update PhysicalModel.get_band_fluxes() in physical_model.py

[mi-dai]

An error occured when filter_name in opsim was not included in PassbandGroup when calculating the fluxes using PhysicalModel.get_band_fluxes(). I added a check.

[hombit]

I'd fail instead - it would help catching typos in field names

Suggested change

	filter_mask = filters == filter_name
	band_fluxes[:, filter_mask] = np.nan
	raise ValueError(f"Filter '{filter_name}' is not found in the passbands object, available filters are {list(passband_or_group.passbands)}")

[mi-dai]

hmmm, then we'll have to modify opsim to make sure it doesn't output non-overlapping filters, or we'll now get this error instead.

[mi-dai]

I suggest we don't raise an error for now.

[jeremykubica]

What's the state of this PR? Is there consensus on whether to fail? If so, can we re-review and merge?

[mi-dai]

I think we bypass this issue by forcing the opsim to have same set of filters as passband_group. I'm closing it for now.

[github-actions]

Before [5ab23f3]	After [3df7107]	Ratio	Benchmark (Parameter)
5.83±0.9ms	5.05±1ms	~0.87	benchmarks.TimeSuite.time_chained_evaluate
2.31±0.8ms	1.52±0.8ms	~0.66	benchmarks.TimeSuite.time_fnu_to_flam
1.38±0.05ms	1.47±0.04ms	1.06	benchmarks.TimeSuite.time_apply_passbands
5.04±0.2ms	5.03±0.08ms	1.00	benchmarks.TimeSuite.time_evaluate_salt3_passbands
28.1±0.3μs	28.1±0.3μs	1.00	benchmarks.TimeSuite.time_make_new_salt3_model
128±2μs	128±2μs	1.00	benchmarks.TimeSuite.time_sample_x0_from_distmod
1.40±0s	1.39±0.01s	0.99	benchmarks.TimeSuite.time_make_x1_from_hostmass
8.94±0.1ms	8.77±0.09ms	0.98	benchmarks.TimeSuite.time_load_passbands
16.7±0.3μs	16.3±0.5μs	0.98	benchmarks.TimeSuite.time_sample_x1_from_hostmass
11.0±0.4ms	10.4±0.5ms	0.94	benchmarks.TimeSuite.time_evaluate_salt3_model

Click here to view all benchmarks.