pdg-control
Value of information
Exploring causes for deviation from S = D
in the Reed model: arises from self-sustaining assumption. Working out the interpretation of the self-sustaining clause, which sounds like it needs to ensure that the stock is non-decreasing in the absence of harvest,
\[ P(X_{t+1} \geq x | X_n = x ) = 1 \]
This sounds like a very awkward condition to enforce for a non-trivial escapement level \( x = S \). Why should the population be guarenteed to increase from it’s escapement population size?
Given the definition
\[ X_{t+1} = Z_t f(X_t) \]
and the condition that we have to have
\[ \operatorname{E}(Z_t f(X_t) ) = f(X_t) \]
that is, $ Z_t $ has to be mean unity, then some shocks must result in
$ X_{t+1} f(X_t) $
for any non-trivial \(Z_t\). So how do we enforce that these decreases do not violate the self-sustaining principle?
It would seem to require at least that \(Z_t\) is a function of \(X\) as well?
Reed seems to imply that this is a much more trivial requirement, such as stating only that \(f(x)\) is compensating density dependence (such as Beverton Holt), and not overcompensating (such as the discrete logistic).
- Write a flat tex outline for policy costs (policy costs)
- Compare the probability of detection in managed and unmanaged models. (resilience thinking)
- Confirm Reed S==D theorem, evaluate in Sethi context (value of information) See notes below.
-
Run scenarios with very low noise, instead of deterministic (value of information), running for
uniform_logistic
now - Re-run bias table with measurement noise (value of information). Done, compare against history.
Other
- PRSB review
- Send Hilmar the ropensci slides
treebase revisions
- Conference possibility?
Substantially updated octokit plugin, see labnotebook issue #11.