Growth function oddness

[bthe]

When poking the growth equation I noticed an oddity that I hadn't thought about before. When the beta-binomial (implemented at https://github.com/gadget-framework/gadget3/blob/master/R/action_grow.R#L66 ) is supplied with a delta_l > binn the sum of the entries is no longer 1, e.g.

sum(growth_bbinom(10,1,1000))

results in 19, and thus could lead to increasing the number of fish in the model. Although I guess this is not very likely to occur in a final model, as it is common practice to set this value roughly as the maximum possible growth, we should probably think of something to ensure that this does not cause problems.

[lentinj]

FWIW you can reproduce this with the real thing via:

g3_eval(gadget3:::g3a_grow_impl_bbinom(
    delta_len_f = 10,
    delta_wgt_f = NA,
    beta_f = 1000,
    maxlengthgroupgrowth = 1)$len, list(stock__plusdl = 1))

as it is common practice to set this value roughly as the maximum possible growth,

I think there's some naming muddling here, which isn't helping.

gadget3/R/action_grow.R

Lines 57 to 61 in f21bdc9

	g3a_grow_impl_bbinom <- function (
	delta_len_f, delta_wgt_f,
	beta_f = g3_parameterized('bbin', by_stock = by_stock),
	maxlengthgroupgrowth,
	by_stock = TRUE) {

gadget3/R/action_grow.R

Lines 63 to 66 in f21bdc9

	##' @param delt_l Vector, for each lengthgroup, mean # of length groups to grow by
	##' @param binn Maximum possible number of length groups to grow by
	##' @return length(delt_l) x (length(delt_l) + 1) 2-dimensional array, initial_len -> growth jump
	growth_bbinom <- g3_native(r = function (delt_l, binn, beta) {

gadget3/R/action_grow.R

Line 132 in f21bdc9

~growth_bbinom(avoid_zero_vec((delta_len_f) / stock__plusdl), maxlengthgroupgrowth, avoid_zero(beta_f)),

binn within growth_bbinom() is always maxlengthgroupgrowth. But we make a default parameter for beta_f called "stock_name.binn".