Where the stirred plateau ends: an inoculation rate of 1.4 B cells per gram.
Below it, growth is MAX_GROWTH_PER_GRAM. At and above it the fit
takes over. Kai's pseudo-code puts the boundary at exactly 1.4
(if (initial cells < 1.4) … if (initial cells between 1.4 and 3.5) …),
and the two branches don't meet. The plateau's 1.400 drops to the fit's
2.33 − 0.67 × 1.4 = 1.392 as soon as the rate reaches 1.4.
That 0.008 B/g step is in the published model and in every implementation
of it, so we keep it instead of smoothing the corner. Matching Brewer's
Friend's numbers matters more than a 0.6 % discontinuity nobody can
measure.
Where the stirred plateau ends: an inoculation rate of 1.4 B cells per gram.
Below it, growth is MAX_GROWTH_PER_GRAM. At and above it the fit takes over. Kai's pseudo-code puts the boundary at exactly 1.4 (
if (initial cells < 1.4) … if (initial cells between 1.4 and 3.5) …), and the two branches don't meet. The plateau's 1.400 drops to the fit's 2.33 − 0.67 × 1.4 = 1.392 as soon as the rate reaches 1.4.That 0.008 B/g step is in the published model and in every implementation of it, so we keep it instead of smoothing the corner. Matching Brewer's Friend's numbers matters more than a 0.6 % discontinuity nobody can measure.