should we include data for k(T, P0) = 0?

[jonwzheng]

Overview

There are several data I found in the Glarborg libraries (one, two, three) that report k(T,P0)=0 where P0 is some low pressure.

Question

Should we include these?
(1) They don't actually provide any functional dependence on T; in other words they're not "real" kinetic parameters.
(2) Because the rates are reported as 0, the reverse rate is infinity which is handled as nan, which can cause errors downstream (for instance, currently cannot be rendered on the RMG website [pending a patch]).
(3) I'm assuming these are correctly parsed, but I don't know what manuscripts these came from. Is it possible these are placeholder values?

Arguments in favor of including: they do convey information (e.g. basically saying that the rates below a certain pressure are too slow to observe) and someone simulating a system at low pressure might productively use this data.

[xiaoruiDong]

Maybe I am wrong, but I think the overall k at those extreme conditions is not 0, as not all components of the PdepMultiArrhenius at the conditions are 0. I tried P = 0.00001 atm in the k evaluator in this page, and they are indeed non-zero.

The error you mentioned is when searching the kinetics for the reverse direction, right? Hmmm... Maybe for multi-arrhenius objects, Arrhenius are fitted individually when trying to get the reverse kinetics?

[jonwzheng]

Ah, I see... you are correct, I misunderstood these PdepMultiArrhenius entries. I guess it's no problem that k2 and k3 are reported as 0 in terms of overall rate.

The only real issue then is handling the reverse rate fitting. Yes, I think they're individually fitted, which is why nan is observed (see ReactionMechanismGenerator/RMG-website#260).