gorithm

my very educated mother just served us noodles

I'm a little sad to see Pluto demoted as a planet.  The key question is how we update:

my very educated mother just served us nine pickles

The Economist, always on the forefront of important issues, suggests:

my very educated mother just served us noodles

Kottke had a contest for a planet mnemonics, with this excellent winner (using nine planets as a protest):

My! Very educated morons just screwed up numerous planetariums.

There really is cold fusion

I was talking with a coworker at lunch today, and he said "wouldn't it have been great if Pons and Fleishman had got cold fusion to work."  Interestingly, Nature published a paper this week on the first unequivocal cold fusion. 

Now Putterman, a physicist at the University of California, Los Angeles, has turned a tiny crystal into a particle accelerator. When its electric field is focused by a tungsten needle, it fires deuterium ions into a target so fast that the colliding nuclei fuse to create a stream of neutrons.

Putterman is not claiming to have created a source of virtually unlimited energy, because the reaction isn't self-sustaining. But until now, achieving any kind of fusion in the lab has required bulky accelerators with large electricity supplies. Replacing that with a small crystal is revolutionary. "The amazing thing is that the crystal can be used as an accelerator without plugging it in to a power station," says Putterman.

It won't generate electricity, but it is real.  Now if only Pons and Fleishman had sent their paper to Nature for peer review, we would have all had less of a mess.

George Dantzig, inventor of the simplex algorithm, dies at 90

George Dantzig died this week at 90.  He was the inventor of the simplex algorithm, which made it possible to solve linear programming problems.  The Operations Research society, INFORMS, has an obituary. To a large extent, he made my field of OR possible, since it was the first time real interesting problems could be solved with a computer.  This may seem odd to non-techies, but I use the simplex algorithm every day to solve problems.  He was a giant in our field.

I hope that when I die, I rate a Slashdot posting.
He even has his own urban legend.

Real tiny hobbits

This is just way cool. A species of little humans, that are recent enought to a part of the legends of the area. Published in Nature, so you know it is important.

It sounds too incredible to be true, but this is not a hoax. A species of tiny human has been discovered, which lived on the remote Indonesian island of Flores just 18,000 years ago.

Researchers have so far unearthed remains from eight individuals who were just one metre tall, with grapefruit-sized skulls. These astonishing little people, nicknamed 'hobbits', made tools, hunted tiny elephants and lived at the same time as modern humans who were colonizing the area.

Check out the free special section on the Nature website.

Linear programming and the Kepler conjecture

The Kepler conjecture is about the best way to pack spheres tightly together. Kepler thought that a cubic close packing, the way oranges get stacked at the grocery store, was best. However, it wasn't until 1997 that this was proved for the 3 dimensional case. (As an interesting aside, for the first time ever Annals of Mathematics gave up trying to review his paper... it was just too hard to verify since it relied on tons of computer calculations.)

The problem also can be solved in higher dimensions as well. Henry Cohn and Noam Elkies have published a new paper that uses linear programming to improve the upper bounds in dimensions 4 through 36. They conjecture that their techniques solves the problem (i.e. the bounds meet) for dimensions 8 and 24.

The reason I bring this up is that linear programming is a technique that is near and dear to my heart, and it is cool to see new uses for it.

(Via a facinating article in the print edition of Nature by Ian Stewart, a math professor at Univ. of Warwick who writes lots of popular math books.)

New random number generator flaw

This is a geeky post. You've been warned.

Nature Science Update points to a new technical report that explains how a commonly used type of random number generator fails.

Like many good papers, it is based on a problem in Knuth's Art of Computer Programming. In this case it is exercise 3.3.2.31.

We measure the probablity that a run of length w (w odd) contains more "heads" than "tails" when produced by a recurrence relation in Z2. This probablity is almost never close to 1/2!

In the paper they explain why generators that generate random bits of 0 or 1 are fundamentally flawed. (This is oppossed to PNG's that generate 32 random bits at a time, etc.) They forcus on linear feedback shift register (LFSR) generators. It is pretty interesting reading. (Does that make me a geek? I read CS tech reports for fun, and I admit it.) Bear this in mind the next time you need a good pseudo-random number generator.

I write like an engineer

A recently published algorithm can determine if a book was written by a man or woman, with 80% accuracy. Roughly speaking, women use more pronouns, and men like articles and adjectives.

Unfortunately for all of you, I tend to write like a male engineer. I'd guess that would make it even easier for the algorithm.