Do sweep of initial cooperator proportion about nicheconstruct HOT 12 CLOSED

An update. Overall, the trend seems to show that cooperators increase in proportion.

So far, no instances of cooperators arising de novo (from p0=0) and spreading, but if there are enough replicates, it may happen...

from nicheconstruct.

One difficulty is that many populations with p0 < 0.5 start out with 0 cooperators. We may need to pay special consideration to these in the analysis.

from nicheconstruct.

Would it be worth adding an option to the code to add a single (isolated) cooperator to an otherwise all-defector population at the beginning of the simulation? This would allow us to directly address the suggestion that we examine results when p0=1/800.

Or, is the fact that we have results for many simulations in which p0=0, so cooperators must arise de novo? More on those results later. Brief summary: Cooperators do not emerge and reach majority in any of the replicates that have been done so far. For this to happen, I think cooperators must arise VERY early on so that they have the ability to migrate to empty subpopulations.

from nicheconstruct.

Dear Brian,

Have these been run with the standard number of subpopulations (or the
increased number)? I wonder if this makes any difference...

Best,

Ben.

On Sun, Aug 23, 2015 at 12:43 PM, Brian Connelly [email protected]
wrote:

Would it be worth adding an option to the code to add a single (isolated)
cooperator to an otherwise all-defector population at the beginning of the
simulation? This would allow us to directly address the suggestion that we
examine results when p0=1/800.

Or, is the fact that we have results for many simulations in which p0=0,
so cooperators must arise de novo? More on those results later. Brief
summary: Cooperators do not emerge and reach majority in any of the
replicates that have been done so far. For this to happen, I think
cooperators must arise VERY early on so that they have the ability to
migrate to empty subpopulations.

—
Reply to this email directly or view it on GitHub
#80 (comment)
.

from nicheconstruct.

Baseline, but I could get some runs going with the 10x smaller subpopulations setup done for the other paper.

from nicheconstruct.

Might be worth checking out...

On Sun, Aug 23, 2015 at 2:17 PM, Brian Connelly [email protected]
wrote:

Baseline, but I could get some runs going with the 10x smaller
subpopulations setup done for the other paper.

—
Reply to this email directly or view it on GitHub
#80 (comment)
.

from nicheconstruct.

Simulations submitted.

from nicheconstruct.

That was quick!

On Sun, Aug 23, 2015 at 2:43 PM, Brian Connelly [email protected]
wrote:

Simulations submitted.

—
Reply to this email directly or view it on GitHub
#80 (comment)
.

from nicheconstruct.

If you have time (ha ha!) I like this idea "Would it be worth adding an option to the code to add a single (isolated) cooperator to an otherwise all-defector population at the beginning of the simulation?"

from nicheconstruct.

Some nice results from simulations where the initial proportion of cooperators is 0 (and subpopulations are 10x smaller, while the population is ~10x larger):

Although these simulations were fairly short (each replicate takes about 43 hours to run), we do see cooperators emerging and increasing in abundance. This appears to happen more often here than without niche construction.

from nicheconstruct.

This seems like this could be worth including in the Results section. We should discuss which parameters would be best:

• stay with the small populations?
• scale up to 3000 updates like the other simulations?

from nicheconstruct.

Scaling up to 3000 updates seems like it would be good idea but if each replicate takes about 43 hours to run, I'm not sure it is doable...

from nicheconstruct.