pm.HalfStudentT.dist() failing to give correct shape

ricardoV94 · April 7, 2023, 2:06pm

@Nn_Nnn Thanks for reporting, it seems there is a bug in how we generate random draws for the HalfStudentT distribution!

github.com

pymc-devs/pymc/blob/b7764dd09b9266006f24db021f1c4a8b825d7534/pymc/distributions/continuous.py#L2483


      
          
          
class HalfStudentTRV(RandomVariable):
              name = "halfstudentt"
              ndim_supp = 0
              ndims_params = [0, 0]
              dtype = "floatX"
              _print_name = ("HalfStudentT", "\\operatorname{HalfStudentT}")
          
          
    @classmethod
              def rng_fn(cls, rng, nu, sigma, size=None) -> np.ndarray:
                  return np.asarray(np.abs(stats.t.rvs(nu, sigma, size=size, random_state=rng)))
          
          

          
halfstudentt = HalfStudentTRV()
          
          

          
class HalfStudentT(PositiveContinuous):
              r"""
              Half Student's T log-likelihood.
          
          
    The pdf of this distribution is

sigma should be passed by keyword as scale, otherwise it’s used for loc

Topic		Replies	Views
Can't recover the scale parameter in Student's T model Questions	3	694	October 6, 2017
Name 'v' is not defined when trying to use StudentT distribution Questions	6	422	August 24, 2021
Why GLM module doesn't have HalfStudentT distribution in Family?	1	552	September 7, 2021
Combining two distributions Questions	11	2051	February 1, 2021
HalfStudentT distribution for the Likelyhood. The derivative is zero problem Questions	10	1263	September 23, 2021

pm.HalfStudentT.dist() failing to give correct shape

Related topics