the kernel appears to have died. it will restart automatically

[cmencar]

#3285 Description of your problem

I am running the following code within Jupyter Notebook (the code is adapted from ch. 2 of "Bayesian Methods for Hackers
by Cameron Davidson-Pilon", see notebook, cells 24--26.

import pymc3 as pm
import scipy.stats as stats
#figsize(12, 4)

#these two quantities are unknown to us.
true_p_A = 0.05
true_p_B = 0.04

#notice the unequal sample sizes -- no problem in Bayesian analysis.
N_A = 15000
N_B = 750

#generate some observations
observations_A = stats.bernoulli.rvs(true_p_A, size=N_A)
observations_B = stats.bernoulli.rvs(true_p_B, size=N_B)

with pm.Model() as model:
    p_A = pm.Uniform("p_A", 0, 1)
    p_B = pm.Uniform("p_B", 0, 1)
    
    # Define the deterministic delta function. This is our unknown of interest.
    delta = pm.Deterministic("delta", p_A - p_B)

    
    # Set of observations, in this case we have two observation datasets.
    obs_A = pm.Bernoulli("obs_A", p_A, observed=observations_A)
    obs_B = pm.Bernoulli("obs_B", p_B, observed=observations_B)

    # To be explained later.
    step = pm.Metropolis()
    trace = pm.sample(20000, step=step)
    burned_trace=trace[1000:]

I run the above code, then re-run it by changing

true_p_B = 0.54

At the second run, the execution will stop at the end of sampling with a pop-up message "the kernel appears to have died. it will restart automatically" (no other messages displayed). Since then, I am unable to use pymc3 unless I reboot the machine.
Same problems by porting the code into Spyder.

edit Unfortunately I am unable to always reproduce the error. It seems to have a randomic behavior.

Please provide any additional information below.

Versions and main components

• PyMC3 Version: 3.9.2
• Theano Version: 1.0.4
• Python Version: 3.7.7
• Operating system: Linux Mint 20
• How did you install PyMC3: (conda/pip) conda

[fonnesbeck]

Since this is not reproducible, I'm going to close the issue. Please feel free to reopen if the problem continues.