Logistic recruitment, uniform noise
(Sethi Fig 3 configuration)
Logistic recruitment, lognormal noise
Beverton-Holt recruitement, uniform noise
Beverton-Holt recruitement, lognormal noise
Logistic recruitment, uniform noise, small r
### Logistic recruitment, lognormal noise, small r
Metadata
From scenarios_meta.txt
Columns:
y_grid: The observed stock size
escapement: The optimal escapement policy. (y_grid value minus the escapement is the quota).
sigma_g: The growth noise scaling factor
sigma_m: The measurement error scaling factor (between observed stock x and measured stock y)
sigma_i: The implementation noise scaling factor (between quota q set and harvest h realized).
r: The growth rate parameter of the recuitment function
K: The carrying capacity parameter of the recruitment function
recruitment:
1: Logistic, x+r*x.*(1-x/K)
2: Ricker, (1+r)*x.*exp(-(log(1+r)/K)*x)
3: Beverton-Holt, (1+r)*x./(1+r/K*x)
noise:
1: uniform noise distribution
2: lognormal noise distribution
id: a unique id number for the scenario (for subsetting and plotting)
Constants:
delta:
value: 0.05
description: Discount rate
Tmax:
value: 10
description: number of years (iterations) over which policy is optimized
x_grid:
value: linspace(0,200,201)
description: True stock size
y_grid:
value: linspace(0,200,201)
description: Observed stock size
h_grid:
value: linspace(0,120,121)
description: Implemented harvest
q_grid:
value: linspace(0,120,121)
description: Target harvest quota
Notes:
See the scenarios.m file to confirm values of constants used.Source code / version archives
Data for all the plots shown appears in scenarios.csv. Because my collaborators have me using Matlab for this and Matlab can neither (afaik) write headers to csv files, append to csv files, or even conviently export non-numeric and numeric data (I write a matrix to the csv file), you’ll just have to consult the scenarios_meta.txt for relevant information.
All results shown here correspond to files that can be found in Commit 4c71612, including the data file. The SVG plots are hosted through the gh-pages branch of the repo, and thus show the most recent versions of those plots. The names have been tagged with the commit hash to avoid accidentally overwriting when programatically generating images. (Clearly that’s not ideal, but cannot link directly to version and have it render).
Results generated by running scenarios.m
Functions
- The main routine is (still) in multiple_uncertainty.m
- The scenario template over which we loop is defined as a seperate function in scenario.m
- scenarios.m also calls the script plot_scenarios.m, which can be run to generate the plots from scenarios.csv directly without first regenerating the results data.