Use of pipelines
- Cormac Reynolds
- Ultimate goal
- fully calibrated images (with appropriate measures)
- Practical Limitations
- 'intelligent' software expensive to write
- initially must settle for incremental improvements
- should at least automate 'book-keeping' and 'standard steps'
- more complicated issues may be omitted or be human-tunable
Example - EVN pipeline
- Objectives:
- To reduce the total time required to analyse an EVN data set
- Minimal user interaction
- Simplified book-keeping
- Automated correction for known, well-defined errors/effects (A priori amplitude calibration and flagging, parallactic angle, telescope positions, etc.)
- 'Rough' calibration of the data through to imaging - principally to allow assessment of data and calibration quality
- Produces diagnostic plots
- Produces AIPS calibration tables
- Processing:
- Pipeline run by correlator staff (but publicly available for use elsewhere)
- Results distributed via the EVN data archive
- Weaknesses:
- Data editing (very difficult to automate with current software)
- Runs on stable version of AIPS (always out of date...)
- No polarization calibration
- No support for spectral line experiments
- Does not attempt to produce publishable images
Example - MERLIN user procedure
- Objectives:
- User-steered data reduction of single target+calibration sources
- Look and feel of AIPS (familiar to many users)
- Defaults should work for simple continuum data, but user can tweak/use interactive options
- Weaknesses:
- Data editing not automated
- Runs on current version of AIPS (can be broken by changes in AIPS code)
- Automatic imaging of sources with structure can be problematic
- Polarization calibration can be poor
- Limited support for spectral line observations
Example - PdBI pipeline
- TBD
- Archives
- An introduction to what is available in the EVN and MERLIN archives, and how to get it.
- User development
Write your own pipeline: ParselTongue (Python interface to AIPS) demonstration.
