I can call pm.draw on the “likelihood” function and resulting a 1x20 array similar to the displacement that I was hoping, taking quite a while per draw (±20s). Seems like the sampling on using “pm.sample” is taking a long time based on this drawing time on my device.
That’s the intent to have system of equation sampled as mu, centered on “fe” function and compare with covariance “cov” a set of constant matrix and observation “y” which is basically a comparison set of data.
Thanks for the help, Jesse.