[cig-commits] r21519 - seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL

[dkomati1 at geodynamics.org]

Author: dkomati1
Date: 2013-03-13 10:03:09 -0700 (Wed, 13 Mar 2013)
New Revision: 21519

Modified:
   seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.pdf
   seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.tex
Log:
added more details about SIMULATION_TYPE == 2


Modified: seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.pdf
===================================================================
(Binary files differ)

Modified: seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.tex
===================================================================
--- seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.tex	2013-03-13 00:23:42 UTC (rev 21518)
+++ seismo/3D/SPECFEM3D_GLOBE/trunk/doc/USER_MANUAL/manual_SPECFEM3D_GLOBE.tex	2013-03-13 17:03:09 UTC (rev 21519)
@@ -197,24 +197,24 @@
 It was a finalist again in 2008 for a run at 0.16 petaflops (sustained) on 149,784 processors of the `Jaguar' Cray XT5 system at Oak Ridge National Laboratories (USA) \citep{CaKoLaTiMiLeSnTr08}.
 It also won the BULL Joseph Fourier supercomputing award in 2010.\\
 
-The next release of the code will include support for GPU graphics card acceleration \citep{KoMiEr09,KoGoErMi10,KoErGoMi10,MiKo10,Kom11}
+The next release of the code will include support for GPU graphics card acceleration \citep{KoMiEr09,KoErGoMi10,MiKo10,Kom11}
 as well as Convolutional or Auxiliary Differential Equation Perfectly Matched absorbing Layers (C-PML or ADE-PML)
 \citep{KoMa07,MaKoEz08,MaKoGe08,MaKo09,MaKoGeBr10} for the case of single-chunk simulations in regional models.
 
 \section{Citation}
 
 If you use SPECFEM3D\_GLOBE for your own research, please cite at least one
-of the following articles: \cite{TrKoLi08,PeKoLuMaLeCaLeMaLiBlNiBaTr11,VaCaSaKoVi99,LeChLiKoHuTr08,LeChKoHuTr09,LeKoHuTr09,KoMiEr09,KoErGoMi10,KoGoErMi10,WiKoScTr04,KoLiTrSuStSh04,ChKoViCaVaFe07,MaKoDi09,KoViCh10,CaKoLaTiMiLeSnTr08,TrKoHjLiZhPeBoMcFrTrHu10,KoRiTr02,KoTr02a,KoTr02b,KoTr99} or \cite{KoVi98}.\\
+of the following articles: \cite{TrKoLi08,PeKoLuMaLeCaLeMaLiBlNiBaTr11,VaCaSaKoVi99,LeChLiKoHuTr08,LeChKoHuTr09,LeKoHuTr09,KoMiEr09,KoErGoMi10,WiKoScTr04,KoLiTrSuStSh04,ChKoViCaVaFe07,MaKoDi09,KoViCh10,CaKoLaTiMiLeSnTr08,TrKoHjLiZhPeBoMcFrTrHu10,KoRiTr02,KoTr02a,KoTr02b,KoTr99} or \cite{KoVi98}.\\
 
 If you use this new version 5.1, which has non blocking MPI for much better performance for medium or large runs, please cite at least one of these five articles,
-in which results of 3D non blocking MPI runs are presented: \cite{KoGoErMi10,KoErGoMi10,KoViCh10,Kom11,PeKoLuMaLeCaLeMaLiBlNiBaTr11,CaKoLaTiMiLeSnTr08}.\\
+in which results of 3D non blocking MPI runs are presented: \cite{KoErGoMi10,KoViCh10,Kom11,PeKoLuMaLeCaLeMaLiBlNiBaTr11,CaKoLaTiMiLeSnTr08}.\\
 
 If you work on geophysical applications, you may be interested in citing some of these application articles as well, among others:
 \cite{WiKoScTr04,JiTsKoTr05,KrJiKoTr06a,KrJiKoTr06b,LeChLiKoHuTr08,LeChKoHuTr09,LeKoHuTr09,ChFaKo04,FaChKo04,RiRiKoTrHe02,GoAmTaCaSmSaMaKo09,TrKo00,SaKoTr10}.
 If you use 3D mantle model S20RTS, please cite \citet{RiVaWo99}.\\
 
 Domain decomposition is explained in detail in \cite{MaKoBlLe08}, and excellent scaling up to 150,000 processor cores in shown for instance in
-\cite{CaKoLaTiMiLeSnTr08,KoLaMi08a,MaKoBlLe08,KoErGoMi10,KoGoErMi10,Kom11}.\\
+\cite{CaKoLaTiMiLeSnTr08,KoLaMi08a,MaKoBlLe08,KoErGoMi10,Kom11}.\\
 
 The corresponding Bib\TeX{} entries may be found
 in file \texttt{doc/USER\_MANUAL/bibliography.bib}.
@@ -597,7 +597,7 @@
 
 \begin{description}
 \item [{\texttt{SIMULATION\_TYPE}}] is set to 1 for forward simulations,
-2 for adjoint simulations (see Section \ref{sec:Adjoint-simulation-finite})
+2 for adjoint simulations for sources (see Section \ref{sec:Adjoint-simulation-finite})
 and 3 for kernel simulations (see Section \ref{sec:Finite-Frequency-Kernels}).
 \item [{\texttt{SAVE\_FORWARD}}] is only set to \texttt{.true.} for a forward
 simulation with the last frame of the simulation saved, as part of
@@ -1761,7 +1761,7 @@
 (1, 2 or 3) and \texttt{SAVE\_FORWARD} (boolean).
 
 
-\section{\label{sec:Adjoint-simulation-sources}Adjoint Simulations for Sources}
+\section{\label{sec:Adjoint-simulation-sources}Adjoint Simulations for Sources Only (not for the Model)}
 
 In the case where a specific misfit function is minimized to invert
 for the earthquake source parameters, the gradient of the misfit function
@@ -3572,7 +3572,7 @@
 
 Seismic networks, such as the Global Seismographic Network (GSN), generally involve various types of instruments with different bandwidths, sampling properties and component configurations. There are standards to name channel codes depending on instrument properties. IRIS  \texttt{(www.iris.edu)} uses SEED/FDSN format for channel codes, which are represented by three letters, such as \texttt{LHN}, \texttt{BHZ}, etc. In older versions of the SPECFEM package, a common format was used for the channel codes of all seismograms, which was \texttt{LHE/LHN/LHZ} for three components. To avoid confusion when comparisons are made to observed data, we are now using the FDSN convention for SEM seismograms. In the following, we give a brief explanation of the FDSN convention and how it is adopted in SEM seismograms. Please visit \texttt{www.iris.edu/manuals/SEED\_appA.htm} for further information.\\
 
-\noindent \textbf{\texttt{Band code:}} The first letter in the channel code denotes the band code of seismograms, which depends on the response band and the sampling rate of instruments. The list of band codes used by IRIS is shown in Figure \ref{fig:IRIS_band_codes}. The sampling rate of SEM synthetics is controlled by the resolution of simulations rather than instrument properties. However, for consistency, we follow the FDSN convention for SEM seismograms governed by their sampling rate. For SEM synthetics, we consider band codes for which $dt \leq 1$ s. The FDSN convention also considers the response band of instruments. For instance, short-period and broad-band seismograms with the same sampling rate correspond to different band codes, such as S and B, respectively. In such cases, we consider SEM seismograms as broad band, ignoring the corner period ($\geq 10$ s) of the response band of instruments (note that at these resolutions, the minimum period in the SEM synthetics will be less than $10$ s). 
+\noindent \textbf{\texttt{Band code:}} The first letter in the channel code denotes the band code of seismograms, which depends on the response band and the sampling rate of instruments. The list of band codes used by IRIS is shown in Figure \ref{fig:IRIS_band_codes}. The sampling rate of SEM synthetics is controlled by the resolution of simulations rather than instrument properties. However, for consistency, we follow the FDSN convention for SEM seismograms governed by their sampling rate. For SEM synthetics, we consider band codes for which $dt \leq 1$ s. The FDSN convention also considers the response band of instruments. For instance, short-period and broad-band seismograms with the same sampling rate correspond to different band codes, such as S and B, respectively. In such cases, we consider SEM seismograms as broad band, ignoring the corner period ($\geq 10$ s) of the response band of instruments (note that at these resolutions, the minimum period in the SEM synthetics will be less than $10$ s).
 Accordingly, when you run a simulation the band code will be chosen depending on the resolution of the synthetics, and channel codes of SEM seismograms will start with either \texttt{L}, \texttt{M}, \texttt{B}, \texttt{H}, \texttt{C} or \texttt{F}, shown by red color in the figure.\\
 
 \begin{figure}[ht]