Science Fare

“Compared to the relative simplicity of the Polynesian sequences, the Welsh results were all over the place. There was no sign of a clear-cut distinction in Wales analogous to what we saw in Polynesia where two separate clusters were so clearly the result of a mixture of people from very different origins. It looked as if we had a small number of little clusters which were all quite closely related to each other, rather than two big clusters separated by a large number of mutations.”
“The Seven Daughters of Eve” by Bryan Sykes, page 123

Playing on the beach!

Overheard regarding the Gizzard Shad : “This is the only species of fish that has a gizzard, and so it must be the only fish decended from birds.” This is, of course, untrue. The Gizzard Shad’s gizzard is an adaptation of the normal fish stomach to function like a bird gizzard. This adaptation is an example of convergent evolution, in which multiple instances of a similar functionality evolves separately. Eyeballs with lenses on the edge of clamshells is another example.

I did an amusing back-of-the-envelope calculation on the drive home tonight. I didn’t actually have an envelope….
The days are getting longer! But how much longer does the day get each day? Let’s suppose that the periodicity of the day length is described by a sine wave. And lets suppose that where we live the shortest day in winter has eight hours of sunlight and the longest day in summer has 16 hours. Also, to keep things simple, lets say day 0 is an equinox, for example the upcoming Spring Equinox on March 21 (or is it March 20? it varies from year to year). Hmm, we’ll need some variables sprinkled in there — how about y for the length of the day and x for the day of the year. Therefore our assumption can be written this way:

y = 12 + 4 sin( 2 π / 365 )
Let’s invoke some good old calculus. The rate of change of day length is the first derivative of y with respect to x. Trust me, this works out to
dy/dx = 8 π cos( 2 π / 365 ) / 365
Exactly at the equinox, the cosine expression goes away (becomes 1), thus the rate of change of day length is
dy/dx( 0 ) = 8 π / 365 = 0.0689 hours per day = 4.131 minutes per day
So at the equinox, I’d estimate that the day gets about 4 minutes longer each day. That comes out to about 28.9 minutes per week.
So how close is my back-of-the-envelope appoximation? I used a military sunrise/sunset calculator to find out what the actual sunrise and sunset times are for Minneapolis. First I checked my assumptions : The shortest day of the year here (Dec 21) is 8:46, and the longest day is 15:37. My guess of 8:00 and 16:00 is reasonably close, but if you redo the calculation above with those numbers you arrive at 3.53 minutes per day instead of 4. Next, let’s look at how long the days are near the equinox:

date	sunrise	sunset	length of day (minutes)
March 19	0618	1825	727
March 20	0616	1826	730
March 21	0614	1827	733
March 22	0612	1828	736
March 23	0610	1830	740

The data from the navy indicates that the days are getting longer by just over three minutes each day. That is still a little bit lower than expected. I suppose maybe the length of the day is not exactly a sine wave, what with the earth being in an eliptical, not circular orbit and all…

Today I saw a manatee. Actually I saw over twenty manatees. They don’t like to be in the Gulf of Mexico when it is too cold (below 65° F). The gulf was 66 degrees today. I saw these manatees basking in the warm effluent from the Fort Meyers power plant cooling towers. The water from the towers is a balmy 81&deg F — just right for manatees.

hi from sanibel!
“We’re thinking of all you poor sucka’s back home,” quoth Tismo. He got crabs.

For all those Dungeons and Dragons fans who also love Macintosh Computers, I feel it is my duty to warn you about zombie processes that are running rampant through your operating system! Why, you ask, why do they exist? Their only purpose is to tell their parent process that they have died! Thus, the zombie process cannot be kill(1)ed, not even kill -9ed!!! The only way to get rid of a zombie is – gasp! – to kill(1) its parent! Does that seem harsh? — well, if the parent process would just wait(2) once in a while, it would pick up on those not-so-subtle SIGNALs (SIGCHLD — roughly translated from computerese means “Hey, ma, your kid is dead”). But no, our zombie’s parent is too busy to wait(2)… we’ll have no choice but to kill the unattentive parent. Then our zombie will be adopted by init, the parent of all lost processes. Init is always wait(2)ing to hear of the demise of its children (init is kind of a worry wart, I guess) So finally our zombie will be at rest.
The moral of the story : Zombies can’t be killed. They just want love, and if you stop and listen they will be at rest. If you see a zombie, and its parent isn’t listening to it, you can just go ahead and kill the parent.
I have to admit, sometimes your operating system will be overrun by zombies. Then you can’t start any more processes, not even kill(1). Basically at this point you are screwed. You can try sacrificing some healthy processes to free up a few PIDs, but ultimately you may just have to reboot.