21 July 2003

I think I've fixed the VHFC's problem with neutral vowels. See the results here. I also figured out that if you run the simulation as "./bug -s" it varies the seed and fixes the randomness problem Mark brought up.

Daily total hours: 6

______________________________________________________

22 July 2003

Today I've been varying population density in 5% increments. All other parameters remaining constant, I think I've decided that population sizes below 0.3 have a much harder time getting rid of harmony. I think this is because the low levels of language interaction due to sparseness of neighbours prohibits homophony from getting a hold on the population lexicon. The denser the population, however, the smoother the curves. Other parameters I could vary are the threshold to zero harmony and the frequency threshold. I'd like to keep death and reproduction constant.

Daily total hours: 6

_______________________________________________________

23 July 2003

This morning there was another Emergence meeting; this time David Berger talked about Wolfram's "New Kind of Science." The discussion centered on how randomness is indistinguishable from pseudorandomness and since the universe is deterministic, all events are, in effect, pseudorandom. This supports the argument for synthetic ethology in implying that complex processes cannot be described mathematically, but only by simulation. Emergence is deterministic, but though there may be rules to explain a behaviour, it is not guarranteed that we can find them. I intuitively agree with most of Wolfram's assumptions, but would like to see evidence beyond cellular automata.

Today my goal is to write about why the simulation is to be trusted. It's going to be a lot about the nature of language change and its dependence on local interactions. I found a great article by none other than Luc Steels called "The Puzzle of Language Change." I wish I could find the paper I wrote for Mark on Language Evolution.

Daily total hours: 6.5

_______________________________________________________

24 July 2003

Today I'm writing David's Swarm primer. When done, it can be found here. Having looked at the parameters, it's clear I need to do a little cleanup to sort out the obsolete parameters and add a few more of my own, like probMisspeak, which is currently 0.05, but could fluctuate between 0.01 and 0.1.

I'm having a lot of trouble updating the defaults in the probe map. In fact, I have no idea how they got the numbers they do, considering they are initialised with completely different numbers in ModelSwarm.m.

The biggest problem with the curve now is the lack of dependability on the downward S. Sometimes it looks more like a W. I do consider some of these good, though. In some runs, harmony bottoms out at a level much higher than H(0). This gives me some confidence in the argument for Karaim and Krymchak, with VHFC harmony levels in the 70's.

Daily total hours: 5.5

________________________________________________________

25 July 2003

This morning I taught myself LaTeX and started writing up all my work. I've been outlining the important sections and writing about the intricacies of harmony and homophony. I also decided that the harmony simulation is good enough to write an article about, especially with '-s' to re-seed the random methods. I'd really like to find some way to take the average of all the curves. I have a feeling it'd be smoother. Ten runs at identical parameters would be enough, I think. It would definitely be easier than trying to tell from individual graphs what the best settings are.

Daily total hours: 6

Weekly total hours: 30