Versions Compared

Old Version 6

changes.mady.by.user Lynne Jones

Saved on

compared with

New Version Current

changes.mady.by.user Lynne Jones

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

 

(updated from standup notes to weekly status/plans).

11/18/2013

Srini - working on database access code. Few problems with python, but working on it (will send email describing problem & error if would be helpful to get problem resolved). Some problems with needing to know table names, temporary solution adopted is to make them hard-coded.  Will commit to new git repo today or tomorrow. 

Lynne -  Working on grids and gridMetrics, found that making ‘globalGrid’ be the histogram bins turned out to be very cumbersome at the gridMetric level, and was clunky in writing code for the globalGrid itself (I found myself re-implementing matplotlib’s histogram functions). Also realized that this histogram bin grid makes it more difficult to calculate mean/rms/etc of the whole dataset (which is what we would want to do, rather than calculating the mean of the histogram). Solution - make the ‘globalGrid’ operate on all the data (so the ‘grid’ = 1 point, which == all pointings). This makes gridMetric work very cleanly with both spatial and global grids, lets us keep the same histogram code for both spatial and global grids (and it’s matplotlib, which ought to make it more flexible in choice of bins and should be faster). Also offers the interesting possibility of subclassing to subdivide opsim data on categories other than RA/Dec (i.e. all observations, but binned in time - see ‘timestepGlobalGrid’ for an example that’s not fully fleshed out, but shows the bones).  I will start moving this code to the new git repo that Simon set up.  

 Srini - had some troubles with LSST stack & getting it set up. Some with csh loading, some with just unfamiliarity with the stack. Got things rolling now. Has 'Table' class ready, working on db access layer. Will do some more work on db access layer, but probably won't get to joins today. Will commit to git before leaves on vacation (vacation rest of the week). Srini will also provide a version of a v3.0 database output for testing.

Lynne - pretty happy with grids I think, still working on evolving gridMetric toward the class it should be (mostly related to making sure functional for both running/generating metrics and for reading metrics to generate or later plots). Started playing with complex metrics which return more than a scalar, to test what extra we need from 'reduce' functions (and just from knowing length of returning data quantity at each gridpoint). Will talk to Simon about his suggestions today and continue working on complex metric / gridMetric. 
As Peter is in Chile and Srini on vacation the rest of the week, stand up is cancelled for the rest of the week.  To resume next Monday, hopefully with a working interface to the DB. 

 11/25/2013

Srini - was on vacation, will work on finishing Database classes, will coordinate with Simon about numpy rec array returns on Table class

 Lynne - worked on gridMetric interface to grid + metrics & pushed current state back to git, will work on complex metrics & ‘reduce' functionality, and adding the call to this into gridMetric 

Peter - was at conference, will work on handling metadata (this will include config details) & comments, making sure these can propagate to output files 

12/11/2013 

Working on design review documentation and checking for general debugging of test examples.