pyklip.fmlib package

Submodules

pyklip.fmlib.diskfm module

pyklip.fmlib.extractSpec module

class pyklip.fmlib.extractSpec.ExtractSpec(inputs_shape, numbasis, sep, pa, input_psfs, input_psfs_wvs, input_psfs_pas=None, datatype='float', stamp_size=None)

Bases: pyklip.fmlib.nofm.NoFM

Planet Characterization class. Goal to characterize the astrometry and photometry of a planet

alloc_fmout(output_img_shape)

Allocates shared memory for the output of the shared memory

Parameters:output_img_shape – shape of output image (usually N,y,x,b) Returns:mp.array to store FM data in fmout_shape: shape of FM data array Return type:fmout

cleanup_fmout(fmout)

After running KLIP-FM, we need to reshape fmout so that the numKL dimension is the first one and not the last

Parameters:fmout – numpy array of ouput of FM Returns:same but cleaned up if necessary Return type:fmout

fm_from_eigen(klmodes=None, evals=None, evecs=None, input_img_shape=None, input_img_num=None, ref_psfs_indicies=None, section_ind=None, section_ind_nopadding=None, aligned_imgs=None, pas=None, wvs=None, radstart=None, radend=None, phistart=None, phiend=None, padding=None, IOWA=None, ref_center=None, parang=None, ref_wv=None, numbasis=None, fmout=None, perturbmag=None, klipped=None, flipx=True, **kwargs)

Generate forward models using the KL modes, eigenvectors, and eigenvectors from KLIP. Calls fm.py functions to perform the forward modelling

Parameters:

• klmodes – unpertrubed KL modes
• evals – eigenvalues of the covariance matrix that generated the KL modes in ascending order (lambda_0 is the 0 index) (shape of [nummaxKL])
• evecs – corresponding eigenvectors (shape of [p, nummaxKL])
• input_image_shape – 2-D shape of inpt images ([ysize, xsize])
• input_img_num – index of sciece frame
• ref_psfs_indicies – array of indicies for each reference PSF
• section_ind – array indicies into the 2-D x-y image that correspond to this section. Note needs be called as section_ind[0]
• pas – array of N parallactic angles corresponding to N reference images [degrees]
• wvs – array of N wavelengths of those referebce images
• radstart – radius of start of segment
• radend – radius of end of segment
• phistart – azimuthal start of segment [radians]
• phiend – azimuthal end of segment [radians]
• padding – amount of padding on each side of sector
• IOWA – tuple (IWA,OWA) where IWA = Inner working angle and OWA = Outer working angle both in pixels. It defines the separation interva in which klip will be run.
• ref_center – center of image
• numbasis – array of KL basis cutoffs
• parang – parallactic angle of input image [DEGREES]
• ref_wv – wavelength of science image
• fmout – numpy output array for FM output. Shape is (N, y, x, b)
• perturbmag – numpy output for size of linear perturbation. Shape is (N, b)
• klipped – PSF subtracted image. Shape of ( size(section), b)
• kwargs – any other variables that we don’t use but are part of the input

generate_models(input_img_shape, section_ind, pas, wvs, radstart, radend, phistart, phiend, padding, ref_center, parang, ref_wv, flipx, stamp_size=None)

Generate model PSFs at the correct location of this segment for each image denoated by its wv and parallactic angle

Parameters:

• pas – array of N parallactic angles corresponding to N images [degrees]
• wvs – array of N wavelengths of those images
• radstart – radius of start of segment
• radend – radius of end of segment
• phistart – azimuthal start of segment [radians]
• phiend – azimuthal end of segment [radians]
• padding – amount of padding on each side of sector
• ref_center – center of image
• parang – parallactic angle of input image [DEGREES]
• ref_wv – wavelength of science image
• stamp_size – size of the stamp for spectral extraction
• flipx – if True, flip x coordinate in final image

Returns:

array of size (N, p) where p is the number of pixels in the segment

Return type:

models

pyklip.fmlib.extractSpec.invert_spect_fmodel(fmout, dataset, method='JB', units='natural', scaling_factor=1.0)

1. Greenbaum Nov 2016

Parameters:

• fmout – the forward model matrix which has structure: [numbasis, n_frames, n_frames+1, npix]
• dataset – from GPI.GPIData(filelist) – typically set highpass=True also
• method – “JB” or “LP” to try the 2 different inversion methods (JB’s or Laurent’s)
• units – “natural” means the answer is scaled to the input PSF (default) fmout will be in these units. “scaled” means the output is scaled to “scaling_factor” argument
• scaling_factor – multiplies output spectrum and forward model, user set for desired calibration factor. units=”scaled” must be set in args for this to work!

Returns:

A tuple containing the spectrum and the forward model (spectrum, forwardmodel) spectrum shape:(len(numbasis), nwav)

pyklip.fmlib.fmpsf module

class pyklip.fmlib.fmpsf.FMPlanetPSF(inputs_shape, numbasis, sep, pa, dflux, input_psfs, input_wvs, flux_conversion=None, spectrallib=None, spectrallib_units='flux', star_spt=None, refine_fit=False, field_dependent_correction=None, input_psfs_pas=None)

Bases: pyklip.fmlib.nofm.NoFM

Forward models the PSF of the planet through KLIP. Returns the forward modelled planet PSF

alloc_fmout(output_img_shape)

Allocates shared memory for the output of the shared memory

Parameters:output_img_shape – shape of output image (usually N,y,x,b) Returns:mp.array to store FM data in fmout_shape: shape of FM data array Return type:fmout

alloc_perturbmag(output_img_shape, numbasis)

Allocates shared memory to store the fractional magnitude of the linear KLIP perturbation Stores a number for each frame = max(oversub + selfsub)/std(PCA(image))

Parameters:

• output_img_shape – shape of output image (usually N,y,x,b)
• numbasis – array/list of number of KL basis cutoffs requested

Returns:

mp.array to store linaer perturbation magnitude perturbmag_shape: shape of linear perturbation magnitude

Return type:

perturbmag

cleanup_fmout(fmout)

After running KLIP-FM, we need to reshape fmout so that the numKL dimension is the first one and not the last

Parameters:fmout – numpy array of ouput of FM Returns:same but cleaned up if necessary Return type:fmout

fm_from_eigen(klmodes=None, evals=None, evecs=None, input_img_shape=None, input_img_num=None, ref_psfs_indicies=None, section_ind=None, aligned_imgs=None, pas=None, wvs=None, radstart=None, radend=None, phistart=None, phiend=None, padding=None, IOWA=None, ref_center=None, parang=None, ref_wv=None, numbasis=None, fmout=None, perturbmag=None, klipped=None, covar_files=None, flipx=True, rdi_psfs=None, **kwargs)

Generate forward models using the KL modes, eigenvectors, and eigenvectors from KLIP. Calls fm.py functions to perform the forward modelling

Parameters:

• klmodes – unpertrubed KL modes
• evals – eigenvalues of the covariance matrix that generated the KL modes in ascending order (lambda_0 is the 0 index) (shape of [nummaxKL])
• evecs – corresponding eigenvectors (shape of [p, nummaxKL])
• input_image_shape – 2-D shape of inpt images ([ysize, xsize])
• input_img_num – index of sciece frame
• ref_psfs_indicies – array of indicies for each reference PSF
• section_ind – array indicies into the 2-D x-y image that correspond to this section. Note needs be called as section_ind[0]
• pas – array of N parallactic angles corresponding to N reference images [degrees]
• wvs – array of N wavelengths of those referebce images
• radstart – radius of start of segment
• radend – radius of end of segment
• phistart – azimuthal start of segment [radians]
• phiend – azimuthal end of segment [radians]
• padding – amount of padding on each side of sector
• IOWA – tuple (IWA,OWA) where IWA = Inner working angle and OWA = Outer working angle both in pixels. It defines the separation interva in which klip will be run.
• ref_center – center of image
• numbasis – array of KL basis cutoffs
• parang – parallactic angle of input image [DEGREES]
• ref_wv – wavelength of science image
• fmout – numpy output array for FM output. Shape is (N, y, x, b)
• perturbmag – numpy output for size of linear perturbation. Shape is (N, b)
• klipped – PSF subtracted image. Shape of ( size(section), b)
• ref_rdi_indices – array of N+M indicating N ADI/SDI images (0’s) and M RDI images (1;s).
• rdi_psfs – array of M RDI reference images in this sector.
• kwargs – any other variables that we don’t use but are part of the input

generate_models(input_img_shape, section_ind, pas, wvs, radstart, radend, phistart, phiend, padding, ref_center, parang, ref_wv, flipx, rdi_indices)

Generate model PSFs at the correct location of this segment for each image denoated by its wv and parallactic angle

Parameters:

• pas – array of N parallactic angles corresponding to N images [degrees]
• wvs – array of N wavelengths of those images
• radstart – radius of start of segment
• radend – radius of end of segment
• phistart – azimuthal start of segment [radians]
• phiend – azimuthal end of segment [radians]
• padding – amount of padding on each side of sector
• ref_center – center of image
• parang – parallactic angle of input image [DEGREES]
• ref_wv – wavelength of science image
• flipx – if True, flip x coordinate in final image
• rdi_indices – array of N corresponding to whether it is (1) or isn’t (0) an RDI frame

Returns:

array of size (N, p) where p is the number of pixels in the segment

Return type:

models

save_fmout(dataset, fmout, outputdir, fileprefix, numbasis, klipparams=None, calibrate_flux=False, spectrum=None, pixel_weights=1)

Saves the FM planet PSFs to disk. Saves both a KL mode cube and spectral cubes

Parameters:

• dataset – Instruments.Data instance. Will use its dataset.savedata() function to save data
• fmout – the fmout data passed from fm.klip_parallelized which is passed as the output of cleanup_fmout
• outputdir – output directory
• fileprefix – the fileprefix to prepend the file name
• numbasis – KL mode cutoffs used
• klipparams – string with KLIP-FM parameters
• calibrate_flux – if True, flux calibrate the data in the same way as the klipped data
• spectrum – if not None, spectrum to weight the data by. Length same as dataset.wvs
• pixel_weights – weights for each pixel for weighted mean. Leave this as a single number for simple mean

pyklip.fmlib.matchedFilter module

class pyklip.fmlib.matchedFilter.MatchedFilter(inputs_shape, numbasis, input_psfs, input_psfs_wvs, spectrallib, save_per_sector=None, datatype='float', fakes_sepPa_list=None, disable_FM=None, true_fakes_pos=None, ref_center=None, flipx=None, rm_edge=None, edge_threshold=None, planet_radius=None, background_width=None, save_bbfm=None, save_fm=None, save_fmout_path=None)

Bases: pyklip.fmlib.nofm.NoFM

Matched filter with forward modelling.

alloc_fmout(output_img_shape)

Allocates shared memory for the output of the shared memory

Parameters:output_img_shape – Not used Returns:mp.array to store auxilliary data in fmout_shape: shape of auxilliary array = (3,self.N_spectra,self.N_numbasis,self.N_frames,self.ny,self.nx)

The 3 is for saving the different term of the matched filter:

0: dot product 1: square of the norm of the model 2: Local estimated variance of the data 3: Number of pixels used in the matched filter

Return type:fmout

fm_end_sector(interm_data=None, fmout=None, sector_index=None, section_indicies=None)

Save the fmout object at the end of each sector if save_per_sector was defined when initializing the class.

fm_from_eigen(klmodes=None, evals=None, evecs=None, input_img_shape=None, input_img_num=None, ref_psfs_indicies=None, section_ind=None, section_ind_nopadding=None, aligned_imgs=None, pas=None, wvs=None, radstart=None, radend=None, phistart=None, phiend=None, padding=None, IOWA=None, ref_center=None, parang=None, ref_wv=None, numbasis=None, fmout=None, perturbmag=None, klipped=None, flipx=True, rdi_psfs=None, **kwargs)

Calculate and project the FM at every pixel of the sector. Store the result in fmout.

Parameters:

• klmodes – unpertrubed KL modes
• evals – eigenvalues of the covariance matrix that generated the KL modes in ascending order (lambda_0 is the 0 index) (shape of [nummaxKL])
• evecs – corresponding eigenvectors (shape of [p, nummaxKL])
• input_img_shape – 2-D shape of inpt images ([ysize, xsize])
• input_img_num – index of sciece frame
• ref_psfs_indicies – array of indicies for each reference PSF
• section_ind – array indicies into the 2-D x-y image that correspond to this section. Note needs be called as section_ind[0]
• pas – array of N parallactic angles corresponding to N reference images [degrees]
• wvs – array of N wavelengths of those referebce images
• radstart – radius of start of segment
• radend – radius of end of segment
• phistart – azimuthal start of segment [radians]
• phiend – azimuthal end of segment [radians]
• padding – amount of padding on each side of sector
• IOWA – tuple (IWA,OWA) where IWA = Inner working angle and OWA = Outer working angle both in pixels. It defines the separation interva in which klip will be run.
• ref_center – center of image
• numbasis – array of KL basis cutoffs
• parang – parallactic angle of input image [DEGREES]
• ref_wv – wavelength of science image
• fmout – numpy output array for FM output. Shape is (N, y, x, b)
• klipped – array of shape (p,b) that is the PSF subtracted data for each of the b KLIP basis cutoffs. If numbasis was an int, then sub_img_row_selected is just an array of length p
• rdi_psfs – array of M RDI reference images in this sector.
• kwargs – any other variables that we don’t use but are part of the input

generate_model_sci(input_img_shape, section_ind, pa, wv, radstart, radend, phistart, phiend, padding, ref_center, parang, ref_wv, sep_fk, pa_fk, flipx)

Generate model PSFs at the correct location of this segment of the science image denotated by its wv and parallactic angle.

Parameters:

• input_img_shape – 2-D shape of inpt images ([ysize, xsize])
• section_ind – array indicies into the 2-D x-y image that correspond to this section. Note needs be called as section_ind[0]
• pa – parallactic angle of the science image [degrees]
• wv – wavelength of the science image
• radstart – radius of start of segment (not used)
• radend – radius of end of segment (not used)
• phistart – azimuthal start of segment [radians] (not used)
• phiend – azimuthal end of segment [radians] (not used)
• padding – amount of padding on each side of sector
• ref_center – center of image
• parang – parallactic angle of input image [DEGREES] (not used)
• ref_wv – wavelength of science image
• sep_fk – separation of the planet to be injected.
• pa_fk – position angle of the planet to be injected.
• flipx – if True, flip x coordinate in final image

Return: (models, mask)

models: vector of size p where p is the number of pixels in the segment mask: vector of size p where p is the number of pixels in the segment

if pixel == 1: arc shape where to calculate the standard deviation if pixel == 2: 7 pixels disk around the position of the planet.

generate_models(input_img_shape, section_ind, pas, wvs, radstart, radend, phistart, phiend, padding, ref_center, parang, ref_wv, sep_fk, pa_fk, flipx, rdi_indices)

Generate model PSFs at the correct location of this segment for each image denotated by its wv and parallactic angle.

Parameters:

• input_img_shape – 2-D shape of inpt images ([ysize, xsize])
• section_ind – array indicies into the 2-D x-y image that correspond to this section. Note needs be called as section_ind[0]
• pas – array of N parallactic angles corresponding to N images [degrees]
• wvs – array of N wavelengths of those images
• radstart – radius of start of segment (not used)
• radend – radius of end of segment (not used)
• phistart – azimuthal start of segment [radians] (not used)
• phiend – azimuthal end of segment [radians] (not used)
• padding – amount of padding on each side of sector
• ref_center – center of image
• parang – parallactic angle of input image [DEGREES] (not used)
• ref_wv – wavelength of science image
• sep_fk – separation of the planet to be injected.
• pa_fk – position angle of the planet to be injected.
• flipx – if True, flip x coordinate in final image

Returns:

array of size (N, p) where p is the number of pixels in the segment

Return type:

models

save_fmout(dataset, fmout, outputdir, fileprefix, numbasis, klipparams=None, calibrate_flux=False, spectrum=None, pixel_weights=1)

Saves the fmout data to disk following the instrument’s savedata function

Parameters:

• dataset – Instruments.Data instance. Will use its dataset.savedata() function to save data
• fmout – the fmout data passed from fm.klip_parallelized which is passed as the output of cleanup_fmout
• outputdir – output directory
• fileprefix – the fileprefix to prepend the file name
• numbasis – KL mode cutoffs used
• klipparams – string with KLIP-FM parameters
• calibrate_flux – if True, flux calibrate the data (if applicable)
• spectrum – if not None, the spectrum to weight the data by. Length same as dataset.wvs
• pixel_weights – weights for each pixel for weighted mean. [Not used in matched filter]

skip_section(radstart, radend, phistart, phiend, flipx=True)

Returns a boolean indicating if the section defined by (radstart, radend, phistart, phiend) should be skipped. When True is returned the current section in the loop in klip_parallelized() is skipped.

Parameters:

• radstart – minimum radial distance of sector [pixels]
• radend – maximum radial distance of sector [pixels]
• phistart – minimum azimuthal coordinate of sector [radians]
• phiend – maximum azimuthal coordinate of sector [radians]
• flipx – if True, flip x coordinate in final image

Returns:

False so by default it never skips.

Return type:

Boolean

pyklip.fmlib.matchedFilter.calculate_fm_opti(delta_KL, original_KL, sci, model_sci_fk, delta_KL_fk, original_KL_fk)

Optimized version for calculate_fm() (if numbas) for a single numbasis.

Calculate what the PSF looks up post-KLIP using knowledge of the input PSF, assumed spectrum of the science target, and the partially calculated KL modes (Delta Z_k^lambda in Laurent’s paper). If inputflux is None, the spectral dependence has already been folded into delta_KL_nospec (treat it as delta_KL).

Note: if inputflux is None and delta_KL_nospec has three dimensions (ie delta_KL_nospec was calculated using pertrurb_nospec() or perturb_nospec_modelsBased()) then only klipped_oversub and klipped_selfsub are returned. Besides they will have an extra first spectral dimension.

Parameters:

• delta_KL_nospec – perturbed KL modes but without the spectral info. delta_KL = spectrum x delta_Kl_nospec. Shape is (numKL, wv, pix). If inputflux is None, delta_KL_nospec = delta_KL
• orignal_KL – unpertrubed KL modes (array of size [numbasis, numpix])
• sci – array of size p representing the science data
• model_sci – array of size p corresponding to the PSF of the science frame
• input_spectrum – array of size wv with the assumed spectrum of the model

If delta_KL_nospec does NOT include a spectral dimension or if inputflux is not None: :returns:

array of shape (b,p) showing the forward modelled PSF

Skipped if inputflux = None, and delta_KL_nospec has 3 dimensions.

klipped_oversub: array of shape (b, p) showing the effect of oversubtraction as a function of KL modes klipped_selfsub: array of shape (b, p) showing the effect of selfsubtraction as a function of KL modes Note: psf_FM = model_sci - klipped_oversub - klipped_selfsub to get the FM psf as a function of K Lmodes

(shape of b,p)

Return type:fm_psf

If inputflux = None and if delta_KL_nospec include a spectral dimension: :returns: Sum(<S|KL>KL) with klipped_oversub.shape = (1,Npix)

klipped_selfsub: Sum(<N|DKL>KL) + Sum(<N|KL>DKL) with klipped_selfsub.shape = (1,N_lambda or N_ref,N_pix)

Return type:klipped_oversub

pyklip.fmlib.nofm module

class pyklip.fmlib.nofm.NoFM(inputs_shape, numbasis)

Bases: object

Super class for all forward modelling classes. Has fall-back functions for all fm dependent calls so that each FM class does not need to implement functions it doesn’t want to. Should do no forward modelling and just do regular KLIP by itself

alloc_fmout(output_img_shape)

Allocates shared memory for the output of the forward modelling

Parameters:output_img_shape – shape of output image (usually N,y,x,b) Returns:mp.array to store FM data in fmout_shape: shape of FM data array Return type:fmout

alloc_interm(max_sector_size, numsciframes)

Allocates shared memory array for intermediate step

Intermediate step is allocated for a sector by sector basis

Parameters:max_sector_size – number of pixels in this sector. Max because this can be variable. Stupid rotating sectors Returns:mp.array to store intermediate products from one sector in interm_shape:shape of interm array (used to convert to numpy arrays) Return type:interm

alloc_output()

Allocates shared memory array for final output

Only use multiprocessing data structors as we are using the multiprocessing class

Args:

Returns:mp.array to store final outputs in (shape of (N*wv, y, x, numbasis)) outputs_shape: shape of outputs array to convert to numpy arrays Return type:outputs

alloc_perturbmag(output_img_shape, numbasis)

Allocates shared memory to store the fractional magnitude of the linear KLIP perturbation

Parameters:

• output_img_shape – shape of output image (usually N,y,x,b)
• numbasis – array/list of number of KL basis cutoffs requested

Returns:

mp.array to store linaer perturbation magnitude perturbmag_shape: shape of linear perturbation magnitude

Return type:

perturbmag

cleanup_fmout(fmout)

After running KLIP-FM, if there’s anything to do to the fmout array (such as reshaping it), now’s the time to do that before outputting it

Parameters:fmout – numpy array of ouput of FM Returns:same but cleaned up if necessary Return type:fmout

fm_end_sector(**kwargs)

Does some forward modelling at the end of a sector after all images have been klipped for that sector.

fm_from_eigen(**kwargs)

Generate forward models using the KL modes, eigenvectors, and eigenvectors from KLIP This is called immediately after regular KLIP

save_fmout(dataset, fmout, outputdir, fileprefix, numbasis, klipparams=None, calibrate_flux=False, spectrum=None, pixel_weights=1)

Saves the fmout data to disk following the instrument’s savedata function

Parameters:

• dataset – Instruments.Data instance. Will use its dataset.savedata() function to save data
• fmout – the fmout data passed from fm.klip_parallelized which is passed as the output of cleanup_fmout
• outputdir – output directory
• fileprefix – the fileprefix to prepend the file name
• numbasis – KL mode cutoffs used
• klipparams – string with KLIP-FM parameters
• calibrate_flux – if True, flux calibrate the data (if applicable)
• spectrum – if not None, the spectrum to weight the data by. Length same as dataset.wvs
• pixel_weights – weights for each pixel for weighted mean. Leave this as a single number for simple mean

skip_section(radstart, radend, phistart, phiend, flipx=None)

Returns a boolean indicating if the section defined by (radstart, radend, phistart, phiend) should be skipped. When True is returned the current section in the loop in klip_parallelized() is skipped.

Parameters:

• radstart – minimum radial distance of sector [pixels]
• radend – maximum radial distance of sector [pixels]
• phistart – minimum azimuthal coordinate of sector [radians]
• phiend – maximum azimuthal coordinate of sector [radians]
• flipx – if True, flip x coordinate in final image

Returns:

False so by default it never skips.

Return type:

Boolean

Module contents