AR-atmospheres-py
Python version of autoregressive atmosphere generator
Relevant paper:
A computationally efficient autoregressive method for generating phase screens with frozen flow and turbulence in optical simulations
Srikar Srinath, Univ. of California, Santa Cruz; Lisa A. Poyneer, Lawrence Livermore National Lab.;
Alexander R. Rudy, UCSC; S. Mark Ammons, LLNL
Published in Optics Express 23, 33335-33349 (2015)
http://dx.doi.org/10.1364/OE.23.033335
Programs for basic operations:
make_ar_atmos: A program that generates a datacube (2 spatial dimensions x time) in fits format (hdf5 outout not implemented yet) of stacked phase screens with the power-scaling law supplied (Kolmogorov, von Karman etc.) and specified turbulence strength. Required inputs are exposure time (in seconds), system rate (in Hz), alpha magnitude (< 1), number of subapertures across the wavefront sensor (n) and resolution of the output phase screen (m, pixels / subaperture). The eventual datacube is (n * m) x (n * m) x (exposure time * rate). For non-AO applications, (n, m) and (expsore time, rate) could just be combined into one variable. Simulations where memory is not constrained, but computation resources are sould use this to generate a phase screen cube beforehand.
create_multilayer_phasecube: AR phase screen generator for use in LSST sims by Jim Chiang
Use these two functions in connjunction for AR phase screen generation & evolution on the fly in simulations. Allows
for arbitrarily long exposure times (where arbitrary is limited by storage and a few other constraints.).
create_multilayer_arbase: Use outside the simulation loop to generate alpha vectors and powerlaw scaling for each layer
get_ar_atmos: Takes initial FT-ed phase screen (generated by get_phase_streamlined, for example), alpha and powerlaw (from
create_multilayer_arbase) and returns new phase and FT of phase for the next time step in a simulation. The first time
this function is called, input FT of phase can just be an array of zeroes.
ARscreens: Class structure for AR screen generation by Jim Chiang
Utilities:
cdr_create_parameters.py: Returns an array with three selected wind layer characteristics (measurements based on GPI data). For each selected layer, the array contains r0, wind velocity (m/s), wind direction (degrees) and layer altitude (m).
check_ar_atmos: Currently set up for GPI only. Checks phase variance and PSDs of phase screen data cubes for sanity.
depiston: Removes piston from residual phase. Takes a 2-D phase array as input and, optionally, the aperture mask.
detilt : Remves tip/tilt from residual phase. Takes a 2-D phase array and the aperture mask as inputs.
gen_avg_per_unb: Generates an averaged, unbiased periodogram. Takes 1-D data, a periodogram length and various options such as half overlapping (to improve SNR), mean removal, hanning or hamming windows, whether to use fftw or fft (in python the scipack fft utilities are just as efficient as fftw).
generate grids: Generates x, y grids that are needed for things like spatial frequency grid generation etc.
get_phase_streamlined: Returns a phase screen of given dimension (n x n) starting from scaled (Kolmogorov power spectrum), white noise. Used to generate the starting phase screen in any simulation/datacube.
radialprofile: return an azimuthally averaged radial profile of a 2-D array. Useful for PSF profile analysis. Initial source of power differences found between AR and frozen flow phase screens
rebin: Resizes a 2d array by averaging or repeating elements.
Fork-Hints:
This is a forked repository from SimonP2207/ARatmospy which is also a fork from shrieks/ARatmospy.
Changelog:
- Replaced
xrangewithrangefor Python3 compatibillity - Added type-hints (not
mypyproof) - Adapted deprecated code to get rid of warnings
- Minor code changes when examples caused an error
The changes are not properly tested and therefore the code is hopefully correct. If you have any improvements, just open a PR.