[cig-commits] [commit] master: preparing manual for release 1.2 (7da858d)
cig_noreply at geodynamics.org
cig_noreply at geodynamics.org
Wed Jan 7 13:34:14 PST 2015
Repository : https://github.com/geodynamics/axisem
On branch : master
Link : https://github.com/geodynamics/axisem/compare/c7f867511ca351fbf339112efbdc0143d1f725e8...7da858ddaa27b15877520b906542cfb79da4be50
>---------------------------------------------------------------
commit 7da858ddaa27b15877520b906542cfb79da4be50
Author: martinvandriel <vandriel at erdw.ethz.ch>
Date: Wed Jan 7 22:33:29 2015 +0100
preparing manual for release 1.2
>---------------------------------------------------------------
7da858ddaa27b15877520b906542cfb79da4be50
...weak_scaling_new.pdf => db_generation_cost.pdf} | Bin 13352 -> 13757 bytes
.../{manual_axisem1.1.tex => manual_axisem1.2.tex} | 61 +++++++++++++++------
MANUAL/memory_usage_new.pdf | Bin 10819 -> 11278 bytes
3 files changed, 44 insertions(+), 17 deletions(-)
diff --git a/MANUAL/COMPUTATIONAL_COST/weak_scaling_new.pdf b/MANUAL/db_generation_cost.pdf
similarity index 57%
copy from MANUAL/COMPUTATIONAL_COST/weak_scaling_new.pdf
copy to MANUAL/db_generation_cost.pdf
index 10b393a..8412346 100644
Binary files a/MANUAL/COMPUTATIONAL_COST/weak_scaling_new.pdf and b/MANUAL/db_generation_cost.pdf differ
diff --git a/MANUAL/manual_axisem1.1.tex b/MANUAL/manual_axisem1.2.tex
similarity index 93%
copy from MANUAL/manual_axisem1.1.tex
copy to MANUAL/manual_axisem1.2.tex
index 0195959..7527965 100644
--- a/MANUAL/manual_axisem1.1.tex
+++ b/MANUAL/manual_axisem1.2.tex
@@ -108,7 +108,7 @@ COMPUTATIONAL INFRASTRUCTURE FOR GEODYNAMICS (CIG)
\hfill{\Huge \fontfamily{\sfdefault}\selectfont User Manual \\
% FILL: manual version
% e.g. 1.0
-\raggedleft \huge \fontfamily{\sfdefault}\selectfont Version {1.1}\\}
+\raggedleft \huge \fontfamily{\sfdefault}\selectfont Version {1.2}\\}
%AUTHOR(S) & WEBSITE%
\null
@@ -148,7 +148,7 @@ Tarje Nissen-Meyer\\Martin van Driel\\Simon St\"{a}hler\\Kasra Hosseini\\Stefani
\begin{minipage}{0.6\textwidth}
\begin{center}
\title{}
- \LARGE{ \textbf{\sc AxiSEMv1.1 Manual}}
+ \LARGE{ \textbf{\sc AxiSEMv1.2 Manual}}
\vspace*{0.6cm}\\
{\large
Tarje Nissen-Meyer\textsuperscript{1},
@@ -621,17 +621,20 @@ than comparable 3D methods.
%\includegraphics[height=50mm]{mesher_comp_cost.pdf}
\includegraphics[height=50mm]{memory_usage_new.pdf}
\hspace{5mm}
- \includegraphics[height=50mm]{solver_totalcpu.pdf}
+ %\includegraphics[height=50mm]{solver_totalcpu.pdf}
+ \includegraphics[height=50mm]{db_generation_cost.pdf}
\end{center}
On the left, you may deduct the mesher's RAM occupation as a function of frequency. Going
towards very high resolution (around and above 1Hz), you will need a rather fat node (>
16GB RAM) for the (shared memory) meshing.
On the right, we depict the computational cost
-associated with the solver to compute seismograms of 30 min length. The relation between
-seismic period and CPU-hours is an approximate proxy to estimate how many cores and
-wall-clock time is optimal for your infrastructure. Scaling of CPU hours is approximately
-to the third power of the maximum frequency.
+associated with the solver to compute seismograms of one hour length for Earth and Mars
+and for the second order Newmark as well as the fourth order symplectic time scheme. The
+relation between seismic period and CPU-hours is an approximate proxy to estimate how many
+cores and wall-clock time is optimal for your infrastructure. Scaling of CPU hours is
+approximately to the third power of the maximum frequency. As a rule of thumb, using the
+symplectic scheme is advisable when propagating waves for more than 100 wavelengths.
\begin{center}
\includegraphics[height=50mm]{COMPUTATIONAL_COST/strong_scaling_new.pdf}
@@ -832,9 +835,18 @@ postprocessing script sums the results to get correct seismograms.
\subsection{Change source time function}
\verb|SOLVER/inparam_advanced|, change parameter \verb|SOURCE_FUNCTION||:\\
-AxiSEM generally calculates displacement seismograms. The source time function selected here is equivalent to the moment function $m(t)$. Note that to calculate seismograms similar to those of an earthquake (i.e. with a persistent displacement at the source), the setting \verb|errorf| has to be used. Note that that is different to the literature that usually defines the source time function as the moment rate function $\dot{m}(t)$. However, it is consistent with SpecFEM.\\
-To create seismograms with a flat, zero-phased source spectrum, use the setting \verb|dirac_0|. Seismograms calculated with this setting can be convolved with other source functions in AxiSEM postprocessing or manually with a program of your choice.\\
-However, note that seismograms or wavefields calculated with a \verb|dirac_0| STF will contain numerical noise. Therefore, visualization and wavefield plotting should be done using \verb|errorf| or one of the Gaussian STFs.
+AxiSEM generally calculates displacement seismograms. The source time function selected
+here is equivalent to the moment function $m(t)$. Note that to calculate seismograms
+similar to those of an earthquake (i.e. with a persistent displacement at the source), the
+setting \verb|errorf| has to be used. Note that that is different to the literature that
+usually defines the source time function as the moment rate function $\dot{m}(t)$.
+However, it is consistent with SpecFEM.\\
+To create seismograms with a flat, zero-phased source spectrum, use the setting
+\verb|dirac_0|. Seismograms calculated with this setting can be convolved with other
+source functions in AxiSEM postprocessing or manually with a program of your choice.\\
+However, note that seismograms or wavefields calculated with a \verb|dirac_0| STF will
+contain numerical noise. Therefore, visualization and wavefield plotting should be done
+using \verb|errorf| or one of the Gaussian STFs.
\subsection{Change seismogram length or sampling rate}
@@ -851,8 +863,8 @@ done with \textit{ObsPy} or another tool that supports filtering.
\subsection{Include lateral heterogeneities (2.5D simulation)}
\verb|SOLVER/inparam_basic|, change parameter \verb|LAT_HETEROGENEITY| to true:\\
The actual heterogeneity model is set in \verb|SOLVER/inparam_hetero|.
-This functionality is currently under development and not documented. See the example files \verb|SOLVER/inparam_hetero.TEMPLATE| for an idea of what to do.
-
+This functionality is currently under development and not documented. See the example
+files \verb|SOLVER/inparam_hetero.TEMPLATE| for an idea of what to do.
\newpage
@@ -888,6 +900,17 @@ doi:10.1111/j.1365-246X.2008.03813.x\\
\textit{Full-wave seismic sensitivity in a spherical Earth},
Ph.D. thesis, Princeton University
(This includes refs (2)-(4) and more details.)\\
+
+(6) Martin van Driel and Tarje Nissen-Meyer (2014),
+\textit{Seismic Wave Propagation in Fully Anisotropic Axisymmetric Media},
+Geophysical Journal International 199 (2), 880â893.\\
+doi:10.1093/gji/ggu269.\\
+
+(7) Martin van Driel and Tarje Nissen-Meyer (2014),
+\textit{Optimized Viscoelastic Wave Propagation for Weakly Dissipative Media},
+Geophysical Journal International 199 (2) 1078â1093.\\
+doi:10.1093/gji/ggu314.\\
+
%
% (5) Jean-Paul Ampuero, Tarje Nissen-Meyer (2011),
% \textit{High-order conservative time schemes in spectral-element methods
@@ -897,22 +920,26 @@ Ph.D. thesis, Princeton University
\noindent \textbf{Other references:}\vspace*{0.2cm}
-(6) Deville, M. O., Fischer, P. F., Mund, E. H. (2002),
+(8) Martin van Driel, Lion Krischer, Simon Stähler, Kasra Hosseini, and Tarje Nissen-Meyer (2015),
+\textit{Instaseis: Instant Global Broadband Seismograms Based on a Waveform Database},
+To be submitted to Solid Earth\\
+
+(9) Deville, M. O., Fischer, P. F., Mund, E. H. (2002),
\textit{High-Order Methods for Incompressible Fluid Flow},
Vol. 2, Cambridge monographs on Sppl. \& Comp. Math., Cambridge University Press.\\
-(7) Tufo, H. M., Fischer, P. F. (2001), \textit{Fast Parallel Direct Solvers For Coarse
+(10) Tufo, H. M., Fischer, P. F. (2001), \textit{Fast Parallel Direct Solvers For Coarse
Grid Problems}, 61, 151-177, J. Par. and Dist. Comput.\\
-(8) Bernardi, C., Dauge, M., Maday, Y. (1999), \textit{Spectral Methods for Axisymmetric
+(11) Bernardi, C., Dauge, M., Maday, Y. (1999), \textit{Spectral Methods for Axisymmetric
Domains}, Vol. 3, Series in Appl. Math., Gauthier-Villars, Paris.\\
-(9) Chaljub, E. (2000), \textit{Mod{\'{e}}lisation num{\'{e}}rique de la
+(12) Chaljub, E. (2000), \textit{Mod{\'{e}}lisation num{\'{e}}rique de la
propagation d'ondes sismiques en g{\'{e}}om{\'{e}}trie sph{\'{e}}rique:
Application {\`{a}} la sismologie globale},
Ph.D. thesis, Universit{\'{e}} de Paris 7.\\
-(10) Komatitsch D., Tromp, J. (2002), \textit{Spectral-element simulations of
+(13) Komatitsch D., Tromp, J. (2002), \textit{Spectral-element simulations of
global seismic wave propagation---I. Validation},
149, 390-412, Geophys. J. Int.
diff --git a/MANUAL/memory_usage_new.pdf b/MANUAL/memory_usage_new.pdf
index 0a12f59..b7c7216 100644
Binary files a/MANUAL/memory_usage_new.pdf and b/MANUAL/memory_usage_new.pdf differ
More information about the CIG-COMMITS
mailing list