AMReX-Astro
For aprox13, rate tabulation should be the default
This gives fairly accurate results relative to the direct rate evaluation method, but is much faster on CPUs and essentially necessary for GPUs.
This can be done by setting use_tables to .true. in the aprox13/_parameters file.
interestingly, with tables, aprox13 integration with test_react does more RHS evals:
no table:
Execution time: 30.443917989730835
RHS stats:
min: 6
avg: 1522
max: 23965
tables:
Execution time: 26.683586120605469
RHS stats:
min: 6
avg: 1916
max: 46530
Here's the comparison for the test:
variable name absolute error relative error
(||A - B||) (||A - B||/||A||)
----------------------------------------------------------------------
level = 1
density 0.000000000 0.000000000
temperature 0.000000000 0.000000000
Xnew_helium-4 0.7446594393E-06 0.7446594402E-06
Xnew_carbon-12 0.1807416444E-06 0.5839670568E-06
Xnew_oxygen-16 0.4457918293E-06 0.7767817760E-06
Xnew_neon-20 0.1002733683E-06 0.3127164189E-06
Xnew_magnesium-24 0.7251977366E-07 0.2474042924E-06
Xnew_silicon-28 0.5241993420E-06 0.8563144416E-06
Xnew_sulfur-32 0.3410176564E-06 0.1028486974E-05
Xnew_argon-36 0.1558357633E-06 0.4576690622E-06
Xnew_calcium-40 0.6082129856E-06 0.8300185222E-06
Xnew_titanium-44 0.7081178333E-06 0.1377724014E-05
Xnew_chromium-48 0.2046546471E-06 0.5383569083E-06
Xnew_iron-52 0.5256162670E-06 0.1038257153E-05
Xnew_nickel-56 0.6329660721E-06 0.6577058145E-06
Xold_helium-4 0.000000000 0.000000000
Xold_carbon-12 0.000000000 0.000000000
Xold_oxygen-16 0.000000000 0.000000000
Xold_neon-20 0.000000000 0.000000000
Xold_magnesium-24 0.000000000 0.000000000
Xold_silicon-28 0.000000000 0.000000000
Xold_sulfur-32 0.000000000 0.000000000
Xold_argon-36 0.000000000 0.000000000
Xold_calcium-40 0.000000000 0.000000000
Xold_titanium-44 0.000000000 0.000000000
Xold_chromium-48 0.000000000 0.000000000
Xold_iron-52 0.000000000 0.000000000
Xold_nickel-56 0.000000000 0.000000000
wdot_helium-4 0.7446594393E-03 0.8290511421E-06
wdot_carbon-12 0.1807416444E-03 0.6024918215E-06
wdot_oxygen-16 0.4457918293E-03 0.1486088692E-05
wdot_neon-20 0.1002733683E-03 0.3342445927E-06
wdot_magnesium-24 0.7251977365E-04 0.2474042925E-06
wdot_silicon-28 0.5241993421E-03 0.8563144418E-06
wdot_sulfur-32 0.3410176564E-03 0.1028486975E-05
wdot_argon-36 0.1558357633E-03 0.4576690623E-06
wdot_calcium-40 0.6082129856E-03 0.8300185223E-06
wdot_titanium-44 0.7081178333E-03 0.1377724014E-05
wdot_chromium-48 0.2046546471E-03 0.5383569084E-06
wdot_iron-52 0.5256162670E-03 0.1038257153E-05
wdot_nickel-56 0.6329660721E-03 0.6577058146E-06
rho_Hnuc 0.3877592594E+22 0.8471309253E-07
is there a reason to do aprox13, but not aprox19 and aprox21?
Also, are the tables compatible with #56?
is there a reason to do aprox13, but not aprox19 and aprox21?
Tabulation hasn't been implemented yet in those networks.
Also, are the tables compatible with #56?
Yes.
Closing as resolved -- whether or not to use rate tabulation for the aprox networks is now a Microphysics-wide parameter.