Thursday, November 8, 2007

Day-By-Day and Yon...

Check out today's Day-by-Day cartoon: John Muir (the cartoonist) comments on the moving photograph by Michael Yon, showing Iraqis of several faiths working together to put a cross on top of a rebuilt Christian church.

If you're reading this on some day other than November 8, 2007, then click on the link above to see what I'm talking about.

x86 Math Bug...

The implementation of sin() and cos() on the Intel x86 platform has an interesting bug:
For many years, the JDK on x86 platforms has used the hardware fsin/fcos x87 instructions in the range [-pi/4, pi/4], a range which encompasses about half of all representable floating-point values. Therefore, in that range the performance of the JDK's transcendental functions should be nearly the same as the performance of the transcendental functions in C, C++, etc. that are using those same fsin/fcos instructions. Benchmarks which focus on testing the trig performance of large values, such as almabench, present a skewed portrait of Java's trigonometric performance. The next question is why don't we just use fsin/fcos over the entire floating-point range? The simple answer is that fsin/fcos can deliver answers that are arbitrarily wrong for the most straightforward way of measuring error in the result.
I'd never heard about this until I read this post on James Gosling's blog. On the native, uncorrected platform, the error can amount to 5 or 6 significant digits – most likely even producing results with the wrong sign!

Java programmers need not fear – the JVM fixes the problem (read the post to see how), but the fix is computationally intensive and reduces performance for angles outside the range of -pi/4 to pi/4 (in radians). Slow, but accurate (and predictable), as opposed to fast, but randomly inaccurate...

Jamulian

From the Union-Tribune:
SAN DIEGO – Authorities released the name Wednesday of an 18-year-old college student killed when the car he was driving crashed off an East County freeway at high speed, possibly as the youth was racing another motorist.

Devan Caddell of Jamul lost control of the 2001 Acura Integra after cutting in front of a white Ford Mustang on State Route 94 in La Mesa shortly after noon Tuesday, according to the California Highway Patrol and county Medical Examiner's Office.

The westbound Acura, which was traveling at 100 mph or faster, skidded off the right-hand side of the roadway near 69th Street, then crashed through a chain-link fence and smashed into a large tree behind a home, CHP public affairs Officer Brian Pennings said.

The impact uprooted the tree, tore the sedan apart and sent Caddell, a student at Mesa College, tumbling into the residential yard. He came to rest near a children's swing set and died at the scene.

The driver of the Mustang – who, according to witnesses, appeared to have been racing Caddell just prior to the crash – did not stop and has not been identified, Pennings said.

I don't know anything else about this senseless death. If any of you know more, please leave comments to let the rest of us know...

Less Profit...

Lakes Entertainment, the investor behind the Jamul Indian Casino project, just issued their quarterly report. It contains these statements related to the casino:

Net unrealized gains and (losses) on notes receivable were ($0.6) million and $5.8 million for the third quarters of 2007 and 2006, respectively. The net unrealized losses in the third quarter of 2007 related to unrealized gains on Lakes notes receivable from the Shingle Springs Band of Miwok Indians (Shingle Springs Tribe) in the amount of $1.2 million and unrealized losses of $1.8 million on the Lakes notes receivable from the Jamul Indian Village (Jamul Tribe). These notes receivable are adjusted to estimated fair value based upon the current status of the related tribal casino projects. The increase in fair value of the notes receivable from the Shingle Springs Tribe relates primarily to continued progress on the construction of this project, which is currently within budget and on schedule. The decrease in fair value of the notes receivable from the Jamul Tribe relates primarily to an increase in the discount rate which resulted from a decrease in current estimated win per unit for this project.

Net unrealized gains of $5.8 million during the third quarter of 2006 included $2.6 million related to the adjustment to fair value of notes receivable from the Shingle Springs Tribe due to favorable events occurring during the third quarter of 2006. The remainder of the net unrealized gains during the three months ended October 1, 2006, related primarily to increases in fair value of notes receivable from the Jamul Tribe and the Pokagon Band.

Summarized in English: the casino project isn't going to be as profitable as they originally thought. Here's hoping that trend continues, to the point where everybody just loses interest...