Closed mkolopanis closed 5 years ago
Here's another interesting thing, the data arrays for v0.2.0
and v0.1.0
are identical:
Comparing these two files: v0.1.0//simulation_results/ref_1.1_uniform.uvfits and v0.2.0/simulation_results/ref_1.1_uniform.uvfits
history parameter value is a string, values are different
parameter _history does not match. Left is Simulated with pyuvsim version: 0.1.0. Git origin: git@github.com:HERA-
Team/pyuvsim. Git hash: 9475ee15165c6c7f2bcc96b350c2d27eb646e28c. Git
branch: HEAD. Git description: v0.1.0-dirty. Sources from source list:
catalog_files/mock_catalog_heratext_2458098.38824015.txt. Based on confi
g files: ., telescope_config/mwa88_nocore_config.yaml, telescope_config/
mwa_nocore_layout.csv Npus = 20. Read/written with pyuvdata version: 1.
3.3. Git origin: https://github.com/HERA-Team/pyuvdata.git. Git hash:
1ceda98827043424cef97fcb6398ca40e1ccd93c. Git branch: master. Git desc
ription: v1.3-212-g1ceda98., right is Simulated with pyuvsim version: 0.2.0. Git origin: git@github.com:HERA-
Team/pyuvsim. Git hash: 08a3143afb9847879e7a597b49adf01c579aea8e. Git
branch: HEAD. Git description: v0.2.0. Sources from source list: catalo
g_files/mock_catalog_heratext_2458098.38824015.txt. Based on config file
s: ., telescope_config/mwa88_nocore_config.yaml, telescope_config/mwa_no
core_layout.csv Npus = 20. Read/written with pyuvdata version: 1.3.3.
Git origin: https://github.com/HERA-Team/pyuvdata.git. Git hash: 1ceda9
8827043424cef97fcb6398ca40e1ccd93c. Git branch: master. Git descriptio
n: v1.3-212-g1ceda98.
object_name parameter value is a string, values are different
parameter _object_name does not match. Left is zenith, right is catalog_files/mock_catalog_heratext_2458098.38824015.txt_ra60.8017_dec-30.7671
False
Data Array
Numpy all close: True
Numpy array_equal: True
pyuvsim seems to run on two different architectures. Interesting things could arise with a deeper dive, but this is good for now.
not 100% it's an issue, but wanted to at least track some outputs here for the oscar vs enterprise discussion.
After some, mishaps, I have managed to get output files from pyuvsim on enterprise. I've made a little script that compares two uvdata objects with the
==
operator to get the pyuvdata object, but also callsnp.allclose
andnp.array_equal
on the data arrays. The last two I've added because some changes have happened even in pyuvdata sincev0.1.0
so it's possible some of the meta data might be handled differently and I wanted explicit checks on the data_arrays as well.also these are being compared to values simulated with
v0.0.1
they were the reference simulations I had on hand but I might havev0.1.0
too somewhere.Here's some comparison files:
and the
v0.2.0
enterprise file:summary: Data arrays are "allclose" for reference sim 1.1 but not identical.