Wednesday, May 9, 2012

Wednesday: I kind of felt like I was getting a cold today

So I took some preventative measures.  I went with curry katsu from the sushi place for lunch:
 It was a good choice, and comes in a much better arrangement than they do it at the plate lunch place.  The curry sauce comes in it's own separate container ("Careful! Hot!" said Sushi Lady), so it doesn't make the rice soggy.  It also allows you to curry it up however you want.  I went with little bits of curry, eat a bit, add more curry, eat more, etc.  There are no chunks in the curry, which is a bit of a shame, because I like bits of carrot and beef.  Still, very tasty, with just enough spice to ensure a sniffle in your nose.
My phone just refuses to align correctly for overhead shots.
Dinner was leftover turkey soup from...um...Thanksgiving.  "Whoa, that's like six months ago!" Sure, but I made it, bagged it, and it went directly into the freezer.  I don't know how long frozen soup lasts, but it's at least this long.  Still tasty after all this time, and my trick of defrosting in the microwave, then adding a handful or two of egg noodles, and then letting them cook in the soup in the microwave works wonderfully.

Remember last month when we had fun with population models?  I wanted to do more of that today, based on this story in the NYTimes.  The hypothesis was going to be that the increase in support for same sex marriage was a result of the changing demographics due to old people dying.  I used this set of death rates (assuming they were constant in time), and selected the matching birth rate from here (again, assuming the 2005 numbers were constant).  This immediately revealed itself as a bad set of assumptions, as you end up with functions that have two regimes:
  1. For t = 0 through t ~ max_lifetime, the system is working towards to normalizing the population cohorts based on the death rates.  The birth rate keeps the system fed, ensuring that there are the right number at the population input.
  2. After t ~ max_lifetime, the population cohorts become stable, as they contain the steady-state fraction of the population. Total population increases exponentially (14.6 / 1000 > 825.9 / 100000).
However, that isn't really the question I wanted to answer, so it doesn't matter.  Actually, given the assumption of near-steady state, the answer is obvious.  The question I really wanted to answer is "what fraction of the total population was born after some date, set such that people born after that date don't really care about same-sex relations."  If I had proper rates for all time, I could probably tie this date down, but I suspect it's something like "1970ish, or maybe like sometime in the 60's."  In any case, knowing that in the steady state you introduce some new fraction F of the population each year as new births, after N years, people born after that date comprise something like N * F of the total population.  Therefore, the fraction is roughly linear in time.  Go back, look at plot, squint.  Yeah, I think I can buy that as an explanation for that plot.



No comments:

Post a Comment