[cig-commits] r18581 - short/3D/PyLith/trunk/doc/userguide/runpylith
brad at geodynamics.org
brad at geodynamics.org
Fri Jun 10 11:21:13 PDT 2011
Author: brad
Date: 2011-06-10 11:21:13 -0700 (Fri, 10 Jun 2011)
New Revision: 18581
Modified:
short/3D/PyLith/trunk/doc/userguide/runpylith/runpylith.lyx
Log:
Added info on null space with field split along with additional warnings.
Modified: short/3D/PyLith/trunk/doc/userguide/runpylith/runpylith.lyx
===================================================================
--- short/3D/PyLith/trunk/doc/userguide/runpylith/runpylith.lyx 2011-06-10 17:53:52 UTC (rev 18580)
+++ short/3D/PyLith/trunk/doc/userguide/runpylith/runpylith.lyx 2011-06-10 18:21:13 UTC (rev 18581)
@@ -1691,6 +1691,50 @@
matrix_type = aij
\end_layout
+\begin_layout Quote
+
+\series bold
+\color red
+Warning:
+\color inherit
+
+\series default
+\color none
+These settings are only available if you build PETSc with Fortran enabled
+ and the ML package.
+ These features are included in the PyLith binary packages.
+\end_layout
+
+\begin_layout Quote
+
+\series bold
+\color red
+Warning:
+\color inherit
+
+\series default
+\color none
+The split fields and algebraic multigrid preconditioning currently fails
+ when in problems with a nonzero null space.
+ This most often occurs when a problem contains multiple faults that extend
+ through the entire domain and create subdomains without any Dirichlet boundary
+ conditions.
+ The current workaround is to use the
+\color inherit
+Additive Schwarz
+\color none
+preconditioner without split fields.
+ See Section
+\begin_inset CommandInset ref
+LatexCommand ref
+reference "sec:Troubleshooting"
+
+\end_inset
+
+ for the error message encountered in this situation.
+
+\end_layout
+
\begin_layout Standard
These more advanced settings allow each component of displacement, as well
as Lagrange multipliers for fault tractions, to be preconditioned separately.
@@ -4471,13 +4515,31 @@
the datasets in external binary files with just metadata in the HDF5 files.
Both methods provide similar performance because they will use MPI I/O
if it is available.
- Storing the datasets within the HDF5 file in a parallel simulation, however,
+
+\end_layout
+
+\begin_layout Quote
+
+\series bold
+\color red
+Warning:
+\color inherit
+
+\series default
+Storing the datasets within the HDF5 file in a parallel simulation, however,
requires that the HDF5 library be configured with the
\family typewriter
--enable-parallel
\family default
option.
- Accessing the datasets for additional analysis or visualiation is identical
+
+\color none
+The binary PyLith packages include this feature and it is a default setting
+ in building HDF5 via the PyLith Installer.
+\end_layout
+
+\begin_layout Standard
+Accessing the datasets for additional analysis or visualiation is identical
in the two methods because the use of external data files is completely
transparent to the user except for the presence of the additional files.
\end_layout
@@ -4589,6 +4651,13 @@
\begin_layout Section
Tips and Hints
+\begin_inset CommandInset label
+LatexCommand label
+name "sec:Tips:Hints"
+
+\end_inset
+
+
\end_layout
\begin_layout Subsection
@@ -4673,6 +4742,13 @@
\begin_layout Subsection
Troubleshooting
+\begin_inset CommandInset label
+LatexCommand label
+name "sec:Troubleshooting"
+
+\end_inset
+
+
\end_layout
\begin_layout Itemize
@@ -4754,19 +4830,19 @@
\end_layout
\begin_layout Quote
-This error occurs when trying to use the 32-bit linux binary on 64-bit linux
- systems.
+This error occurs when trying to use the 32-bit linux binary on some 64-bit
+ linux systems.
One of the Python packages PyLith uses does not know how to determine the
system architecture at runtime.
The workaround is:
\end_layout
\begin_layout Enumerate
-Go to the lib/python2.5/site-packages directory.
+Go to the lib/python2.6/site-packages directory.
\end_layout
\begin_layout Enumerate
-Unzip merlin-1.7-py2.5.egg (if it is a file and not a directory).
+Unzip merlin-1.7-py2.6.egg (if it is a file and not a directory).
\end_layout
\begin_layout Enumerate
@@ -4779,11 +4855,48 @@
\end_layout
\begin_layout Enumerate
-If merlin-1.7-py2.5.egg is a file, rezip merlin.
- Go to the site-packages directory and enter "zip -r merlin-1.7-py2.5.egg merlin".
+If merlin-1.7-py2.6.egg is a file, rezip merlin.
+ Go to the site-packages directory and enter "zip -r merlin-1.7-py2.6.egg merlin".
\end_layout
\begin_layout Itemize
+
+\family typewriter
+-- Solving equations.
+\begin_inset Newline newline
+\end_inset
+
+[0]PETSC ERROR: ---------------- Error Message -------------------------------
+
+\begin_inset Newline newline
+\end_inset
+
+[0]PETSC ERROR: Detected zero pivot in LU factorization
+\begin_inset Newline newline
+\end_inset
+
+ see http://www.mcs.anl.gov/petsc/petsc-as/documentation/faq.html#ZeroPivot!
+\end_layout
+
+\begin_layout Quote
+This usually occurs when the null space of the system Jacobian is nonzero,
+ such as the case of a problem without Dirichlet boundary conditions on
+ any boundary.
+ If this arises when using the split fields and algebraic multigrid precondition
+ing and no additional Dirichlet boundary conditions are desired, then the
+ workaround is to revert to using the Additive Schwarz preconditioning without
+ split fields as discussed in Section
+\begin_inset CommandInset ref
+LatexCommand ref
+reference "sec:petsc:options"
+
+\end_inset
+
+.
+
+\end_layout
+
+\begin_layout Itemize
PyLith crashes with a bus error.
\end_layout
More information about the CIG-COMMITS
mailing list