23:19 -07:00
114f21e8f7
setter for current nodal coordinates
2019-09-27 14:07:34 -07:00
2bb6f74de9
use new data structure
2019-09-27 14:07:03 -07:00
2e63affe47
[skip ci] better to read
2019-09-25 12:48:16 -07:00
9ebd91312a
avoid size(xxx) of unallocated xxx
2019-09-25 12:21:57 -07:00
fc090a1960
[skip ci] indentation only
2019-09-25 12:05:37 -07:00
116f43f401
[skip ci] indentation hick-up
2019-09-23 22:15:13 -07:00
ffcb7a0a24
aim of the calculation clearer
...
don't know if brackets are honored here
2019-09-23 09:31:14 -07:00
da041f43bc
floating point arithmetic refinement
...
Intel 18.4 could not detect 0.0
2019-09-23 09:29:38 -07:00
96811f77ca
not used
2019-09-23 06:50:25 -07:00
0620eae797
less code
2019-09-23 06:48:39 -07:00
611cc9bcae
preprocessor does not allow spaces
2019-09-23 06:48:20 -07:00
48e0a7ad97
no floating point comparison for equality
2019-09-23 06:35:02 -07:00
b8204015f5
wrong type (math_range returns integers)
2019-09-23 06:29:32 -07:00
02fef16b38
used only by MARC/Abaqus
2019-09-23 06:28:10 -07:00
78344c01ac
avoid floating point comparison
2019-09-23 05:37:36 -07:00
da15ff85f3
tests now operational
2019-09-23 00:11:45 -07:00
285dab4451
self consistent solutions
2019-09-22 23:53:56 -07:00
bbb2441cc2
preparing more tests
2019-09-22 22:08:19 -07:00
0655ef2c90
small precision adjustments required
...
one in a Mio might have degenerated precision...
Also now finally testing core functionality
2019-09-22 21:58:18 -07:00
c1398e5fa4
more robust agains overflow + tests
2019-09-22 21:41:00 -07:00
40099fd581
typo
2019-09-22 17:46:37 -07:00
9fd65b0617
need to relax tolerance for case 3
2019-09-22 16:57:16 -07:00
d1566aa5bf
GNU 8 has issues with IEEE_infinite
...
errors occur when running aggressively optimized code. GNU 9.1 seems to
be ok
2019-09-22 16:05:41 -07:00
c0210659b6
statUnit was not opened, stat file was written to STDERR
2019-09-22 15:56:37 -07:00
26bdcd64b9
this might resolve the occasional MPI restart test failuers
2019-09-22 15:34:05 -07:00
4ff292ba70
IEEE infinite causes problems with older GNU
2019-09-22 14:46:27 -07:00
171a93ca30
also small negative values cause problems
...
Logic differs from the original formulation now
2019-09-22 14:17:17 -07:00
6b5b0fae22
mixed up real and aimag part in quaternion
...
+ some tests
2019-09-22 12:10:39 -07:00
25c9bb1cd7
testing some corner cases
2019-09-22 11:29:34 -07:00
62174a8978
testing all conversions
...
using norm(Delta q) ensures that the quality of the conversion is
tested, not the numerical properties of any orientation representation.
2019-09-22 11:15:27 -07:00
6e5b9ca665
unused variables
2019-09-22 08:54:59 -07:00
Martin Diehl