Form a modeltime-length array for all random deviations R0
When is_projecting, fill in gaps with rnorm, either random-walking or dnorm as appropriate.
Ideally we don't want a parameter per-projection year, but some way of forcing e.g. no-spawning years. Maybe work this into the formula? if (cur_year == bad_year) 0 else rnorm(...).
Maybe we should have a extra_actions attribute that g3_step() sorts out, to simplify this setup?
lentinj
commented
1 month ago
Make what's going on in g3_GoB the default:
https://github.com/vbartolino/g3_GoB/blob/45abba5f3d457bde81295f81edacedb253f6b818/vendace_g3/gadget3/09-SRR-RND/setup-random.R#L1-L15
In
g3l_random_dnorm, or a wrapper:R0rnorm, either random-walking or dnorm as appropriate.Ideally we don't want a parameter per-projection year, but some way of forcing e.g. no-spawning years. Maybe work this into the formula?
if (cur_year == bad_year) 0 else rnorm(...).Maybe we should have a
extra_actionsattribute thatg3_step()sorts out, to simplify this setup?