Are we happy with the multiple uncertainty calculation as it stands? #1

cboettig opened this Issue Mar 12, 2013 · 2 comments


None yet
1 participant

cboettig commented Mar 12, 2013

From Nov 22

uniform noise policy functions

log-normal noise policy functions

  • It is surprising that log-normal deviates so much from uniform noise policy functions.
  • It is troublesome that the uniform case does not look smoother. Should match Sethi Figure 3 exactly. Qualitative match is reasonable. Higher grid resolution does not seem to resolve this.
  • Negative escapement seems like it would make no sense. However, tis results because the policy is chooses harvest, not escapement. It is actually rational to set a target harvest larger than the estimated stock size when you know that estimate may have error. If there is no cost to fishing effort, that can be set arbitrarily high.

Algorithm discussed Nov 1. See current algo implementation for precise details.


cboettig commented Mar 12, 2013

  • What further tests could be run to first verify these results?
  • How can we explain the differences between the two noise scenarios, and also the non-monotonicity in log-normal noise?

cboettig commented Mar 27, 2013

Jim requested seeing a Matlab implementation to compare against. Implementation developed in #4, see

the full matlab code here:

and here is a matlab script that runs that function on an example test case and plots the resulting policy function:

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment