a
    h#                     @   s   d dl Z d dlmZmZ d dlZd dlmZ d dlmZ d dlm	Z	 d dl
mZmZmZmZmZ d dlmZ d dlmZmZmZ d	gZG d
d	 d	e	ZdS )    N)OptionalUnion)Tensor)constraints)ExponentialFamily)broadcast_allclamp_probslazy_propertylogits_to_probsprobs_to_logits) binary_cross_entropy_with_logits)_Number_sizeNumberContinuousBernoullic                       sr  e Zd ZdZejejdZejZdZ	dZ
d2eeeef  eeeef  eeef ee dd fdd	Zd3 fd
d	Zdd Zdd Zdd Zdd ZeedddZeedddZeedddZeedddZeedddZeej ddd Z!e  fd!d"Z"e  fe#ed#d$d%Z$d&d' Z%d(d) Z&d*d+ Z'd,d- Z(eee dd.d/Z)d0d1 Z*  Z+S )4r   a  
    Creates a continuous Bernoulli distribution parameterized by :attr:`probs`
    or :attr:`logits` (but not both).

    The distribution is supported in [0, 1] and parameterized by 'probs' (in
    (0,1)) or 'logits' (real-valued). Note that, unlike the Bernoulli, 'probs'
    does not correspond to a probability and 'logits' does not correspond to
    log-odds, but the same names are used due to the similarity with the
    Bernoulli. See [1] for more details.

    Example::

        >>> # xdoctest: +IGNORE_WANT("non-deterministic")
        >>> m = ContinuousBernoulli(torch.tensor([0.3]))
        >>> m.sample()
        tensor([ 0.2538])

    Args:
        probs (Number, Tensor): (0,1) valued parameters
        logits (Number, Tensor): real valued parameters whose sigmoid matches 'probs'

    [1] The continuous Bernoulli: fixing a pervasive error in variational
    autoencoders, Loaiza-Ganem G and Cunningham JP, NeurIPS 2019.
    https://arxiv.org/abs/1907.06845
    )probslogitsr   TNgV-?gx&1?)r   r   limsvalidate_argsreturnc                    s   |d u |d u krt d|d urjt|t}t|\| _|d ur\| jd | j s\t dt| j| _n"|d usvJ t|t}t|\| _	|d ur| jn| j	| _
|rt }n
| j
 }|| _t j||d d S )Nz;Either `probs` or `logits` must be specified, but not both.r   z&The parameter probs has invalid valuesr   )
ValueError
isinstancer   r   r   arg_constraintscheckallr   r   _paramtorchSizesize_limssuper__init__)selfr   r   r   r   Z	is_scalarbatch_shape	__class__ V/var/www/auris/lib/python3.9/site-packages/torch/distributions/continuous_bernoulli.pyr#   7   s(    



zContinuousBernoulli.__init__c                    s~   |  t|}| j|_t|}d| jv r>| j||_|j|_d| jv r^| j	||_	|j	|_t
t|j|dd | j|_|S )Nr   r   Fr   )Z_get_checked_instancer   r!   r   r   __dict__r   expandr   r   r"   r#   _validate_args)r$   r%   Z	_instancenewr&   r(   r)   r+   W   s    


zContinuousBernoulli.expandc                 O   s   | j j|i |S N)r   r-   )r$   argskwargsr(   r(   r)   _newe   s    zContinuousBernoulli._newc                 C   s,   t t | j| jd t | j| jd S )Nr      )r   maxler   r!   gtr$   r(   r(   r)   _outside_unstable_regionh   s    $z,ContinuousBernoulli._outside_unstable_regionc                 C   s&   t |  | j| jd t | j S )Nr   )r   wherer7   r   r!   	ones_liker6   r(   r(   r)   
_cut_probsm   s
    zContinuousBernoulli._cut_probsc              	   C   s   |   }tt|d|t|}tt|d|t|}ttt	| t| tt|dt	d| td| d  }t
| jd d}tddd|  |  }t|  ||S )zLcomputes the log normalizing constant as a function of the 'probs' parameter      ?g              @      ?   gUUUUUU?g'}'}@)r:   r   r8   r4   
zeros_likeger9   logabslog1ppowr   mathr7   )r$   	cut_probsZcut_probs_below_halfZcut_probs_above_halflog_normxtaylorr(   r(   r)   _cont_bern_log_normt   s&    
z'ContinuousBernoulli._cont_bern_log_norm)r   c                 C   sj   |   }|d| d  dt| t|   }| jd }dddt|d  |  }t|  ||S )Nr<   r=   r;   gUUUUUU?gll?r>   )r:   r   rC   rA   r   rD   r8   r7   )r$   rF   ZmusrH   rI   r(   r(   r)   mean   s    
zContinuousBernoulli.meanc                 C   s   t | jS r.   )r   sqrtvariancer6   r(   r(   r)   stddev   s    zContinuousBernoulli.stddevc                 C   s   |   }||d  tdd|  d dtt| t| d  }t| jd d}ddd|  |  }t|  ||S )Nr=   r<   r>   r;   gUUUUUU?g?ggjV?)r:   r   rD   rC   rA   r   r8   r7   )r$   rF   varsrH   rI   r(   r(   r)   rM      s     zContinuousBernoulli.variancec                 C   s   t | jddS NT)Z	is_binary)r   r   r6   r(   r(   r)   r      s    zContinuousBernoulli.logitsc                 C   s   t t| jddS rP   )r   r
   r   r6   r(   r(   r)   r      s    zContinuousBernoulli.probsc                 C   s
   | j  S r.   )r   r    r6   r(   r(   r)   param_shape   s    zContinuousBernoulli.param_shapec                 C   sX   |  |}tj|| jj| jjd}t  | |W  d    S 1 sJ0    Y  d S N)dtypedevice)_extended_shaper   randr   rS   rT   Zno_gradicdfr$   sample_shapeshapeur(   r(   r)   sample   s    

zContinuousBernoulli.sample)rY   r   c                 C   s,   |  |}tj|| jj| jjd}| |S rR   )rU   r   rV   r   rS   rT   rW   rX   r(   r(   r)   rsample   s    
zContinuousBernoulli.rsamplec                 C   s8   | j r| | t| j|\}}t||dd |   S )Nnone)Z	reduction)r,   _validate_sampler   r   r   rJ   )r$   valuer   r(   r(   r)   log_prob   s    
zContinuousBernoulli.log_probc              
   C   s   | j r| | |  }t||td| d|  | d d| d  }t|  ||}tt|dt|tt	|dt
||S )Nr=   r<   g        )r,   r_   r:   r   rD   r8   r7   r4   r?   r@   r9   )r$   r`   rF   ZcdfsZunbounded_cdfsr(   r(   r)   cdf   s     


zContinuousBernoulli.cdfc              	   C   sT   |   }t|  t| |d| d   t|  t|t|   |S )Nr<   r=   )r:   r   r8   r7   rC   rA   )r$   r`   rF   r(   r(   r)   rW      s    
zContinuousBernoulli.icdfc                 C   s4   t | j }t | j}| j||  |   | S r.   )r   rC   r   rA   rK   rJ   )r$   Z
log_probs0Z
log_probs1r(   r(   r)   entropy   s    zContinuousBernoulli.entropyc                 C   s   | j fS r.   )r   r6   r(   r(   r)   _natural_params   s    z#ContinuousBernoulli._natural_paramsc                 C   s   t t || jd d t || jd d }t ||| jd d t | }t t t j	
|t t | }d| t |dd  t |dd  }t |||S )zLcomputes the log normalizing constant as a function of the natural parameterr   r;   r2   r>   g      8@   g     @)r   r3   r4   r!   r5   r8   r9   rA   rB   Zspecialexpm1rD   )r$   rH   Zout_unst_regZcut_nat_paramsrG   rI   r(   r(   r)   _log_normalizer   s    ((z#ContinuousBernoulli._log_normalizer)NNr   N)N),__name__
__module____qualname____doc__r   Zunit_intervalrealr   ZsupportZ_mean_carrier_measureZhas_rsampler   r   r   r   tuplefloatboolr#   r+   r1   r7   r:   rJ   propertyrK   rN   rM   r	   r   r   r   r   rQ   r\   r   r]   ra   rb   rW   rc   rd   rg   __classcell__r(   r(   r&   r)   r      sT       
 				)rE   typingr   r   r   r   Ztorch.distributionsr   Ztorch.distributions.exp_familyr   Ztorch.distributions.utilsr   r   r	   r
   r   Ztorch.nn.functionalr   Ztorch.typesr   r   r   __all__r   r(   r(   r(   r)   <module>   s   