/[MITgcm]/manual/s_getstarted/text/getting_started.tex

Diff of /manual/s_getstarted/text/getting_started.tex

Parent Directory | Revision Log | View Revision Graph Revision Graph | View Patch Patch

-revision 1.3 by adcroft,
Thu Oct 11 19:18:41 2001 UTC
+revision 1.6 by adcroft,
Thu Oct 18 18:44:14 2001 UTC
 Line 1
  % $Header$
  % $Name$
- \section{Getting started}
+ %\section{Getting started}
- In this part, we describe how to use the model. In the first section, we
+ In this section, we describe how to use the model. In the first
- provide enough information to help you get started with the model. We
+ section, we provide enough information to help you get started with
- believe the best way to familiarize yourself with the model is to run the
+ the model. We believe the best way to familiarize yourself with the
- case study examples provided with the base version. Information on how to
+ model is to run the case study examples provided with the base
- obtain, compile, and run the code is found there as well as a brief
+ version. Information on how to obtain, compile, and run the code is
- description of the model structure directory and the case study examples.
+ found there as well as a brief description of the model structure
- The latter and the code structure are described more fully in sections 2 and
+ directory and the case study examples.  The latter and the code
-, respectively. In section 4, we provide information on how to customize
+ structure are described more fully in chapters
- the code when you are ready to try implementing the configuration you have
+ \ref{chap:discretization} and \ref{chap:sarch}, respectively. Here, in
- in mind.
+ this section, we provide information on how to customize the code when
+ you are ready to try implementing the configuration you have in mind.
- \subsection{Obtaining the code}
+ \section{Where to find information}
- The reference web site for the model is:
+ \label{sect:whereToFindInfo}
- \begin{verbatim}
- http://mitgcm.org
+ A web site is maintained for release 1 (Sealion) of MITgcm:
- \end{verbatim}
+ \begin{verbatim}
+ http://mitgcm.org/sealion
- On this site, you can download the model as well as find useful information,
+ \end{verbatim}
- some of which might overlap with what is written here. There is also a
+ Here you will find an on-line version of this document, a
- support news group for the model located at (send your message to \texttt{%
+ ``browsable'' copy of the code and a searchable database of the model
- support@mitgcm.org}):
+ and site, as well as links for downloading the model and
+ documentation, to data-sources and other related sites.
+ There is also a support news group for the model that you can email at
+ \texttt{support@mitgcm.org} or browse at:
  \begin{verbatim}
  news://mitgcm.org/mitgcm.support
  \end{verbatim}
+ A mail to the email list will reach all the developers and be archived
+ on the newsgroup. A users email list will be established at some time
+ in the future.
+ \section{Obtaining the code}
+ \label{sect:obtainingCode}
  If CVS is available on your system, we strongly encourage you to use it. CVS
  provides an efficient and elegant way of organizing your code and keeping
  track of your changes. If CVS is not available on your machine, you can also
  download a tar file.
- \subsubsection{Using CVS}
  Before you can use CVS, the following environment variable has to be set in
  your .cshrc or .tcshrc:
  \begin{verbatim}
  % setenv CVSROOT :pserver:cvsanon@mitgcm.org:/u/u0/gcmpack
+ \end{verbatim}
+ To start using CVS, "login" to the server using:
+ \begin{verbatim}
  % cvs login ( CVS password: cvsanon )
  \end{verbatim}
+ You only need to do ``cvs login'' once.
- You only need to do ``cvs login'' once. To obtain the source for the release:
+ To obtain the source for the release:
  \begin{verbatim}
  % cvs co -d directory -P -r release1 MITgcmUV
  \end{verbatim}
-Line 62 
 or the latest development version:
+Line 76 
 or the latest development version:
  % cvs co -d directory -P MITgcm
  \end{verbatim}
- \subsubsection{other methods}
+ \paragraph*{Conventional download method}
+ \label{sect:conventionalDownload}
- You can download the model as a tar file from the reference web site at:
+ If you do not have CVS on your system, you can download the model as a
+ tar file from the reference web site at:
  \begin{verbatim}
  http://mitgcm.org/download/
  \end{verbatim}
+ The tar file still contains CVS information which we urge you not to
+ delete; even if you do not use CVS yourself the information can help
+ us if you should need to send us your copy of the code.
- \subsection{Model and directory structure}
+ \section{Model and directory structure}
  The ``numerical'' model is contained within a execution environment support
  wrapper. This wrapper is designed to provide a general framework for
-Line 119 
 the makefile specific to the Tangent lin
+Line 138 
 the makefile specific to the Tangent lin
  generates the adjoint code. The latter is described in details in part V.
  \item \textit{utils}: this directory contains various utilities. The
- subdirectory \textit{knudsen2} contains code and a makefile that compute
+ subdirectory \textit{knudsen2} contains code and a makefile that
- coefficients of the polynomial approximation to the knudsen formula for an
+ compute coefficients of the polynomial approximation to the knudsen
- ocean nonlinear equation of state. The \textit{matlab} subdirectory contains
+ formula for an ocean nonlinear equation of state. The \textit{matlab}
- matlab scripts for reading model output directly into matlab. \textit{scripts%
+ subdirectory contains matlab scripts for reading model output directly
- } contains C-shell post-processing scripts for joining processor-based and
+ into matlab. \textit{scripts} contains C-shell post-processing
- tiled-based model output.
+ scripts for joining processor-based and tiled-based model output.
  \item \textit{verification}: this directory contains the model examples. See
- below.
+ section \ref{sect:modelExamples}.
  \end{itemize}
- \subsection{Model examples}
+ \section{Example experiments}
+ \label{sect:modelExamples}
  Now that you have successfully downloaded the model code we recommend that
  you first try to run the examples provided with the base version. You will
-Line 139 
 you will use eventually. The examples ar
+Line 159 
 you will use eventually. The examples ar
  the directory \textit{verification} and are briefly described below (a full
  description is given in section 2):
- \subsubsection{List of model examples}
+ \subsection{List of model examples}
  \begin{itemize}
  \item \textit{exp0} - single layer, ocean double gyre (barotropic with
-Line 193 
 Molteni physics package. Global configur
+Line 213 
 Molteni physics package. Global configur
  \item \textit{adjustment.cs-32x32x1}
  \end{itemize}
- \subsubsection{Directory structure of model examples}
+ \subsection{Directory structure of model examples}
  Each example directory has the following subdirectories:
-Line 245 
 comparison with your own output when you
+Line 265 
 comparison with your own output when you
  Once you have chosen the example you want to run, you are ready to compile
  the code.
- \subsection{Compiling the code}
+ \section{Building the code}
+ \label{sect:buildingCode}
+ To compile the code, we use the {\em make} program. This uses a file
+ ({\em Makefile}) that allows us to pre-process source files, specify
+ compiler and optimization options and also figures out any file
+ dependancies. We supply a script ({\em genmake}), described in section
+ \ref{sect:genmake}, that automatically creates the {\em Makefile} for
+ you. You then need to build the dependancies and compile the code.
+ As an example, let's assume that you want to build and run experiment
+ \textit{verification/exp2}. The are multiple ways and places to actually
+ do this but here let's build the code in
+ \textit{verification/exp2/input}:
+ \begin{verbatim}
+ % cd verification/exp2/input
+ \end{verbatim}
+ First, build the {\em Makefile}:
+ \begin{verbatim}
+ % ../../../tools/genmake -mods=../code
+ \end{verbatim}
+ The command line option tells {\em genmake} to override model source
+ code with any files in the directory {\em ./code/}.
+ If there is no \textit{.genmakerc} in the \textit{input} directory, you have
+ to use the following options when invoking \textit{genmake}:
+ \begin{verbatim}
+ % ../../../tools/genmake  -mods=../code
+ \end{verbatim}
+ Next, create the dependancies:
+ \begin{verbatim}
+ % make depend
+ \end{verbatim}
+ This modifies {\em Makefile} by attaching a [long] list of files on
+ which other files depend. The purpose of this is to reduce
+ re-compilation if and when you start to modify the code. {\tt make
+ depend} also created links from the model source to this directory.
+ Now compile the code:
+ \begin{verbatim}
+ % make
+ \end{verbatim}
+ The {\tt make} command creates an executable called \textit{mitgcmuv}.
+ Now you are ready to run the model. General instructions for doing so are
+ given in section \ref{sect:runModel}. Here, we can run the model with:
+ \begin{verbatim}
+ ./mitgcmuv > output.txt
+ \end{verbatim}
+ where we are re-directing the stream of text output to the file {\em
+ output.txt}.
+ \subsection{Building/compiling the code elsewhere}
+ In the example above (section \ref{sect:buildingCode}) we built the
+ executable in the {\em input} directory of the experiment for
+ convenience. You can also configure and compile the code in other
+ locations, for example on a scratch disk with out having to copy the
+ entire source tree. The only requirement to do so is you have {\tt
+ genmake} in your path or you know the absolute path to {\tt genmake}.
+ The following sections outline some possible methods of organizing you
+ source and data.
+ \subsubsection{Building from the {\em ../code directory}}
+ This is just as simple as building in the {\em input/} directory:
+ \begin{verbatim}
+ % cd verification/exp2/code
+ % ../../../tools/genmake
+ % make depend
+ % make
+ \end{verbatim}
+ However, to run the model the executable ({\em mitgcmuv}) and input
+ files must be in the same place. If you only have one calculation to make:
+ \begin{verbatim}
+ % cd ../input
+ % cp ../code/mitgcmuv ./
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ or if you will be making muliple runs with the same executable:
+ \begin{verbatim}
+ % cd ../
+ % cp -r input run1
+ % cp code/mitgcmuv run1
+ % cd run1
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ \subsubsection{Building from a new directory}
- \subsubsection{The script \textit{genmake}}
+ Since the {\em input} directory contains input files it is often more
+ useful to keep {\em input} prestine and build in a new directory
+ within {\em verification/exp2/}:
+ \begin{verbatim}
+ % cd verification/exp2
+ % mkdir build
+ % cd build
+ % ../../../tools/genmake -mods=../code
+ % make depend
+ % make
+ \end{verbatim}
+ This builds the code exactly as before but this time you need to copy
+ either the executable or the input files or both in order to run the
+ model. For example,
+ \begin{verbatim}
+ % cp ../input/* ./
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ or if you tend to make multiple runs with the same executable then
+ running in a new directory each time might be more appropriate:
+ \begin{verbatim}
+ % cd ../
+ % mkdir run1
+ % cp build/mitgcmuv run1/
+ % cp input/* run1/
+ % cd run1
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ \subsubsection{Building from on a scratch disk}
+ Model object files and output data can use up large amounts of disk
+ space so it is often the case that you will be operating on a large
+ scratch disk. Assuming the model source is in {\em ~/MITgcm} then the
+ following commands will build the model in {\em /scratch/exp2-run1}:
+ \begin{verbatim}
+ % cd /scratch/exp2-run1
+ % ~/MITgcm/tools/genmake -rootdir=~/MITgcm -mods=~/MITgcm/verification/exp2/code
+ % make depend
+ % make
+ \end{verbatim}
+ To run the model here, you'll need the input files:
+ \begin{verbatim}
+ % cp ~/MITgcm/verification/exp2/input/* ./
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ As before, you could build in one directory and make multiple runs of
+ the one experiment:
+ \begin{verbatim}
+ % cd /scratch/exp2
+ % mkdir build
+ % cd build
+ % ~/MITgcm/tools/genmake -rootdir=~/MITgcm -mods=~/MITgcm/verification/exp2/code
+ % make depend
+ % make
+ % cd ../
+ % cp -r ~/MITgcm/verification/exp2/input run2
+ % cd run2
+ % ./mitgcmuv > output.txt
+ \end{verbatim}
+ \subsection{\textit{genmake}}
+ \label{sect:genmake}
  To compile the code, use the script \textit{genmake} located in the \textit{%
  tools} directory. \textit{genmake} is a script that generates the makefile.
-Line 348 
 For some of the examples, there is a fil
+Line 524 
 For some of the examples, there is a fil
  that particular example. In this way you don't need to type the options when
  invoking \textit{genmake}.
- \subsubsection{Compiling}
- Let's assume that you want to run, say, example \textit{exp2} in the \textit{%
+ \section{Running the model}
- input} directory. To compile the code, type the following commands from the
+ \label{sect:runModel}
- model root tree:
- \begin{verbatim}
- % cd verification/exp2/input
- % ../../../tools/genmake
- % make depend
- % make
- \end{verbatim}
- If there is no \textit{.genmakerc} in the \textit{input} directory, you have
+ If compilation finished succesfuully (section \ref{sect:buildModel})
- to use the following options when invoking \textit{genmake}:
+ then an executable called {\em mitgcmuv} will now exist in the local
- \begin{verbatim}
+ directory.
- % ../../../tools/genmake  -mods=../code
- \end{verbatim}
- In addition, you will probably want to disable some of the packages. Taking
+ To run the model as a single process (ie. not in parallel) simply
- again the case of \textit{exp2}, the full \textit{genmake} command will
+ type:
- probably look like this:
  \begin{verbatim}
- % ../../../tools/genmake  -mods=../code  -disable=kpp,gmredi,aim,...
+ % ./mitgcmuv
  \end{verbatim}
+ The ``./'' is a safe-guard to make sure you use the local executable
- The make command creates an executable called \textit{mitgcmuv}.
+ in case you have others that exist in your path (surely odd if you
+ do!). The above command will spew out many lines of text output to
- Note that you can compile and run the code in another directory than \textit{%
+ your screen.  This output contains details such as parameter values as
- input}. You just need to make sure that you copy the input data files into
+ well as diagnostics such as mean Kinetic energy, largest CFL number,
- the directory where you want to run the model. For example to compile from
+ etc. It is worth keeping this text output with the binary output so we
- \textit{code}:
+ normally re-direct the {\em stdout} stream as follows:
  \begin{verbatim}
- % cd verification/exp2/code
+ % ./mitgcmuv > output.txt
- % ../../../tools/genmake
- % make depend
- % make
  \end{verbatim}
- \subsection{Running the model}
+ For the example experiments in {\em vericication}, an example of the
+ output is kept in {\em results/output.txt} for comparison. You can compare
+ your {\em output.txt} with this one to check that the set-up works.
- The first thing to do is to run the code by typing \textit{mitgcmuv} and see
- what happens. You can compare what you get with what is in the \textit{%
- results} directory. Unless noted otherwise, most examples are set up to run
- for a few time steps only so that you can quickly figure out whether the
- model is working or not.
- \subsubsection{Output files}
+ \subsection{Output files}
  The model produces various output files. At a minimum, the instantaneous
  ``state'' of the model is written out, which is made of the following files:
-Line 445 
 as the pickup files but are named differ
+Line 605 
 as the pickup files but are named differ
  used to restart the model but are overwritten every other time they are
  output to save disk space during long integrations.
- \subsubsection{Looking at the output}
+ \subsection{Looking at the output}
  All the model data are written according to a ``meta/data'' file format.
  Each variable is associated with two files with suffix names \textit{.data}
-Line 459 
 written in this format. The matlab scrip
+Line 619 
 written in this format. The matlab scrip
  \textit{utils/matlab} under the root tree. The script \textit{rdmds.m} reads
  the data. Look at the comments inside the script to see how to use it.
- \section{Code structure}
+ Some examples of reading and visualizing some output in {\em Matlab}:
+ \begin{verbatim}
+ % matlab
+ >> H=rdmds('Depth');
+ >> contourf(H');colorbar;
+ >> title('Depth of fluid as used by model');
+ >> eta=rdmds('Eta',10);
+ >> imagesc(eta');axis ij;colorbar;
+ >> title('Surface height at iter=10');
- \section{Doing it yourself: customizing the code}
+ >> eta=rdmds('Eta',[0:10:100]);
+ >> for n=1:11; imagesc(eta(:,:,n)');axis ij;colorbar;pause(.5);end
+ \end{verbatim}
- \subsection{\protect\bigskip Configuration and setup}
+ \section{Doing it yourself: customizing the code}
  When you are ready to run the model in the configuration you want, the
  easiest thing is to use and adapt the setup of the case studies experiment
-Line 473 
 relative to the ''numerical model'' part
+Line 644 
 relative to the ''numerical model'' part
  the ''execution environment'' part is covered in the parallel implementation
  section) and on the variables and parameters that you are likely to change.
+ \subsection{Configuration and setup}
  The CPP keys relative to the ''numerical model'' part of the code are all
  defined and set in the file \textit{CPP\_OPTIONS.h }in the directory \textit{%
  model/inc }or in one of the \textit{code }directories of the case study
-Line 489 
 In what follows the parameters are group
+Line 662 
 In what follows the parameters are group
  computational domain, the equations solved in the model, and the simulation
  controls.
- \subsubsection{Computational domain, geometry and time-discretization}
+ \subsection{Computational domain, geometry and time-discretization}
  \begin{itemize}
  \item dimensions
-Line 581 
 parameter is set through the variable \t
+Line 754 
 parameter is set through the variable \t
  stagger baroclinic time stepping can be activated by setting the logical
  variable \textbf{staggerTimeStep }to '.\texttt{TRUE}.'.
- \subsubsection{Equation of state}
+ \subsection{Equation of state}
  First, because the model equations are written in terms of perturbations, a
  reference thermodynamic state needs to be specified. This is done through
-Line 610 
 available in the same directory and you
+Line 783 
 available in the same directory and you
  values of the vertical levels in \textit{knudsen2.f }so that they match
  those of your configuration). \textit{\ }
- \subsubsection{Momentum equations}
+ \subsection{Momentum equations}
  In this section, we only focus for now on the parameters that you are likely
  to change, i.e. the ones relative to forcing and dissipation for example.
-Line 714 
 geopotential (for the atmosphere) you ne
+Line 887 
 geopotential (for the atmosphere) you ne
  \texttt{TRUE}.' and the other to '.\texttt{FALSE}.' depending on how you
  want to deal with the ocean upper or atmosphere lower boundary).
- \subsubsection{Tracer equations}
+ \subsection{Tracer equations}
  This section covers the tracer equations i.e. the potential temperature
  equation and the salinity (for the ocean) or specific humidity (for the
-Line 805 
 wish the tracer vertical diffusivities t
+Line 978 
 wish the tracer vertical diffusivities t
  vertically due to static instabilities. Note that \textbf{cadjFreq }and
  \textbf{ivdc\_kappa }can not both have non-zero value.
- \subsubsection{Simulation controls}
+ \subsection{Simulation controls}
  The model ''clock'' is defined by the variable \textbf{deltaTClock }(in s)
  which determines the IO frequencies and is used in tagging output.

 Legend:



Removed from v.1.3
 


changed lines


 
Added in v.1.6
 Legend:



Removed from v.1.3
 


changed lines


 
Added in v.1.6
-Removed from v.1.3
+Added in v.1.6

	ViewVC Help
Powered by ViewVC 1.1.22