1. Biological background
-
Annual plants germinate, grow, reproduce, and then die, all in one year. All of their fitness is tied up in how much they reproduce in that year. This fact makes them easier to model than, say, perennials. Everything we talk about here more or less only applies to annual plants. Actually, more restrictive than that: annual plants that reproduce by seed, rather than by vegetative reproduction.
-
Think of the plant as doing only two things: reproducing (i.e., building flowers, fruits, and other reproductive structures) and investing in reproduction by growing vegetative structures (i.e., leaves, roots, stems, and everything else that doesn't contribute directly to reproduction but that provides resources which are themselves available to reproduction).
-
On every day of the growing season, a plant turns some light into carbohydrates, which we'll call photosynthate. How does it allocate this photosynthate between immediate reproduction and vegetative growth?
2. Mathematical background
-
perfect info : perfect plasticity among perfectly specialist schedules
-
zero info : perfect fixity at perfectly bet-hedging schedule
-
but what about intermediate info?
3. Interestingness background
4. What I did
-
GA to search schedule space for schedules that maximize long-term fitness for a given probability distribution of season lengths
-
schedules represented as strings of allocation fractions
-
mutations: point, crossover
-
fitness calculated as geometric-mean fecundity with weights given by the probability distribution
5. Simulation vs search
-
perils of self-education: at first I did it wrong
-
results same but they took longer and were much noisier
-
GA != simulation of evolution
6. Randomness, predictability, and information
-
When is randomness the same thing as unpredictability?
-
Is narrowness of the probability distribution a valid way to represent cue reliability?
7. Complexity of plant perception
Last and final step in this project: develop "response assembly" model of the complexity of perception. (Don't scoop me!)
