obiwan.dplearn.create_training
saves 64x64 pixel cutouts of each source in a Data Release as HDF5 files
Classes
Functions
flux2mag(nmgy) |
|
get_ELG_box(rz, gr[, pad]) |
|
get_xy_pad(slope, pad) |
Returns dx,dy |
mpi_main([nproc, which, outdir, ls_dir, …]) |
| param nproc: | > 1 for mpi4py |
|---|
|
testcase_main() |
|
y1_line(rz[, pad]) |
|
y2_line(rz[, pad]) |
|
-
class
obiwan.dplearn.create_training.SimStamps(ls_dir=None, outdir=None, savedir=None, jpeg=False)[source]
Object for exracting sim cutouts
| Parameters: |
- ls_dir – LEGACY_SURVEY_DIR, like ‘tests/end_to_end/testcase_DR5_grz’
- outdir – path to dir containing obiwan,coadd,tractor dirs
|
|---|
For each id,x,y in self.cat, extracts image cutout
| Parameters: | hw – half-width, pixels, (hw*2) x (hw*2) image cutout |
|---|
-
load_data(brick, cat_fn, coadd_dir)[source]
loads coadd and catalogue data
| Parameters: |
- brick –
- coadd_dir – path/to/rs0, rs300, rs300_skipid, etc
|
|---|
-
run(brick, stampSize=64, applyCuts=True, zoom=None)[source]
Write the hdf5 image files for all rs/* in this brick
| Parameters: |
- brick – brickname
- stampSize – height and width in pixes of training image
- zoom – if legacypipe was run with zoom option
|
|---|
-
set_paths_to_data(brick)[source]
lists of catalogues filenames and coadd dirs
-
class
obiwan.dplearn.create_training.TractorStamps(ls_dir=None, outdir=None, savedir=None, jpeg=False)[source]
For each id,x,y in self.cat, extracts image cutout
| Parameters: | hw – half-width, pixels, (hw*2) x (hw*2) image cutout |
|---|
-
isFaint_cut(df)[source]
There are only faint sources in the deep2 matched sample,
but in the tractor catalogus have a bright population presumably
stars. Remove these
| Parameters: | df – pd.DataFrame have tractor cat extinction corrected grz mags |
|---|
-
load_data(brick, cat_fn, coadd_dir)
loads coadd and catalogue data
| Parameters: |
- brick –
- coadd_dir – path/to/rs0, rs300, rs300_skipid, etc
|
|---|
-
run(brick, stampSize=64, applyCuts=True, zoom=None)
Write the hdf5 image files for all rs/* in this brick
| Parameters: |
- brick – brickname
- stampSize – height and width in pixes of training image
- zoom – if legacypipe was run with zoom option
|
|---|
-
set_paths_to_data(brick)[source]
lists of catalogues filenames and coadd dirs
-
sim_sampling_cut(df)[source]
same cut applied to simulated sources
| Parameters: | df – pd.DataFrame have tractor cat extinction corrected grz mags |
|---|
-
class
obiwan.dplearn.create_training.UserDefinedStamps(ls_dir=None, outdir=None, savedir=None, jpeg=False)[source]
For each id,x,y in self.cat, extracts image cutout
| Parameters: | hw – half-width, pixels, (hw*2) x (hw*2) image cutout |
|---|
-
load_data(brick, cat_fn, coadd_dir)
loads coadd and catalogue data
| Parameters: |
- brick –
- coadd_dir – path/to/rs0, rs300, rs300_skipid, etc
|
|---|
-
run(brick, stampSize=64, applyCuts=True, zoom=None)
Write the hdf5 image files for all rs/* in this brick
| Parameters: |
- brick – brickname
- stampSize – height and width in pixes of training image
- zoom – if legacypipe was run with zoom option
|
|---|
-
set_paths_to_data(brick)[source]
lists of catalogues filenames and coadd dirs
-
obiwan.dplearn.create_training.get_ELG_box(rz, gr, pad=None)[source]
| Parameters: |
- rz – r-z
- gr – g-r
- pad – magnitudes of padding to expand TS box
|
|---|
-
obiwan.dplearn.create_training.get_xy_pad(slope, pad)[source]
Returns dx,dy
-
obiwan.dplearn.create_training.mpi_main(nproc=1, which=None, outdir=None, ls_dir=None, savedir=None, jpeg=False, bricks=[])[source]
| Parameters: |
- nproc – > 1 for mpi4py
- which – one of [‘tractor’,’sim’,’userDefined’]
- outdir – path to coadd,tractor dirs
- ls_dir – not needed if legacy_survey_dir env var already set
- savedir – where to write the hdf5 files, outdir if None
- jpeg – extract .jpg instead of .fits
- bricks – list bricks to make hdf5 cutouts from
|
|---|