pycbc_inference_models_stat fails when waveform generation fails

[yi-fan-wang]

I was trying to use pycbc_inference_models_stat to reconstruct the parameters that had been marginalized , but encountered the following error:

"""
Traceback (most recent call last):
  File "/usr/lib/python3.11/multiprocessing/pool.py", line 125, in worker
    result = (True, func(*args, **kwds))
                    ^^^^^^^^^^^^^^^^^^^
  File "/work/yifanwang/ecc/pe/env-2025-10-11/lib/python3.11/site-packages/pycbc/inference/models/__init__.py", line 49, in _call_global_model
    return _global_instance(*args, **kwds)  # pylint:disable=not-callable
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/work/yifanwang/ecc/pe/env-2025-10-11/bin/pycbc_inference_model_stats", line 106, in callmodel
    rec = model.reconstruct(seed=iteration)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/work/yifanwang/ecc/pe/env-2025-10-11/lib/python3.11/site-packages/pycbc/inference/models/tools.py", line 795, in reconstruct
    rec['loglr'] = loglr[xl]
                   ~~~~~^^^^
TypeError: 'float' object is not subscriptable
"""

This was accompanied by a large number of waveform generation failure messages. My suspicion is that loglr might be a None object because none of the waveform generations succeeded. This may indicate the parameter reconstruction doesn't properly deal with waveform generation failures. The complete log can be found here.

@ahnitz Do you have any idea how to fix it?