Ambipolar bug about sparta

[aborner1]

@zhangchonglin Yes, I agree. I think it is solved, I will push an update very soon.

[aborner1]

Actually, if the case is ran long enough (> 1000 steps), the bug will appear for any number of ranks (even serial).

[aborner1]

As far as I can tell, the bug was introduced with version 5Sep18, as version 22Jun18 looks to give correct results.

[aborner1]

Bug was introduced by 2b038d8

[aborner1]

The bug is due to the following modification in react_bird.cpp:

// added MAX() limit, 24Aug18
tgamma(z+2.5-omega) / MAX(1.0e-6,tgamma(z+r->coeff[3]+1.5));

going back to
tgamma(z+2.5-omega) / tgamma(z+r->coeff[3]+1.5);

solves the problem.

[aborner1]

The following reactions should be commented out of the air.tce file as their temperature exponent is below -1.5, which causes the gamma function to have a native argument, which is not acceptable for the TCE model:

N2+ + O --> N2 + O+
E A 0.0 0.000 4.118e-11 -2.2 1.148e-19

O + e --> O+ + e + e
I A 0.0 2.188e-18 6.4761E3 -3.78 -2.188e-18

N + e --> N+ + e + e
I A 0.0 2.322e-18 4.1513E4 -3.82 -2.322e-18

[zhangchonglin]

@aborner1 I think I encountered similar issue as you did and reported it here: #10 (comment)

[zhangchonglin]

@aborner1 Thanks for the confirmation and fix!