search

bomeara / selac

Model by O'Meara, Chai, and Gilchrist for SELection on Amino acids and/or Codons
1 stars 7 forks source link

Runge-Kutta (ode45) working properly #13

Closed mikegilchrist closed 7 years ago

mikegilchrist commented 7 years ago

The problem wasn't actually with the RK method, but the erroneous test for probability values for a specific aa state. The appropriate test was whether the sum of probability values across aa state was 0. I have not checked any other ode solvers, but ode45 gives a 20 fold speed up over lsoda. As a result, I recommend checking the other solvers and updating ode routines to use the fastest one.

mikegilchrist commented 7 years ago

Will fix travis error by removing testing code (which should not be here anyway).