JSim Batch Processing Manual
JSim users normally use the Graphic User Interface(GUI) for routine modeling work. However, for debugging and batch processing, it is sometimes useful to run JSim models via the command line. The program jsbatch provides this functionality and is described in this document.
Note: jsbatch switches were significantly reworked for JSim release 1.6.71 to provide more consistency and transparency. This document applies to JSim releases 1.6.71 and above. See here for information on older releases. This document is under construction (Mar 13 2007).
Prerequisites:
- Introduction to the JSim GUI (required)
- Introductory Survey of MML (required)
- JSim Projects (recommended)
- Data Files and Project Data Sets (recommended)
- Familiarity with command-line tools in Unix user environment (recommended).
Contents:
Overview
jsbatch is a command-line driven, text-based program that loads JSim projects, builds and runs models and writes results in various formats. This approach to modeling can be useful for batch processing, or for debugging models or JSim internals. jsbatch provides no graphics or data-analysis facilities. Users requiring those abilities should use the JSim GUI. This document describes running jsbatch on Unix systems. Microsoft Windows users should make appropriate translations for that operating system.
The general sequence of jsbatch operation follows. This general sequence may be altered somewhat by switch selection.
- A session work directory ($HOME/.jsim/work*) is created for temporary file creation.
- If a project file (extension .proj) is specified via the -f switch, the project file is loaded. Otherwise, a new, empty project is created.
- Any additional files specified via the -f, -func or -plot switches are loaded into the project.
- The first model within the project is selected (unless overridden by the -model switch). The selected model is built.
- Any model inputs and other project variables values specified via the -i switch, are assigned.
- The selected model is run.
- Values for all model variables are printed on the standard output;
- Session work directory is deleted.
- Program exits with status 0 (normal).
Command line synopsis
The following switches modify standard program behavior before the models are run.
Usage: jsbatch [ switches ] Switches: -f file ... load project, model & other files -v verbose mode -userurl url load -f files relative to URL -userdir dir set user directory -path path set search path -server name connect to remote server -plugin file ... load optional plugin -usage print usage message and exit -stack print stack dump on error -noscrub don't scrub work directory at exit -sandbox [path [path]] run application is safe sandbox -nosandbox don't use sandbox (default) -securityMgr class-name install security manager -mp N enable up to N processor multiprocessing -i var=expr ... assign 1 or more model inputs or controls -func name ... create 1 or more function generator -plot name ... create 1 or more plot pages -rtml file ... load model RTML from file -img file ... load 1 or more model image files -lmod name load source code into model -lpar name load project par set into model -lxpar file load XSim .par file into model -spar name store model par set in project -filter dataset ... filter 1 or more project datasets -iproj proj:name ... import content from other projects -c name=value ... set model compiler options -model name specify which model (if multiple) -q [var ...] query variable constraints -loops loops run instead of normal run -sens sensitivity analysis instead of normal run -optim run optimizer first -nruns # # runs for each model -fwdIC forward ICs after (each) run -o expr ... output selected model expressions -oplot output data from plots -ofgenprev output data from function generator preview -odataset name output data from specified dataset -ortml name output selected model's RTML -out file write output to file instead of stdout -odim # select data of specified dimension -ofmt name write output in named format -oblocked mn mx write blocked output (if ofmt supported) -osingle write single-precision output (default) -odouble write double-precision output -oprec # write custom precision output -ozero # write small output data as 0 -oencoding code write format-specific output encoding -ocrash write output even if run crashes -oreport prepend optimization report -oplan [variant] write model plan(s) to output (if available) -omathml write model MathML to output -omml write MML model(s) to output (if available) -oflat write flat model(s) to output (if available) -ojava write java model(s) to output (if available) -oproj [name ...] write project content to output -sdata name store output in project dataset -
