Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 2.06 KB

LaunchingCodeInstructions.md

File metadata and controls

6 lines (4 loc) · 2.06 KB

5.3.3. Launching Code Instructions

For the hpfs workflow all code is launched either by a set of batch files or by matlab commands. All of the launching batch files are located in the launch folder of astropath\hpfs. To get started, download the repository to a working destination. The following modules should be launched through the command line and require simple path input: AstroIDGen, TransferDeamon, transferanno, and inform_queue. The rest of the modules can be launched in matlab by adding the AstroPathPipeline repository to the matlab path and entering the module name. Most of the matlab modules require minimal input such as the location of the <Mpath> or a working folder\ drive, all of which is desribed in the respective module documentation.

Each module, outside the meanimages, transferanno and segmaps modules, are designed to indefinitely loop through all Projects defined in the <Mpath>. Files are only updated or processed in the event of new\updated files or queue input. A formal queued workflow is still being developed. The current workflow design requires the use of workers on multiple machines for best computation potential which increases the complications in designing a formal workflow. Modules can be turned off at convience by closing the corresponding command window. However, it should be noted that if there is nothing for the module to process computational impact will be minimal. Usually, modules process all available tasks then hibernate for a set peroid of time before checking for new tasks. The workflow steps are laid out below, this should not be used in lieu of the formal documentation in each section, linked to under contents.

Also note that the modules are expected to be running simultaneously such that the user runs each module from a separate command line or matlab instance. The modules designated as workers can be launched multiple times to reduce processing time though processing locations must be predefined usually hard coded in the respective queue module.