beta_j0_hum * (1 + a1 cos(2 pi t/p) + b1 sin(2 pi t/p) +
a2 cos(4 pi t/p) + b2 sin(4 pi t/p)).
Arguments
- par
Parameters from
sim_params().
Details
t is 1-indexed, not 0-indexed — t = 1:nticks. The
Python implementation writes np.arange(0, nticks) + 1 with the comment
"R is 1-indexed, so we start at 1", i.e. it already matches an R convention.
Using 0:(nticks-1) here would phase-shift the whole seasonal envelope
by one day.
Calendar phase. The coefficients are estimated on calendar
day-of-year (est_seasonal_dynamics(), t = 1 is 1 January), so tick
i is evaluated at t = par$season_t0 + i, where
season_t0 is the day-of-year of date_start minus one (set by
sim_params(); 0 for a 1 January start, and taken as 0 when absent).