DP3¶

Flint can run steps of DP3 against a measurement set. DP3 is installed from pip with the dp3 extra (Linux x86_64 only):

pip install 'askap-flint[dp3]'

DP3 is driven through its Python interface, so no container is needed. Each task builds the DP3 settings in memory and makes a single pass over the measurement set, updating it in place.

Tasks¶

The tasks live in flint.dp3:

dp3_flag: flags with aoflagger, using the packaged ASKAP.lua strategy unless another is given. A preflagger step can run first.
dp3_ddecal: solves for gains against the model column with ddecal, and returns the measurement set with h5parm set to the solutions.
dp3_gaincal: the same, with gaincal.
dp3_applycal: applies one or more solution tables from an h5parm, writing the corrected data to a new column (CORRECTED_DATA by default). The measurement set’s h5parm is used when none is given.
dp3_solve_apply: solves and applies in one call. With gaincal this is a single pass over the data. With ddecal the solutions are written to an h5parm and then applied.

Calibration solves against the measurement set’s model_column (e.g. MODEL_DATA written by wsclean), which must be set.

Each step has an options class (e.g. DP3DDECalOptions) whose fields are DP3 settings for that step. Unset options are left to DP3’s defaults. They can also be set in a strategy file under the dp3preflagger, dp3aoflagger, dp3ddecal, dp3gaincal and dp3applycal modes.

Accessing via the CLI¶

Run DP3 steps against measurement sets through DP3’s Python interface

usage: flint_dp3 [-h] {flag,ddecal,gaincal,applycal} ...

Positional Arguments¶

task: Possible choices: flag, ddecal, gaincal, applycal

Sub-commands¶

flag¶

Flag with aoflagger

flint_dp3 flag [-h] [--column COLUMN] [--strategy STRATEGY]
               [--memoryperc MEMORYPERC] [--timewindow TIMEWINDOW]
               ms

Positional Arguments¶

ms: The measurement set to flag

Named Arguments¶

--column

The column to flag

Default: 'DATA'

Inputs for DP3AOFlaggerOptions¶

--strategy: The aoflagger .lua strategy file. The packaged ASKAP strategy is used if unset.
--memoryperc: Percentage of system memory aoflagger may use
--timewindow: Number of time slots flagged together

ddecal¶

Solve for gains with ddecal

flint_dp3 ddecal [-h] [--column COLUMN] [--model-column MODEL_COLUMN]
                 [--h5parm H5PARM]
                 [--mode {scalar,scalarphase,scalaramplitude,diagonal,diagonalphase,diagonalamplitude,fulljones,tec,tec+delay,tec+phase,tec+phase+delay,rotation,rotation+diagonal,faradayrotation,leakage,leakageamplitude}]
                 [--solint SOLINT] [--nchan NCHAN] [--maxiter MAXITER]
                 [--tolerance TOLERANCE] [--uvlambdamin UVLAMBDAMIN]
                 [--propagatesolutions PROPAGATESOLUTIONS]
                 ms

Positional Arguments¶

ms: The measurement set to solve

Named Arguments¶

--column

The data column

Default: 'DATA'

--model-column

The model column

Default: 'MODEL_DATA'

--h5parm

The output h5parm

Inputs for DP3DDECalOptions¶

--mode

Possible choices: scalar, scalarphase, scalaramplitude, diagonal, diagonalphase, diagonalamplitude, fulljones, tec, tec+delay, tec+phase, tec+phase+delay, rotation, rotation+diagonal, faradayrotation, leakage, leakageamplitude

What to solve for

Default: 'scalarphase'

--solint

Number of time slots per solution. 0 uses the whole observation.

Default: 1

--nchan

Number of channels per solution. 0 uses the whole band.

Default: 0

--maxiter

Maximum number of solver iterations

--tolerance

Convergence tolerance of the solver

--uvlambdamin

Ignore baselines shorter than this many wavelengths

--propagatesolutions

Start each solution interval from the previous solutions

gaincal¶

Solve for gains with gaincal

flint_dp3 gaincal [-h] [--column COLUMN] [--model-column MODEL_COLUMN]
                  [--h5parm H5PARM]
                  [--caltype {scalar,scalarphase,scalaramplitude,diagonal,diagonalphase,diagonalamplitude,fulljones,tec,tec+phase}]
                  [--solint SOLINT] [--nchan NCHAN] [--maxiter MAXITER]
                  [--tolerance TOLERANCE] [--uvlambdamin UVLAMBDAMIN]
                  [--propagatesolutions PROPAGATESOLUTIONS]
                  ms

Positional Arguments¶

ms: The measurement set to solve

Named Arguments¶

--column

The data column

Default: 'DATA'

--model-column

The model column

Default: 'MODEL_DATA'

--h5parm

The output h5parm

Inputs for DP3GainCalOptions¶

--caltype

Possible choices: scalar, scalarphase, scalaramplitude, diagonal, diagonalphase, diagonalamplitude, fulljones, tec, tec+phase

What to solve for

Default: 'scalarphase'

--solint

Number of time slots per solution. 0 uses the whole observation.

Default: 1

--nchan

Number of channels per solution. 0 uses the whole band.

Default: 0

--maxiter

Maximum number of solver iterations

--tolerance

Convergence tolerance of the solver

--uvlambdamin

Ignore baselines shorter than this many wavelengths

--propagatesolutions

Start each solution interval from the previous solutions

applycal¶

Apply an h5parm

flint_dp3 applycal [-h] [--column COLUMN] [--output-column OUTPUT_COLUMN]
                   [--correction CORRECTION [CORRECTION ...]]
                   [--solset SOLSET] [--interpolation {nearest,linear}]
                   [--updateweights UPDATEWEIGHTS] [--invert INVERT]
                   ms h5parm

Positional Arguments¶

ms: The measurement set to correct
h5parm: The h5parm to apply

Named Arguments¶

--column

The data column

Default: 'DATA'

--output-column

The corrected data column

Default: 'CORRECTED_DATA'

Inputs for DP3ApplyCalOptions¶

--correction

Soltabs to apply in turn, e.g. [‘phase000’] or [‘rotationmeasure000’]

Default: ['phase000']

--solset

The solset holding the soltabs. DP3 picks one if unset.

--interpolation

Possible choices: nearest, linear

How solutions are interpolated in time and frequency

--updateweights

Update the weights along with the data

--invert

Apply the inverse of the solutions, i.e. correct the data