search

davidrpugh / population-ecology-approach

0 stars 2 forks source link

Parameter sweeps #15

Closed davidrpugh closed 9 years ago

davidrpugh commented 9 years ago

Discussion thread for parameter sweeps. I have started a new code branch add-parameter-sweep to work on adding this functionality to the code.

@markeschaffer and @PaulSeabright, based on my understanding of our recent conversation:

  1. dA and da control the "informativeness" of the female signal for "altruistic" and "selfish" type females, respectively. Thus, dA=0 (da=0) indicates that the female's signal is completely uninformative about her true type. Similarly dA=1 (da=1) indicates that the female's signal is perfectly informative about her true type.
  2. eA and ea represent the probability of successful screening by males preferring "altruistic" or "selfish" females, respectively. Since any male can always screen females by "flipping a coin", both eA and ea should be bounded below by 0.5.
  3. We should restrict our attention to initial conditions with "pure sub-populations" (i.e., initial conditions of the form S_A = mGA = fGA and S_a = mga = fga = 1 - S_A and mGa = maG = fGa = faG = 0).

Do you agree?

markeschaffer commented 9 years ago

Not sure if it belongs here, but I was wrong and Paul was right about the interpretation of d.

d=0.5 is random signalling, and female signals have no information content. d=1 is perfect signalling. d<0.5 is ruled out in the paper, but maybe this is a mistake. That is, d<0.5 can be interpreted as mimicking, and d=0 is a perfect mimic. If a females can perfectly mimic A females, then da=0, etc. So d lies in [0, 1].

And e also lies in [0, 1]. e=1 means males can correctly identify unsuitable females 100% of the time. e=0 means males cannot exclude any unsuitable females from the adoption pool.

davidrpugh commented 9 years ago

@markeschaffer

Great thanks for clarifying the interpretation of the parameters. I have written code to sweep all four parameters and the initial condition. Code is slow (i.e., 5 nested for loops), but there is much that I can do to speed it up. I will post some example plots tomorrow for discussion.

davidrpugh commented 9 years ago

I am closing this issue and moving the discussiong thread to issues #18, #19, #20, #21 for screening parameter sweeps; and issues #23, #24, #25, #26, #27 for signaling sweeps.