This page was created prior to a phone meeting scheduled for 10AM on 11 Jan 2011. Eventually this list will be associated with tracked items on Pivotal Tracker
(and possibly also MRs where necessary). None of this has been agreed to yet.
Items needed to complete the KFPA functionality:
- Frequency-switched mode
- Developer documentation
- Online filler improvements:
- Clean up the documentation - especially the various partially completed and mostly out of date wiki pages made during the development stage.
Observer's Pipeline Priorities GIL:
In many ways the KFPA and Pipeline are functioning very well and producing good results. Observers are producing images faster
than ever before, and we're hearing very good reports.
We're still in the middle of learning about very large data bases and how to deal with simultaneous processing very big ones.
Some of these issues lead to very dramatic changes in system performance. Ie we just turned on multi-processor, parallel computing
to discover that when too many processes were spawned, many calibration tasks died, giving the appearance of a very serious problem.
Fortunately we were able to diagnose the problem and Joe added options to limit the number of processes, fixing this problem.
There are many tasks we'd like to complete, the top ones are:
- Real-time filling with the capability of mapping on separate computers and without adversely effecting data monitoring.
- We've discussed changing filling so that individual scans are kept separate, so that reading older data does not effect current data writing.
- We'd need to change the pipeline to read groups of files, instead of a single file
- We'd need a tool to merge all requested scans for input to the realtime GBTIDL.
- Early in an observing session GBTIDL is a useful tool for examining the data, but after an hour or so, "online" GBTIDL is too slugish. The observer needs to "sdfits" individual scans separately, then bring them in to GBTIDL and examine them.
- Completing the use of planned calibration inputs
- PTCS efficiency and gain parameters
- Vector T-cals and/or Vector S-cals
- Frequency Switched Calibration
- Clean up the system of measuring and applying relative gain values to the observations.
- Get the dedicated machine fully functioning. We're working on some issues with new hardware, and we may need to exchange some components.
- Revising imaging to work with gridded images and weights images, so that the processing can be made more parallel and basket-weaving baseline can be implemented. Basket weaving calibration does not seem an absolute necessity yet, but would produce the best images. Needs some research.
- Adding some new tools to simplify combing different observing blocks. The pipeline is very good at imaging smaller images (about an hours data), but the observer has to do bookkeeping to keep track of partial maps and optimizing the placement of the image center and bounding region.
- Publish a KFPA hardware Paper
- Publish a KFPA Pipeline Paper and some first results.
Observing Preparation, Monitoring
- Set software warning/monitor levels (lo, voltages, cryogenic temperatures) to the appropriate values.
- Test pointing with more beam than 2.
- Run tests with IF Rack pointing and 4 beam observations (B. Mason's suggestion).