Working with matrices speeds up the code by a factor of x40, 🎉
Another 20% one gains using @tullion for index loops.
Test
The Intervention let do a finding that test coverage on full model functionality was pure.
New tests with 20 chains of different k improve the coverage.
Breaking
Note, the PR might modify how SymbolicThreeBodyDecays.jl work due to
amplitude with two arguments is used for full_amplitude computation, but also for dalitz-plot point evaluation. The conflict is resolved by specifying the type of sigma for the first call.
It's not ideal, since SymbolicTBDs.jl won't work again
Closes #28 Closes #27
Speed up (
x60
)Working with matrices speeds up the code by a factor of
x40
, 🎉 Another 20% one gains using @tullion for index loops.Test
The Intervention let do a finding that test coverage on full model functionality was pure. New tests with 20 chains of different
k
improve the coverage.Breaking
SymbolicThreeBodyDecays.jl
work due tothat used to be from
two_λs
amplitude
with two arguments is used for full_amplitude computation, but also for dalitz-plot point evaluation. The conflict is resolved by specifying the type of sigma for the first call.It's not ideal, since
SymbolicTBDs.jl
won't work again