/[MITgcm]/manual/s_autodiff/text/doc_ad_2.tex

Diff of /manual/s_autodiff/text/doc_ad_2.tex

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

-revision 1.20 by edhill,
Wed Apr  5 02:27:33 2006 UTC
+revision 1.21 by heimbach,
Thu Jan 17 22:32:06 2008 UTC
 Line 10 
 existing fully non-linear prognostic cod
  software tool is used that parses and transforms source files
  according to a set of linguistic and mathematical rules.  AD tools are
  like source-to-source translators in that they parse a program code as
- input and produce a new program code as output.  However, unlike a
+ input and produce a new program code as output
+ (we restrict our discussion to source-to-source tools, ignoring
+ operator-overloading tools).  However, unlike a
  pure source-to-source translation, the output program represents a new
  algorithm, such as the evaluation of the Jacobian, the Hessian, or
  higher derivative operators.  In principle, a variety of derived
-Line 21 
 Model Compiler (TAMC) and its successor
+Line 23 
 Model Compiler (TAMC) and its successor
  Algorithms in Fortran), developed by Ralf Giering (\cite{gie-kam:98},
  \cite{gie:99,gie:00}).  The first application of the adjoint of MITgcm
  for sensitivity studies has been published by \cite{maro-eta:99}.
- \cite{sta-eta:97,sta-eta:01} use MITgcm and its adjoint for ocean
+ \cite{stam-etal:97,stam-etal:02} use MITgcm and its adjoint for ocean
  state estimation studies.  In the following we shall refer to TAMC and
  TAF synonymously, except were explicitly stated otherwise.
- TAMC exploits the chain rule for computing the first derivative of a
+ As of mid-2007 we are also able to generate fairly efficient
+ adjoint code of the MITgcm using a new, open-source AD tool,
+ called OpenAD (see \cite{naum-etal:06,utke-etal:08}.
+ This enables us for the first time to compare adjoint models
+ generated from different AD tools, providing an additional
+ accuracy check, complementary to finite-difference gradient checks.
+ OpenAD and its application to  MITgcm is described in detail
+ in section \ref{sec_ad_openad}.
+ The AD tool exploits the chain rule for computing the first derivative of a
  function with respect to a set of input variables.  Treating a given
  forward code as a composition of operations -- each line representing
  a compositional element, the chain rule is rigorously applied to the
-Line 681 
 Schematic view of intermediate dump and
+Line 692 
 Schematic view of intermediate dump and
  In this section we describe in a general fashion
  the parts of the code that are relevant for automatic
  differentiation using the software tool TAF.
+ Modifications to use OpenAD are described in \ref{sec_ad_openad}.
  \input{part5/doc_ad_the_model}
-Line 764 
 w.r.t. the 3-level checkpointing (see se
+Line 776 
 w.r.t. the 3-level checkpointing (see se
  %------------------------------------------------------------------
- \subsection{Building the AD code
+ \subsection{Building the AD code using TAF
  \label{section_ad_build}}
  The build process of an AD code is very similar to building
-Line 774 
 the following {\tt make} targets are ava
+Line 786 
 the following {\tt make} targets are ava
  \begin{table}[h!]
  {\footnotesize
- \begin{tabular}{ccll}
+ \begin{tabular}{|ccll|}
+ \hline
  ~ & {\it AD-target} & {\it output} & {\it description} \\
  \hline
  \hline
-Line 793 
 generates code for $<$MODE$>$ using $<$T
+Line 806 
 generates code for $<$MODE$>$ using $<$T
  ~ & ~ & ~ & and compiles all code \\
  ~ & ~ & ~ & (use of TAF is set as default) \\
  \hline
- \hline
  \end{tabular}
  }
  \end{table}
-Line 802 
 Here, the following placeholders are use
+Line 814 
 Here, the following placeholders are use
  %
  \begin{itemize}
  %
- \item [$<$TOOL$>$]
+ \item $<$TOOL$>$
  %
  \begin{itemize}
  %
-Line 811 
 Here, the following placeholders are use
+Line 823 
 Here, the following placeholders are use
  %
  \end{itemize}
  %
- \item [$<$MODE$>$]
+ \item $<$MODE$>$
  %
  \begin{itemize}
  %
-Line 853 
 The {\tt make <MODE>all} target consists
+Line 865 
 The {\tt make <MODE>all} target consists
  \item
  A header file {\tt AD\_CONFIG.h} is generated which contains a CPP option
  on which code ought to be generated. Depending on the {\tt make} target,
- the contents is
+ the contents is one of the following:
  \begin{itemize}
  \item
  {\tt \#define ALLOW\_ADJOINT\_RUN}
-Line 869 
 consisting of all {\tt .f} files that ar
+Line 881 
 consisting of all {\tt .f} files that ar
  and all {\tt .flow} files that are part of the list {\bf AD\_FLOW\_FILES}.
  %
  \item
- The AD tool is invoked with the {\bf <MODE>\_<TOOL>\_FLAGS}.
+ The AD tool is invoked with the {\tt <MODE>\_<TOOL>\_FLAGS}.
  The default AD tool flags in {\tt genmake2} can be overrwritten by
  an {\tt adjoint\_options} file (similar to the platform-specific
  {\tt build\_options}, see Section ???.
-Line 1117 
 from each contribution and sums over all
+Line 1129 
 from each contribution and sums over all
  \end{equation}
  %
  The total cost function {\bf fc} will be the
- 'dependent' variable in the argument list for TAMC, i.e.
+ 'dependent' variable in the argument list for TAF, i.e.
  \begin{verbatim}
- tamc -output 'fc' ...
+ taf -output 'fc' ...
  \end{verbatim}
  %%%% \end{document}
-Line 1234 
 and gradient are generated and initialis
+Line 1246 
 and gradient are generated and initialis
  %
  The dependency flow for differentiation w.r.t. the controls
  starts with adding a perturbation onto the input variable,
- thus defining the independent or control variables for TAMC.
+ thus defining the independent or control variables for TAF.
  Three types of controls may be considered:
  %
  \begin{itemize}
-Line 1267 
 u         & = \, u_{[0]} \, + \, \Delta
+Line 1279 
 u         & = \, u_{[0]} \, + \, \Delta
  holding the perturbation. In the case of a simple
  sensitivity study this array is identical to zero.
  However, it's specification is essential in the context
- of automatic differentiation since TAMC
+ of automatic differentiation since TAF
  treats the corresponding line in the code symbolically
  when determining the differentiation chain and its origin.
  Thus, the variable names are part of the argument list
- when calling TAMC:
+ when calling TAF:
  %
  \begin{verbatim}
- tamc -input 'xx_tr1 ...' ...
+ taf -input 'xx_tr1 ...' ...
  \end{verbatim}
  %
  Now, as mentioned above, MITgcm avoids maintaining
  an array for each control variable by reading the
  perturbation to a temporary array from file.
- To ensure the symbolic link to be recognized by TAMC, a scalar
+ To ensure the symbolic link to be recognized by TAF, a scalar
  dummy variable {\bf xx\_tr1\_dummy} is introduced
  and an 'active read' routine of the adjoint support
  package {\it pkg/autodiff} is invoked.
  The read-procedure is tagged with the variable
- {\bf xx\_tr1\_dummy} enabling TAMC to recognize the
+ {\bf xx\_tr1\_dummy} enabling TAF to recognize the
  initialization of the perturbation.
- The modified call of TAMC thus reads
+ The modified call of TAF thus reads
  %
  \begin{verbatim}
- tamc -input 'xx_tr1_dummy ...' ...
+ taf -input 'xx_tr1_dummy ...' ...
  \end{verbatim}
  %
  and the modified operation to (\ref{perturb})

 Legend:



Removed from v.1.20
 


changed lines


 
Added in v.1.21
 Legend:



Removed from v.1.20
 


changed lines


 
Added in v.1.21
-Removed from v.1.20
+Added in v.1.21

	ViewVC Help
Powered by ViewVC 1.1.22