Wednesday, September 29, 2004

The New MacArthur Geniuses are in.

I was intrigued when I first saw this new announcement of the MacArthur fellows winners. So I did some digging and found this article written in 2000 in Slate: The MacArthur Geniuses - How to become one of them. By David Plotz. What is fascinating is that one can see how the winner's prototype description back in 2000 follows nearly word by word the description of some of the 2004 fellows. Tunnel vision anyone ?

Watching a volcano erupt

With the new volcano advisory, one can now watch the Mount St. Helens Volcano erupts from your own living room. more to go

The first attempt by SpaceShipOne was successful. One more to go to win the Ansari X-prize. Our friends at Ecliptic provided a video of the event. Virgin Galactic will provide rides on this bird later at the tune of $200,000. This is about 4 minutes of zero-g for $200 K and the disctintion that you can now be called an astronaut. For about a hundred times less you can spend comparable amount of time in zero-g in a parabolic flight in a plane but no astronaut title.

16 years, 16 days, what's two orders of magnitude among friends

It looks like the most precisely expected earthquake is 16 years late. At least we know what to expect of this prediction.

MREMS: Micro-Radioisotopic-Electro-Mechanical Systems

This feature article of IEEE spectrum does a good job mentionning the issue of radioactive microbatteries that could power future MEMS.

My own preference is for Polonium-210 which has the ability to release about 120 W per grams. This is huge. Unfortunately, as said in the article:

At the diminutive dimensions of MEMS devices, the ratio between an object's surface and its volume gets very high. This relatively large surface makes it difficult to sufficiently reduce heat losses and maintain the temperatures necessary for RTGs to work. So we had to find other ways of converting nuclear into electric energy.

For a sphere, surface area decreases as the square of the radius whereas its volumes decreases as the cube of the radius. Hence, as the size decreases, the surface area become more important for the volume: i.e. major heat loss is expected from this sphere. Realizing this, Tom Blanchard from Wisconsin-Madison decided to look at this through the building a micro heat barrier. Hopefully, we will be able to keep high temperatures in that sphere so that thermoelectric generators can be used. It is also wrong to think that NASA has been looking at only Plutonium as an RTG sources. Back in the 1960's Strontium was looked into as a source of power for autonmous buoy at sea. Even polonium was looked at but the mission scenario requiring this type of sources with a half life of 138 days were non-existent. I recall the preliminary design needing a lot of shielding and being heavy. But then again, if the only issue is about having polonium not surrounded by a thermally conductive material, what about using aerogel instead ?

Monday, September 27, 2004

Airport Terminals Collapse

There is an eerie feeling that the Dubai airport terminal collapse looks too familiar to the Paris Terminal collapse. In fact they both have the same architect whose recent project include the following terminals:

Dubaï, Terminal 3

Charles de Gaulle, Terminal 2 E,

The commission that looked into the first collapse in Paris seemed to say that the sudden collapse is linked to the perforation of the concrete roof by the struts (supporting it). The science of concrete is a very difficult one, especially when it comes to computing its performance in an inventive type of architecture. Hopefully these accidents will be studied very carefully so we can have an answer like we did for the sleipner platform collapse which also included concrete walls.

On a related note, I think I understand how people get to be chosen in order to bid on contracts like these and I can see how a certain amount of ego goes into the decision making process. But, I do not understand how little emphasis there is on the functionality of these airports. When I fly out of Charles de Gaulle, I certainly do not expect the nice design of the buildings to give me a more interesting flight experience. I would rather that they put an emphasis on, the following:
  • Is security good ?
  • How cheap is the airfare ?
  • Can I get for free on one of these airlines clubs ?
  • Do I have an internet connection in these clubs ?
  • Can I board first ?
When I land in Paris, I look at the following:
  • Can I get out of the plane first ?
  • Can I get to customs without having to ask people where to go ?
  • Can I get through customs without having to wait for an hour (because they are all at lunch) ?
  • Can I get my luggage fast ?
  • Can I find some ATM machine fast ?
  • Can I quickly find a Taxi or a shuttle bus ?

Most of these items do not need the airport buildings to look good. Most airports including CDG fail at providing a decent solution to these expectations.

Sunday, September 26, 2004

Perfect Storms

One can only wonder if the collision of several seemingly unrelated facts is the precursor to a perfect weather/financial storm. Take for example, the oil price going up because of the, now, fourth hurricane hitting the Gulf of Mexico and Florida and the increase in foreclosures in Houston because of a hot summer (actually not the hottest one in the past five years.) Knowing that the weather is getting drier/warmer in Texas when there is a hurricance in Louisiana/Mississippi like it happened two weeks ago with Frances.....

Friday, September 24, 2004

Local Weather Prediction for the People II

In today's previous entry I mentionned that there seems to be small markets that are pretty much untouched for weather predictions since the brunt of previous and current efforts have been focused on very hard to solve problems such as predicting Jeanne's tortuous path or figuring out why in Europe, August 10-12th is a period of strange weather patterns like in 2002 or 2003 or trying to figure out if one should evacuate New Orleans or Miami. They have such tremendous economic value that the trade-off of investing in tools to predict them is seen as worthy by governments. But aside from these very weird events, minimal variations in the weather patterns over short distances (less than the coarse grid designed for supercomputer input) are now becoming important enough that it affects events with attendant economic value. I will try to mention these in a future entry.

Nuit Blanche Paris 2004

Folks, there is nothing here to see on the subject on this blog. Please go to this "hidden" Nuit Blanche web site instead. It will take place on October 2nd, 2004. Have Fun!

Non, ceci n'est pas le site de la Nuit Blanche de Paris, veuillez plutot aller vers ce site "cache" (il n'y a pas de lien direct avec address http depuis ) de la Nuit Blanche. Cela le se passe le 2 Octobre 2004. Amusez-vous bien!

Local Weather Prediction for the People

There is a segment of the population that needs to have accurate weather prediction on a specific day or have access to specific historical data and they are willing to pay for it. Case in point, the prediction of wind patterns for wind turbines. In France, one is beginning to see micro-local weather prediction services such as Meteoconsult or Meteopartner that are dedicated to serve micro-markets such as bread makers who need humidity data to see if they should prepare their flour on a specific day. The interesting part of this story is that they can do this because most of the supercomputer time spent on devising the weather for the world is spent here in the U.S. and freely available on the net. Meteo France is screaming that it is unfair competition that a small companies like Meteopartner can have acccess to data like this because they did not really pay for it unlike them. The other interesting part of the story is that the niche market found by these smaller outfits is precisely the type of customer that could not be a customer for large outifts like meteo france. There is a need to dedicate special numerical skills to interpolate weather data from the current and too coarse data coming out of supercomputers (20 kms by 20 kms grids.) If the smaller outfits are capable of finding enough customers, they are very likely to be the big dogs in the future because this is clearly a scenario described many times of Disruptive Technologies. Maybe learning how to solve the Navier-Stokes equation was not such a bad skill after all....

Wednesday, September 22, 2004

If ALICE is not even a teen-ager, how do you expect a robot to drive 150 miles alone in the desert?

According to slashdot, ALICE Wins Loebner Prize 2004. It actually won the Bronze medal only. If Artificial Intelligence is at this stage of the game, how come people expect to win the DARPA Grand Challenge race ? Maybe it's because they don't think it is Artificial Intelligence in the first place ?

Tuesday, September 21, 2004

Weapons plutonium disposition will take place

It looks like after years of thinking about it, part of the U.S. Weapons plutonium stockpile will be initially processed in France. This is good news because it reduces the danger of proliferation of this material in the future. What annoys me in these types of articles is the assymmetry of experts and stories. The anti-nukes side is always overepresented because the pro-nukes are generally the people who are working in that field and they generally work for the government. For instance, in this article, when Tom Clements says that

It is the height of arrogance to carry out a shipment like this while demanding that other nations refrain from proliferating nuclear weapons’ materials and technologies

he basically shows a lack of knowledge of what proliferation means. When you ship special nuclear materials (SNM) from a nuclear weapons state to another nuclear weapons state, you do not proliferate. But then again what would you expect from a spokesperson from Greenpeace international. When Edwin Lyman, from the Union of Concerned Scientists in Washington DC states that
the undertaking is ill-conceived, poorly executed and inadequately secured.

I would like to know if he knows something that everybody else does not. How ill-conceived is a plan that is the final solution to a process that has lasted at least eight years and lived through two different administrations of opposite end on the political spectrum. When he says
It would be safer to choose the “plutonium immobilisation” option

Mr. Lyman does not tell the other side of the story: i.e. that the reason this solution to plutonium disposition is chosen is because it is the mirror process that involves the Russians. See, the Russians have double the amount of weapons plutonium and will not ever think of plutonium as trash. They put a lot of lives/money/effort in producing it, it is a national treasure to them. The response from guys like Mr. Lyman looks like the lessons given by the super rich to the poor, some sort of a nuclear equivalent to "let them eat cake." The Russians always told the U.S. that immobilization amounted to storage. Storage means that plutonium can be retrieved "easily." This situation is unacceptable to both parties eventually because it would mean that it is not disarmament. The situation is even hilarious if it were not for this serious subject. When Mr. Lyman talks about the immobilization option by mixing
manageable plutonium with highly radioactive waste – making it hugely dangerous to anyone attempting to misuse it – before encasing it in glass and disposing of it.

It is the type of thinking of another age; before 9/11 to be exact. Before 9/11, nobody thought that people would hijack a plane and ram it into buildings. The same goes for this "immobilization disposition" solution: to think that nobody would get this glass casing, blow it up, pick up the pieces and try to assemble a weapon out of it is a sure sign that the Union of Concerned Scientists has not evolved since 9/11. The tragedies of 9/11 or Beslan tells us that you will find people who will do this and die doing it (because of radiation exposure) but will eventually get that job done. Matt Bunn, whom I respect, underestimates the French when it comes to nuclear issues. France has 50 nuclear power plants and a mature fuel recycling program, the U.S. has 100 nuclear power plants and stores all their waste at every plant, who is more vulnerable, Matt?

We don't care about sleeping like babies no more

In a previous entry, I was making a mention of an italian study on sleep. This new study from shows that Americans downplay sleep as crucial to health

Although they admit that a lack of sleep affects their work and sex lives, the majority of Americans don't believe that sleep is an important factor in maintaining good health, a new poll finds.

The survey of 2,442 Americans, conducted by SleepApneaInfo.Com., which is run by companies that make sleep disorder therapies, said that 62 percent of Americans thought that diet and exercise were more important than sleep in staying healthy and happy. Seven percent said none of the three was important.

According to the survey results, 58 percent of employed American experienced some daytime sleepiness, while 39 percent said that they don't have sex because they're too tired.

About 40 million Americans suffer from sleep disorders, and 18 million of them have obstructive sleep apnea, in which the person stops sleeping many times a night because of a blocked airway.

With this type of results, one wonders how the sleeping industry could awake...

Monday, September 20, 2004

Lessons learned for asteroids and earthquake will eventually converge

Will a 5.5 earthquake in southern california that was 14 days late be considered a success by the Keilis-Borok team ? Please note that the miss of the Assissi earthquake is symmetrical to this southern california miss. The M8 algorithm said the Assissi would be 5.5 and ended up being 6.5 whereas this one was predicted to be 6+ and ended up being late and less than expected. What is seriously missing in these stories is the idea of probability distribution, there is no way the M8 algorithm said there was no chance whatsoever of a quake after September 5th if it had not occured by then. At one point or another, like the controversy that erupted for near earth miss of asteroids, this incident will lead the scientific community to be more behaved when it comes to warning the general public.

Modelling Criminality

Modelling criminality in the urban environment seems to be a new area of scientific investigation; the title "crime mapping as anticipation" reminds me of a recent movie.

How should we evacuate New Orleans ?

You'd think that evacuation of a town like New Orleans is something that is well understood for no other reason that its clear exposure to the elements. Well it seems that it is not really. What I do not understand is how can the federal government invest money in evaluating traffic jams in Dallas or Portland but not in the evacuation of a major city like New Orleans.

Thursday, September 16, 2004

Google's playground

Google Labs is an interesting place where they seem to test their new products. A good way to do rapid prototyping.

Number Theory can even get you a job !

It is a stunning news indeed. Besides putting observatories at the right locations, number theory can also get you a job at Google. The fascinating thing, is that most researchers in Number theory would agree it is not challenging when it can be found easily.

GPU programming VI

While reading GPU Gems, I found the article I really wanted to see: How do you go from a numerical analysis background to programming with a graphics card. How do you map what you already know into something that a chip does well. Well Mark Harris explains how you solve the Navier Stokes equation on a GPU. That's enough for me to do the rest of the mapping I think.

Tuesday, September 14, 2004

Vortices and Spacecrafts

Am I the only one to make a parallel between this movie and this picture.

Spilling your guts for only $3,000

Cable just sent me this MSNBC article on Zero-gravity flights going mainstream. It has been a long time in the making and knowing some of the people involved I can only say it took some stamina. The idea of allowing the plane to be used as cargo was one of the few ways one could make this whole operation work because otherwise the economics is not very good (I checked some years ago.)

In the meantime, checking the survey from the MSNBC site reveals that out of 4200 people only 11 % would not go or are indifferent, 33 % would like to know where to sign up and 56 % thinks it is fun but too pricey. As having done it several times, I can only say that I surely would have paid that much to do it once. One of the very interesting aspect of these flights is that even though you are "having fun" for two hours and half. The experience is so novel to your brain, that it feels like you spent only fifteen minutes in the plane, so you really need to bring your camera.....

Should New Orleans be evacuated ? ( Doit-on evacuer la Nouvelle Orleans ?)

According to most computer models , Ivan
will hit New Orleans. New Orleans has the specificity of being 15 feet beloww sea level. Is this the big ONE ? Check how high the water level will get in the French Quarter.
. A similar question was asked earlier.

Monday, September 13, 2004

11 Empty Quivers

I never realized that there were that many. If no other reasons than making sure they are not going to detonate, it would seem to me that addtional manpower should be put to find them in order to potentially improve dramatically the stockpile stewardship program.

Sunday, September 12, 2004

Abstracts for Tex-MEMS VI are now available

They are located here.

Interesting numerical links of the week

A toolbox to do interval arithmetics: INTLAB - INTerval LABoratory Ver. 5

An interesting approach to algorithm development and start-up funding: Fang Inc. with interpolation schemes

Two books from C.T. Kelley of North Carolina now freely available on the web from SIAM: Iterative Methods for Linear and Nonlinear Equations and Iterative Methods for Optimization

A free non-equispaced fourier transform package at:

Eric Brochu, Nando de Freitas and Kejie Bao just made available this paper: Owed to a Martingale: A Fast Bayesian On-Line EM Algorithm for Multinomial Models . UBC CS Technical Report TR-2004-08.

and finally an MPEG-4 encoding library for the people: Revel

Exoplanet detection using number theory

This approach by Daniel Rouan to nulling interferometry as used for the detection of exoplanets uses number theory to define the location of the spacecraft-observatories. This is fascinating.

Saturday, September 11, 2004

The Davalos-Carron's Law seems to be holding.

The Davalos-Carron's Law states that the size of the text in all of one's Email stack is exactly a thousands times less than the size of the all its related file attachments (jpg, mpg, pdf,......) We currently have in our collaborative project system something like 12 MB of text for the different "E-mail" exchanges we have with each other (threads), but the size of the library associated with this application which hold all the attachments associated with these threads is now 12 GB.

GPU programming Part V

It looks like Google has difficulty making a difference between a good link to programming a GPU and one that only relates specific examples like my blog on the subject. My blog is on the first page of Google for the expression "GPU programming". It shouldn't.

Friday, September 10, 2004

They do lightnings but they don't do tornadoes

Second email of this kind today, after the rabid bats, here is the lightning warning. Now what I want to know is what to do when a tornado hits since it is more likely to hit this region than lightning. It looks like I am not the only one concerned about this issue.

Safety Issues - Take Lightning Hazards Seriously

Lightning is one of nature's leading causes of injuries and deaths each year in the United States. To alert students, faculty and staff of these hazards, lightning warning systems have been installed on the Texas A&M campus and at numerous parks and schools in College Station. When the likelihood of a cloud-to-ground lightning strike is high, these systems will sound one long 15-second blast from the horns. Intermittently flashing lights or strobes will continue for the duration of the hazard. Three 5-second blasts signal that the danger has passed and normal outdoor activities can resume. These systems are used solely to warn of lightning hazards.

If you hear the danger signal, you should find shelter quickly since the storms that produce the lighting typically arrive within minutes of the alarm. Organized outdoor activities should be suspended until the system gives the "all-clear."

When lightning hazards are present, avoid tall trees, flagpoles and high open areas. Golf carts do not offer the same protection as automobiles. Know the signals and protect yourself. ...

Are rabid bats beneficial ?

Just got this in my E-mail stack:

COLLEGE STATION: A reminder to avoid contact with bats is being made by Texas A&M University officials. The annual bat migration may be starting earlier that usual this year, perhaps due to milder than usual temperatures in August and September. Due to the expected increase in the number of bats arriving in this area during the annual fall migration, officials want to caution individuals about handling or approaching bats.

According to Dr. John Patton, a Research Scientist with Wildlife and Fisheries Sciences at Texas A&M University, only a small percentage of bats will have rabies. Although rabies can only be confirmed by a laboratory, any bat that is active during the day, is found in a place where they are not usually seen, or is unable to fly is far more likely to be rabid. These bats are often the most easily approached, and should not be handled.

Bats commonly found on campus are Mexican Free-tailed bats which have been named the Official State Flying Mammal of Texas. For this reason, universities and other entities that have to deal with bats are limited in ways to control them. When bats must be captured and removed from campus, they are transported to a remote area and released. In fact, bats perform a very beneficial service because they devour large numbers of insects.

Anyone on campus encountering a bat that is on the ground or otherwise acting in an unusual manner is requested to call Pest Management at....

So, if you see a rabid bat coming at you, you can't "take care" of it. uh....Only in Texas.

GPU programming IV

While it looked like the speed up was up to ten times, it looks like for solving specific equations, the speed up is only 4 to 5 times as fast. Good enough for me.

Wednesday, September 08, 2004

Tuesday, September 07, 2004

Hypertaskers don't do things faster

In this article, people seem to say that hypertaskers do things faster but not better. I would not be so sure about that. When you read the convincing argument of Joel on multitasking and parallel processing (which seems to be the process identified in the Harvard study), you realize that for tasks that are supposed to last about equal time, you will not do things faster for any of the two tasks.

Monty Python in Iraq

Enough said.

Nature: 4, Human Modeling: 0

  • Sharp downgrading of Frances before it hit the Floridian shores.
    Nature 1, Human Modeling: 0

  • Charley's sharp turn
    Nature: 2, Human modeling: 0

  • On Sunday, there was no 6.4 Earthquake in Southern California
    Nature: 3, Human Modeling: 0

  • But it ocurred in Japan instead
    Nature: 4, Human Modeling: 0
  • Monday, September 06, 2004

    The Columbia hit the foam, not the contrary.

    I was watching yesterday a show that looked like it was a NOVA program on the Columbia flight. What stroke me was that everybody was in denial and could not believe a foam piece could have done that much damage. However, a simple back of the envelope calculation using the high speed imagery of the take-off shows that the foam piece is being overtaken by the shuttle at a speed of 924 fps or about 1060 km/hr.

    Sunday, September 05, 2004

    Tex-MEMS VI is on

    The dreadful aspect of organizing a meeting like this is that you eventually seldom enjoy it. By that I mean you don't really get to see the good presentations or at least the ones you want because there is always something for the organizers to do. Case in point, I still don't have a title for my presentation yet, with the meeting on thursday, this is not good. At least being an organizer allows one to have a place holder. Most abstracts can be found here.

    Saturday, September 04, 2004

    Nice toy

    I was at the toy store today and found the following RC car. Not only does it do the RC car thing, but it also has a little camera (a CMOS) in the car that transmits video to the RC console.

    Friday, September 03, 2004

    GPU Programming III

    As a result of using GPU instead of the main CPU, it looks like we can go from 5 Gflops on a CPU to 50 GFlops on a GPU. Time to hone some skills.

    Frances from different altitudes

    From the International Space Station at 300 kms:1, 2, 3.

    From TRMM at 403 kms: 1

    From SeaWifs at 705 kms: 1,, 2, 3, 4.

    From the GOES satellite at 36,000 kms:1

    Wednesday, September 01, 2004

    Online parameter estimation for space related thermal problems

    We are going to present a paper at TFAWS 2004. The abstract is here. The idea is eventually to be able to do model estimation while we do payload qualification in the thermal vacuum environment. In general, the process involves a lot of back and forth estimation and testing to get the model right. The intent is to eventually use Particle Filtering techniques like MCMC.

    You want another type of FUD ?

    This article relates the fact that full-body CT scans pose high cancer risk. Where have these people been hiding ? Yes it is risky to take CT scans but it also absolutely far riskier to have an undetected tumor...