This plot is from MWATelescope/Birli#203. It shows how much the UVWs Birli writes change between Marlu 0.19.0, which agrees with cotter, and the unreleased precession on Marlu main (e739762: rotate by the apparent sidereal time; 150623a: take the phase centre's aberration out with a rigid rotation). The same changes are on feat/conventions, via Marlu#43.
The data are Birli's test files for 1254670392 (tests/data/1254670392_avg), using 2 coarse channels (gpubox 23 and 24) and 2 timesteps. Both Birli builds need aoflagger (apt-get install aoflagger-dev) for --no-rfi.
git clone https://github.com/MWATelescope/Birli && cd Birli
D=tests/data/1254670392_avg
ARGS="-m $D/1254670392.fixed.metafits --no-draw-progress --pfb-gains none \
--no-cable-delay --no-geometric-delay --no-rfi --emulate-cotter --no-flag-dc \
--flag-init 0 $D/1254670392_20191009153257_gpubox24_00.fits \
$D/1254670392_20191009153257_gpubox23_00.fits"
# "old": Birli main, Marlu 0.19.0 from crates.io (cotter's frame)
git checkout cf741e5
cargo build --release --features aoflagger
./target/release/birli -u old.uvfits $ARGS
# "new": Birli#203 before the cotter-frame fix, on Marlu feat/conventions
git checkout e02aa91
cargo build --release --features aoflagger
./target/release/birli -u new.uvfits $ARGS
pip install numpy astropy matplotlib
python3 uvw_change.py old.uvfits new.uvfits birli_uvw_change.pngThe script reads UU/VV/WW (in seconds, multiplied by c) and checks that the two files' baselines line up row for row. It then plots |Δu|, |Δv| and |Δw| in mm against baseline length, with Marlu's stated 8 cm/km bound on w and the 1e-5 m tolerance Birli's cotter tests use. The second panel shows the phase change 2π|Δw|/λ at 200 MHz, which is what the --emulate-cotter geometric-correction tests see. It prints the maximum differences. For this run the largest |Δw| is 4.9 cm on the 740 m baseline (6.9e-5 of the baseline length), which is 11.8° at 200 MHz.
- Later commits on Birli#203 make
--emulate-cotteruse Marlu's newUvwFrame::Cotter. Running the "new" step from the current branch tip with--emulate-cottertherefore gives (near) zero difference. To see the default frame's change from that tip, drop--emulate-cotter. - The uvfits random-group parameters are float32, so differences below ~1e-7 of the baseline length are rounding.
