Ph"tddlmZddlZddlmZddlmZddlmZddl m Z m Z ddl m Z dgZGd deZy) )DictN) constraints) Distribution) Independent)ComposeTransform Transform)_sum_rightmostTransformedDistributionceZdZUdZiZeeejfe d<dfd Z dfd Z ejddZ edZej"fd Zej"fd Zd Zd Zd ZdZxZS)r a Extension of the Distribution class, which applies a sequence of Transforms to a base distribution. Let f be the composition of transforms applied:: X ~ BaseDistribution Y = f(X) ~ TransformedDistribution(BaseDistribution, f) log p(Y) = log p(X) + log |det (dX/dY)| Note that the ``.event_shape`` of a :class:`TransformedDistribution` is the maximum shape of its base distribution and its transforms, since transforms can introduce correlations among events. An example for the usage of :class:`TransformedDistribution` would be:: # Building a Logistic Distribution # X ~ Uniform(0, 1) # f = a + b * logit(X) # Y ~ f(X) ~ Logistic(a, b) base_distribution = Uniform(0, 1) transforms = [SigmoidTransform().inv, AffineTransform(loc=a, scale=b)] logistic = TransformedDistribution(base_distribution, transforms) For more examples, please look at the implementations of :class:`~torch.distributions.gumbel.Gumbel`, :class:`~torch.distributions.half_cauchy.HalfCauchy`, :class:`~torch.distributions.half_normal.HalfNormal`, :class:`~torch.distributions.log_normal.LogNormal`, :class:`~torch.distributions.pareto.Pareto`, :class:`~torch.distributions.weibull.Weibull`, :class:`~torch.distributions.relaxed_bernoulli.RelaxedBernoulli` and :class:`~torch.distributions.relaxed_categorical.RelaxedOneHotCategorical` arg_constraintsct|tr |g|_nCt|tr%t d|Ds t d||_nt d||j |jz}t|j}t|j}t||jjkr/t dj|jj||j|}|j|}||k7r"|dt||z } |j| }|jj|z } | dkDr t!|| }||_|j$j|jjz } t'|j$j|| z} t|| k\sJt|| z } |d| }|| d}t(|U|||y)Nc3<K|]}t|tywN) isinstancer).0ts wC:\Users\daisl\Desktop\realtime-object-detection\venv\Lib\site-packages\torch/distributions/transformed_distribution.py z3TransformedDistribution.__init__..6sDAz!Y/sz6transforms must be a Transform or a list of Transformsz0transforms must be a Transform or list, but was zHbase_distribution needs to have shape with size at least {}, but got {}.r validate_args)rr transformslistall ValueError batch_shape event_shapelenrdomain event_dimformat forward_shape inverse_shapeexpandr base_distcodomainmaxsuper__init__)selfbase_distributionrr base_shapebase_event_dim transformr!expanded_base_shapebase_batch_shapereinterpreted_batch_ndimstransform_change_in_event_dimrcutrr __class__s rr(z TransformedDistribution.__init__0s j) ,DO D )DDD L)DOB:,O  '225F5R5RR .::;$T__5 z?Y--77 7Zaa$$..   "// ; '55mD , ,2;#)*^;  !2 8 89I J $-$4$4$>$>$O! $q ( +!#<! +    ( (9+;+;+E+E E &    ( ( : : =!Y...- 9,#DS) #CD)  kOc|jt|}tj|}||jz}t |j D]}|j|}|dt|t|jjz }|jj||_ |j |_tt|3||jd|j|_ |S)NFr)_get_checked_instancer torchSizerreversedrr"rr$r#r'r(_validate_args)r)r _instancenewshaperr/r3s rr#zTransformedDistribution.expandfs(()@)Ljj- d...$//*AOOE*E+ !O3u:DNN4N4N0O#OP--.>?  %s4 )) 5 "00 r4F) is_discretec,|js|jjS|jdj}t |j |j kDr6tj|t |j |j z }|S)N) rr$supportr%rrrr independent)r)rAs rrAzTransformedDistribution.supportusx>>)) )//"%.. t 7#4#4 4!--T--.1B1BBGr4c.|jjSr)r$ has_rsample)r)s rrDz#TransformedDistribution.has_rsamples~~)))r4ctj5|jj|}|jD] }||} |cdddS#1swYyxYw)a Generates a sample_shape shaped sample or sample_shape shaped batch of samples if the distribution parameters are batched. Samples first from base distribution and applies `transform()` for every transform in the list. N)r7no_gradr$samplerr) sample_shapexr-s rrGzTransformedDistribution.samplesE]]_%%l3A!__ aL- __s 6AAcn|jj|}|jD] }||} |S)a$ Generates a sample_shape shaped reparameterized sample or sample_shape shaped batch of reparameterized samples if the distribution parameters are batched. Samples first from base distribution and applies `transform()` for every transform in the list. )r$rsamplerrHs rrLzTransformedDistribution.rsamples4 NN " "< 0I! A)r4c(|jr|j|t|j}d}|}t |j D]{}|j |}||jj|jjz z }|t|j||||jjz z }|}}|t|jj||t|jjz z}|S)z Scores the sample by inverting the transform(s) and computing the score using the score of the base distribution and the log abs det jacobian. g)r:_validate_samplerrr9rinvrrr%r log_abs_det_jacobianr$log_prob)r)valuerrQyr-rJs rrQz TransformedDistribution.log_probs     ! !% ((()  !$//2I a A ))33i6H6H6R6RR RI...q!4I,,666#HA3n NN # #A & C8R8R4S(S  r4cd}|jD]}||jz}t|tr|dk(r|S||dz zdzS)zu This conditionally flips ``value -> 1-value`` to ensure :meth:`cdf` is monotone increasing. g?)rsignrint)r)rRrVr-s r_monotonize_cdfz'TransformedDistribution._monotonize_cdfsN I)..(D) dC TQYLus{#c))r4c|jdddD]}|j|}|jr|jj ||jj |}|j |}|S)z Computes the cumulative distribution function by inverting the transform(s) and computing the score of the base distribution. Nr@)rrOr:r$rNcdfrXr)rRr-s rrZzTransformedDistribution.cdfsm 2.IMM%(E/    NN + +E 2""5)$$U+ r4c|j|}|jj|}|jD] }||} |S)z Computes the inverse cumulative distribution function using transform(s) and computing the score of the base distribution. )rXr$icdfrr[s rr]zTransformedDistribution.icdfsE $$U+##E*Ie$E) r4r)__name__ __module__ __qualname____doc__r rstrr Constraint__annotations__r(r#dependent_propertyrApropertyrDr7r8rGrLrQrXrZr] __classcell__)r3s@rr r s@:rps0 +97F4 $ %JlJr4