:j d dl Z d dlmZ d dlZd dlZd dlmZ d dlmZ d dlm Z d dl
m Z m Z dgZ
d Zg d Zg d
Zg d Zg d ZeegZeegZdd
Zej, j. d Z G d de Zy) N)Optional)Tensor) constraints) Distribution)
broadcast_all
lazy_propertyVonMisesc r t | }|j }|r|j | |z z }|r|S N)listpop)ycoefresults r/home/mcse/projects/srt_converter/srt-converter-venv/lib/python3.12/site-packages/torch/distributions/von_mises.py
_eval_polyr s7 :D
XXZF
a&j( M )g ?g$ @g03@g,?N?g2t?gIx?gtHZr?) e3E?g -5?gՒ+Hub?gJNYgTPÂ?g 'gZ?gUL+ߐg;^p?) ?gY ?g(z?g*O?gZ9?g.h?gӰ٩=5?) r g .kg?VmgtZOZ?g< Q g' 8`?gP⥝gqJ:N?g;PJ4qc V |dk( s|dk( sJ | dz }||z }t |t | }|dk( r| j |z }|j }d| z }| d| j z z
t |t | j z }t j | dk || }|S )zX
Returns ``log(I_order(x))`` for ``x > 0``,
where `order` is either 0 or 1.
r g @r )r _COEF_SMALLabslog _COEF_LARGEtorchwhere)xorderr smalllarger s r _log_modified_bessel_fnr" E s
A:!# #
DA AA q+e, -E z% IIKE
qA
aeeg
1k%.@ A E E G GE
[[T5%
0F Mr c 6 t j |j t j | j }|j st j d|j z | j | j }|j \ }}}t j t j |z } d|| z z || z z }
|||
z
z } | d| z
z |z
dkD | |z j dz | z
dk\ z } | j r>t j | |dz
j |
j! z | }|| z }|j s|t j z | z dt j z z t j z
S )Ndtypedevice) r r r )r zerosshapeboolr& allrandr% unbindcosmathpir anyr signacos)
loc
concentration
proposal_rr doneuu1u2u3zfcaccepts
r _rejection_samplerA \ s< ;;qwwejj DDhhj JJtagg~SYYszz JXXZ
B IIdggl #
a
JN 3 Z!^ ,A;#q(a"f\\^a-?!-Cq-HI ::< FR#XOO$5$@!DA&=D hhj
K# !dgg+ . 88r c r e Zd ZdZej
ej dZej
ZdZ dde
de
de e ddf fd
Z
d
Zede
fd Zede
fd Zede
fd
Z ej( ej* fd Zd fd Zede
fd Zede
fd Zede
fd Z xZ S )r aX
A circular von Mises distribution.
This implementation uses polar coordinates. The ``loc`` and ``value`` args
can be any real number (to facilitate unconstrained optimization), but are
interpreted as angles modulo 2 pi.
Example::
>>> # xdoctest: +IGNORE_WANT("non-deterministic")
>>> m = VonMises(torch.tensor([1.0]), torch.tensor([1.0]))
>>> m.sample() # von Mises distributed with loc=1 and concentration=1
tensor([1.9777])
:param torch.Tensor loc: an angle in radians.
:param torch.Tensor concentration: concentration parameter
)r5 r6 FNr5 r6
validate_argsreturnc t || \ | _ | _ | j j }t j
}t | ||| y r )r r5 r6 r* r Sizesuper__init__)selfr5 r6 rC batch_shape event_shape __class__s r rH zVonMises.__init__ sD (5S-'H$$$hhnn jjl
k=Ar c | j r| j | | j t j || j
z
z }|t
j dt j z z
t | j d z
}|S )Nr( r r )
_validate_args_validate_sampler6 r r/ r5 r0 r r1 r" )rI valuelog_probs r rR zVonMises.log_prob sy ! !% (%% %$((2B(CC hhq477{#
$%d&8&8B
C
r c T | j j t j S r )r5 tor doublerI s r _locz
VonMises._loc s xx{{5<<((r c T | j j t j S r )r6 rT r rU rV s r _concentrationzVonMises._concentration s !!$$U\\22r c | j }ddd|dz z z j z }|d|z j z
d|z z }d|dz z d|z z }d|z |z }t j |dk || S )Nr r( gh㈵>)rY sqrtr r )rI kappataurho _proposal_r_proposal_r_taylors r r` zVonMises._proposal_r s ##1q5!8|#))++a#g^^%%!e)436za#g. Y .{{54<);[IIr c B | j | }t j || j j | j
j }t | j | j | j | j | j
j S )a
The sampling algorithm for the von Mises distribution is based on the
following paper: D.J. Best and N.I. Fisher, "Efficient simulation of the
von Mises distribution." Applied Statistics (1979): 152-157.
Sampling is always done in double precision internally to avoid a hang
in _rejection_sample() for small values of the concentration, which
starts to happen for single precision around 1e-4 (see issue #88443).
r$ ) _extended_shaper emptyrW r% r5 r& rA rY r` rT )rI sample_shaper* r s r samplezVonMises.sample sn $$\2 KKTYY__TXX__ M IIt**D,<,+. ." O MM - -.> ?M((//+.C . . 5 5k BM 4:c=
N N Os A.BBc | j S )z8
The provided mean is the circular one.
r5 rV s r meanz
VonMises.mean s
xxr c | j S r ro rV s r modez
VonMises.mode s xxr c dt | j d t | j d z
j z
S )z<
The provided variance is the circular one.
r rN r )r" r6 exprV s r variancezVonMises.variance s>
'(:(:!D)$*<*r s + 9 B ,
~. ~. . 9 9 n
| n
r