GOW Temperature Handling Differs From Athena

[BenWibking]

Summary

JAFF substitutes each parsed KIDA rate's tgas with the row's
Tmin/Tmax-clipped value. Athena++ floors the gas temperature globally with
temp_min_rates_, but by default does not clip two-body rates to each row's
temperature range.

Assuming Athena++ is correct, JAFF changes GOW rates outside their tabulated
temperature ranges.

Affected Behavior

This affects ordinary KIDA rows that rely on parsed formula 3, 4, or 5 rates,
for example the non-special two-body GOW reactions with finite temperature
ranges.

The collapsed ID 30 row in networks/GOW/GOW.jet is not separately semantic
because Athena's two source rows use identical coefficients. The semantic
difference is the JAFF temperature clipping applied to parsed rates.

Reference Behavior

Athena++ defaults:

is_Tcap_2body_ = false;
temp_min_rates_ = 1.;

With the default, Athena++ computes ordinary two-body rates from the floored
gas temperature T, not from each row's clipped Tcap.

Current JAFF Behavior

JAFF builds a local substitution:

local_subs_dict[tgas] = Max(Min(tgas, tmax), tmin)

whenever parsed KIDA rows have finite temperature bounds.

Proposed Patch

Add an explicit temperature-policy option for KIDA parsing, with an
Athena-compatible policy for GOW:

temperature_policy = "athena"

Under the Athena-compatible policy:

• do not apply per-row Tmin/Tmax clipping unless requested;
• apply a global Max(tgas, temp_min_rates) floor, defaulting to 1.0 K;
• keep the existing per-row clipping available for users who want the current
  JAFF behavior.

Add tests that compare a bounded GOW formula-3 row below Tmin and above
Tmax against the Athena-style extrapolated expression.

Acceptance Criteria

• GOW can be loaded in Athena-compatible mode without per-row clipping.
• The default choice is documented clearly.
• Existing users can still opt into the old clipped-rate behavior.

[XRByte]

I don't believe this is gonna affect anything

[BenWibking]

I don't believe this is gonna affect anything

I'm not sure how that can be the case. It would seem like extrapolating or clipping the temperature for each rate would give completely different results, unless I'm missing something?

[XRByte]

Sorry. I misunderstood the issue. Yeah, JAFF clips the temperature. Andrea did run the GOW network recently with this implementation and it seemed to agree well with Munan Gong's PDR simulation - both temperature, rates and energy functions. Since extrapolation and clipping are both assumptions, I guess it's better to modify the max and min temperature in the network files instead of modifying the code. Maybe @markkrumholz can weigh in on this?

[BenWibking]

Ok, I see. I agree it is not clear what the best thing to do is, so it would be good to clarify the extrapolation/clipping policy for this network (and perhaps in general).