How to implement bivariate poisson logbp

[alexcom]

Ok thanks, I already checked like this

index = pt.arange(10)

lams = pm.Exponential.dist(lam = 0.5, shape = (5,3))
lam1 = lams[:,0]
lam2 = lams[:,1]
lam3 = lams[:,2]
lam = (lam3 / (lam2 * lam1))[:,None]

resultado = lam ** index

And the print looks nice:

[0 1 2 3 4 5 6 7 8 9] (10,)
[[  3.92221234]
 [  6.99686109]
 [  0.72190472]
 [  1.06022012]
 [695.87921068]] (5,1)
[[1.00000000e+00 3.92221234e+00 1.53837496e+01 6.03383326e+01
  2.36659753e+02 9.28229802e+02 3.64071438e+03 1.42796549e+04
  5.60078385e+04 2.19674635e+05]
 [1.00000000e+00 6.99686109e+00 4.89560650e+01 3.42538786e+02
  2.39669631e+03 1.67693511e+04 1.17332820e+05 8.20961444e+05
  5.74415318e+06 4.01910418e+07]
 [1.00000000e+00 7.21904723e-01 5.21146429e-01 3.76218069e-01
  2.71593601e-01 1.96064703e-01 1.41540035e-01 1.02178420e-01
  7.37630839e-02 5.32499186e-02]
 [1.00000000e+00 1.06022012e+00 1.12406671e+00 1.19175814e+00
  1.26352596e+00 1.33961565e+00 1.42028747e+00 1.50581735e+00
  1.59649785e+00 1.69263915e+00]
 [1.00000000e+00 6.95879211e+02 4.84247876e+05 3.36978030e+08
  2.34496005e+11 1.63180895e+14 1.13554192e+17 7.90200018e+19
  5.49883765e+22 3.82652680e+25]] (5, 10)

But why I keep getting this error when I run the model and the logbp looking the same as the example above? And also why I can’t check the value of lam1, lam2 and lam3 inside of the logp using the function .eval()?

@alexcom

ValueError                                Traceback (most recent call last)
Cell In[106], line 2
      1 with model:
----> 2     trace = pm.sample(1000, tune=1500, cores = 4)
...
    456     )
    458 # inplace_pattern maps output idx -> input idx
    459 inplace_pattern = self.inplace_pattern

ValueError: Incompatible Elemwise input shapes [(1, 380), (1, 5)]