Skip to content

DDECal: large fraction of solutions flagged

Created by: darafferty

I am testing DDECal (latest master) within Factor (using a single direction), and am getting >99% solutions as flagged (weight = 0). I noticed that the number of iteration is very low. E.g.,

Total NDPPP time     106.76 real       84.77 user        0.42 system
   18.9% MSReader
   81.1% DDECal solvetec.
           64.9% of it spent in predict
           34.8% of it spent in estimating gains and computing residuals
                0.29% spent in constraints
            0.0% of it spent in writing gain solutions to disk
Iterations taken: [3|2,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|\
3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3\
,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,\
4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4\
|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|\
3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3\
,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3,4|3]
    0.0% MSUpdater msout.

I used the following parameters:

msin.datacolumn=DATA
msout=.
numthreads=1
solvetec.approximatetec=True
solvetec.detectstalling=False
solvetec.h5parm=h5parm
solvetec.maxapproxiter=50
solvetec.maxiter=75
solvetec.mode=tecandphase
solvetec.nchan=2
solvetec.propagatesolutions=True
solvetec.solint=1
solvetec.sourcedb=sourcedb
solvetec.stepsize=0.2
solvetec.tolerance=1e-05
solvetec.type=ddecal
solvetec.uvlambdamin=80.0
steps=[solvetec]

I tried changing the tolerance and stepsize but got similar results. The sourcedb and MS file are at ftp.hs.uni-hamburg.de/pub/outgoing/rafferty in case they're helpful.