Newer
Older
Dave Whipp
committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
--------------------------------------------------------------------------------
DOUAR-WSMP v0.1 - 23/03/2011
--------------------------------------------------------------------------------
This is the README for DOUAR-WSMP version 0.1. This version of the code has been
modified primarily by D. Whipp and D. Guptill (Dalhousie) from a version of the
code handed down by J. Braun in early 2009. This version uses the WSMP solver,
and has diverged from the original version of the code that used the MUMPS
solver. Below, I list the changes that have been implemented in the code since
we began tracking them in July of 2009. There are still a large number of
features that need to be added, but hopefully this is helpful to new DOUAR
users.
--------------------------------------------------------------------------------
- Changes are summarized in the ChangeLog file.
- Detail on building the code can be found in the INSTALL file.
- Instructions for running a simple test model can be found in the EXAMPLE file.
--------------------------------------------------------------------------------
Known bugs:
--------------------------------------------------------------------------------
1. DOUAR-WSMP does not run reliably when linked with Open MPI. The cause of
this trouble is unclear.
2. The subdirectory DEBUG/mpilogs must exist when using debug level 1 or
higher. It seems this should only need to exist for level 2 or higher.
--------------------------------------------------------------------------------
Known limitations:
--------------------------------------------------------------------------------
1. Constant temperature boundary conditions for the thermal calculation are
applied at the top set of nodes on the free surface. Since this surface is
always a cut surface, this will not represent the actual surface
temperature.
2. The largest successful simulation using the current code version was ~2.7
million elements on 128 cores with 8GB of RAM per core.
3. Some MPI implementations, including Open MPI do not support passing of
command line options to the code. This means that the use of the different
define_bc options is not functioning properly. This should be resolved in
version 0.2.
4. The code will crash if any surface is advected into the base of the model.
It is not clear how best to handle these surfaces.
--------------------------------------------------------------------------------
Near-term planned feature additions/modifications:
--------------------------------------------------------------------------------
1. Boundary condition selection and input parameters need to be input values
from the input.txt file.
2. There is no need to refine the surfaces following isostatic adjustment, so
the code will be modified to turn that off.
3. Need to further modify the velocity boundary conditions for the HT models so
that they adjust more accurately to crustal subsidence. Rather than looking
at the isostatic subsidence array, the Moho surface position should be used
to modify the basal velocity B/Cs.
4. Strain softening in vrm.f90 should be a function so that it could be more
flexible and apply to either the friction angle or cohesion.
5. More output should be reported to the stdout to ensure proper conversion of
input values. This may already be done in some of the debugging files, but
that is not currently clear.
6. There are numerous unused variables that could be pruned from the code and
unintialized variables that need initial values.
--------------------------------------------------------------------------------
Long-term planned feature additions/modifications:
--------------------------------------------------------------------------------
1. The solver should be selectable by the user so that either MUMPS or WSMP
could be used for people who don't have a WSMP license.