a ߙfb^/@sdZddlZddlZddlZddlmZddlmZm Z m Z dZ gdZ Gdd d Z Gd d d e ZGd d d e ZddZdS)al This module contains the convolution and filter functionalities of astropy. A few conceptual notes: A filter kernel is mainly characterized by its response function. In the 1D case we speak of "impulse response function", in the 2D case we call it "point spread function". This response function is given for every kernel by an astropy `FittableModel`, which is evaluated on a grid to obtain a filter array, which can then be applied to binned data. The model is centered on the array and should have an amplitude such that the array integrates to one per default. Currently only symmetric 2D kernels are supported. N)AstropyUserWarning)discretize_modeladd_kernel_arrays_1Dadd_kernel_arrays_2Dd)KernelKernel1DKernel2Dkernel_arithmeticsc@seZdZdZdZdZdZddZeddZ ed d Z ed d Z ed dZ eddZ d&ddZeddZeddZeddZddZddZddZd d!Zd"d#Zd'd$d%ZdS)(rzq Convolution kernel base class. Parameters ---------- array : ndarray Kernel array. FTNcCst||_dS)N)npZ asanyarray_array)selfarrayr7lib/python3.9/site-packages/astropy/convolution/core.py__init__,szKernel.__init__cCs|jS)z: Deviation from the normalization to one. )Z _truncationrrrr truncation/szKernel.truncationcCs|jS)z Indicates if kernel is bool. If the kernel is bool the multiplication in the convolution could be omitted, to increase the performance. )_is_boolrrrris_bool6szKernel.is_boolcCs|jS)z( Kernel response model. )_modelrrrrmodel@sz Kernel.modelcCs|jjS)z# Kernel dimension. )rndimrrrr dimensionGszKernel.dimensioncCsdd|jjDS)z- Index of the kernel center. cSsg|] }|dqS)r).0Z axes_sizerrr Sz!Kernel.center..r shaperrrrcenterNsz Kernel.centerintegralcCsh|dkr|j}n|dkr(|j}ntd|dkrFtdtnt|j||j|j|_ dS)ag Normalize the filter kernel. Parameters ---------- mode : {'integral', 'peak'} One of the following modes: * 'integral' (default) Kernel is normalized such that its integral = 1. * 'peak' Kernel is normalized such that its peak = 1. r"Zpeakz*invalid mode, must be 'integral' or 'peak'rz8The kernel cannot be normalized because it sums to zero.N) r summax ValueErrorwarningswarnrr ZdivideZ _kernel_sum)rmodeZ normalizationrrr normalizeUs  zKernel.normalizecCs|jjS)z, Shape of the kernel array. rrrrrr ssz Kernel.shapecCs|jS)a Indicates if the filter kernel is separable. A 2D filter is separable, when its filter array can be written as the outer product of two 1D arrays. If a filter kernel is separable, higher dimension convolutions will be performed by applying the 1D filter array consecutively on every dimension. This is significantly faster, than using a filter array with the same dimension. ) _separablerrrr separablezs zKernel.separablecCs|jS)z& Filter kernel array. r rrrrrsz Kernel.arraycCs t||dS)z) Add two filter kernels. addr rkernelrrr__add__szKernel.__add__cCs t||dS)z. Subtract two filter kernels. subr.r/rrr__sub__szKernel.__sub__cCs t||dSzF Multiply kernel with number or convolve two kernels. mulr.rvaluerrr__mul__szKernel.__mul__cCs t||dSr4r.r6rrr__rmul__szKernel.__rmul__cCs|jS)z5 Array representation of the kernel. r,rrrr __array__szKernel.__array__cCst|dtjkrtS|SdS)z? Wrapper for multiplication with numpy arrays. rN)typer ZufuncNotImplemented)rrcontextrrr__array_wrap__szKernel.__array_wrap__)r")N)__name__ __module__ __qualname____doc__r*rrrpropertyrrrrr!r)r r+rr1r3r8r9r:r>rrrrrs8         rcs"eZdZdZdfdd ZZS)r u4 Base class for 1D filter kernels. Parameters ---------- model : `~astropy.modeling.FittableModel` Model to be evaluated. x_size : int or None, optional Size of the kernel array. Default = ⌊8*width+1⌋. Only used if ``array`` is None. array : ndarray or None, optional Kernel array. width : number Width of the filter kernel. mode : str, optional One of the following discretization modes: * 'center' (default) Discretize model by taking the value at the center of the bin. * 'linear_interp' Discretize model by linearly interpolating between the values at the corners of the bin. * 'oversample' Discretize model by taking the average on an oversampled grid. * 'integrate' Discretize model by integrating the model over the bin. factor : number, optional Factor of oversampling. Default factor = 10. Nc s|jr|durtd|dur&|j}n|t|kr:td|ddkrjt| ddt|ddf}n&t|d dt|dddf}t|j|fi|}n|durtdt|dS)N-Array argument not allowed for kernel models.x_size should be an integerrr?r#Must specify either array or model.r TypeErrorZ _default_sizeintrsuperr)rrx_sizerkwargsx_range __class__rrrs  $&zKernel1D.__init__)NNNr?r@rArBr __classcell__rrrOrr s r cs"eZdZdZdfdd ZZS)r u Base class for 2D filter kernels. Parameters ---------- model : `~astropy.modeling.FittableModel` Model to be evaluated. x_size : int, optional Size in x direction of the kernel array. Default = ⌊8*width + 1⌋. Only used if ``array`` is None. y_size : int, optional Size in y direction of the kernel array. Default = ⌊8*width + 1⌋. Only used if ``array`` is None, array : ndarray or None, optional Kernel array. Default is None. mode : str, optional One of the following discretization modes: * 'center' (default) Discretize model by taking the value at the center of the bin. * 'linear_interp' Discretize model by performing a bilinear interpolation between the values at the corners of the bin. * 'oversample' Discretize model by taking the average on an oversampled grid. * 'integrate' Discretize model by integrating the model over the bin. width : number Width of the filter kernel. factor : number, optional Factor of oversampling. Default factor = 10. Nc sD|jr"|durtd|dur(|j}n|t|krs >K