Create marginal expected benefit functions for a gpl_df

The marginal expected benefit of allocating to target i is $$\Lambda_i(x) = -\frac{1}{w_i}\frac{d}{dx} E L_i(x, Y_i) = \frac{\kappa_i}{w_i} g'(x) (\alpha_i - F_i(x)),$$ a decreasing function of x. allocate() equalizes it across targets.

Usage

meb_gpl_df(df, F, w)

Arguments

df

a gpl_df, as created by new_gpl_df().

F

list of predictive cdfs, one per row of df.

w

weights (costs per unit allocated) used in the budget constraint.

Value

A list of functions of an allocation x, one per target.

Examples

gdf <- new_gpl_df(N = 2, alpha = 0.5)
meb <- meb_gpl_df(gdf, F = list(pnorm, pnorm), w = 1)
meb[[1]](0)
#> [1] 0