fn
i1
→Tensori1(x: Tensor)Modified Bessel function of the first kind, order 1.
Computes , the order-one solution of the modified Bessel equation that is regular at the origin. Used in Rician statistics, Bessel / von Mises – Fisher distributions, and rotational-symmetric kernels.
Parameters
xTensorInput tensor; any floating-point dtype.
Returns
Tensor element-wise, same shape and dtype as x.
Notes
Series definition:
The implementation uses the two-branch Abramowitz & Stegun §9.8
polynomial approximation: a power series in for
, and an asymptotic expansion in
scaled by for
. Accuracy is roughly over the
real line. I_1 is odd, I_1(0) = 0, and
as .
Examples
>>> import lucid
>>> from lucid.special import i1
>>> i1(lucid.tensor([0.0, 1.0, 5.0]))
Tensor([0.0000, 0.5652, 24.3356])