Closed
Description
Describe the bug
When trying to run exoplanet/docs/tutorials/transit.py
, execution is stopped at the sampling stage on line 155: the stated error is TypeError: cannot pickle 'exoplanet.theano_ops.driver.LimbDark' object
.
To Reproduce
cd
into docs/tutorials
and run python transit.py
: output is reproduced below.
optimizing logp for variables: [b, r, u, logP, t0, mean]
106it [00:01, 82.32it/s, logp=2.479354e+04]
message: Desired error not necessarily achieved due to precision loss.
logp: 24787.977771807487 -> 24793.53942561123
Auto-assigning NUTS sampler...
Initializing NUTS using adapt_full...
/Users/adityasengupta/opt/anaconda3/envs/oracle/lib/python3.8/site-packages/pymc3/step_methods/hmc/quadpotential.py:503: UserWarning: QuadPotentialFullAdapt is an experimental feature
warnings.warn("QuadPotentialFullAdapt is an experimental feature")
Multiprocess sampling (2 chains in 2 jobs)
NUTS: [b, r, u, logP, t0, mean]
Traceback (most recent call last):
File "transit.py", line 149, in <module>
trace = pm.sample(
File "/Users/adityasengupta/opt/anaconda3/envs/oracle/lib/python3.8/site-packages/pymc3/sampling.py", line 545, in sample
trace = _mp_sample(**sample_args, **parallel_args)
File "/Users/adityasengupta/opt/anaconda3/envs/oracle/lib/python3.8/site-packages/pymc3/sampling.py", line 1470, in _mp_sample
sampler = ps.ParallelSampler(
File "/Users/adityasengupta/opt/anaconda3/envs/oracle/lib/python3.8/site-packages/pymc3/parallel_sampling.py", line 432, in __init__
step_method_pickled = pickle.dumps(step_method, protocol=-1)
TypeError: cannot pickle 'exoplanet.theano_ops.driver.LimbDark' object
Expected behavior
The same output as in the docs at https://docs.exoplanet.codes/en/stable/tutorials/transit/.
Your setup (please complete the following information):
- Version of exoplanet: 0.4.0
- Operating system: Mac OS Catalina v10.15.7
- Python version & installation method (pip, conda, etc.): Python 3.8.5, installation via pip in a conda env (+ git clone to get the file)
Additional context
Add any other context about the problem here.