Working SB2 version?

[Raindogjones]

Hi,
I've been trying to tweak the 'new-sb2' branch to get it working but to no avail. Does anyone have any ideas as to what needs to be done? Or, even better, know of a working sb2 branch? I'm happy to dig into the code myself but I don't really know where to start.
Thanks!

[adrn]

Can you try that branch again? I just rebased it on main, and I believe main is working. If you hit an error / failure, can you post it here?

[Raindogjones]

Thanks a lot for looking into this, @adrn.
I tried that but ran into an issue with pymc3, so I switched the import in prior_sb2.py from pymc3 to pymc. That then leads to an error when trying to sample the priors:

---------------------------------------------------------------------------
UnitsError                                Traceback (most recent call last)
Cell In[10], line 1
----> 1 prior_samples = prior.sample(size=1_000)

File ~/jokersb2/lib/python3.12/site-packages/thejoker/prior_sb2.py:198, in JokerSB2Prior.sample(self, size, generate_linear, return_logprobs, random_state, dtype, **kwargs)
    195     elif p.name not in s._valid_units.keys():
    196         continue
--> 198     s[name] = np.atleast_1d(raw_samples[p.name]) * unit
    200 if return_logprobs:
    201     s['ln_prior'] = log_prior

File ~/jokersb2/lib/python3.12/site-packages/thejoker/samples.py:189, in JokerSamples.__setitem__(self, key, val)
    187 expected_unit = self._valid_units[key]
    188 if not val.unit.is_equivalent(expected_unit):
--> 189     raise u.UnitsError(
    190         f"Units of '{key}' must be convertable to " f"{expected_unit}"
    191     )
    193 self.tbl[key] = val

UnitsError: Units of 'P' must be convertable to d

Even though my prior on the Period is set with units: P = xu.with_unit(pm.Normal("P", 2500.0, 1000.), u.d)