Information Processing

Just another WordPress.com weblog

Archive for March 2005

Hard or soft?

with 2 comments

There’s been a lot of discussion in the blogosphere about whether the world economy will experience a hard or soft landing as currencies adjust to the US current account deficit. I recommend two particularly nice summaries of the soft and hard scenarios.

One point I think people may have missed is that whether there is a hard (disorderly or panic/crash/recession inducing) landing or a soft (gradual, Plaza Accord-like) landing, the dollar is set to decline substantially against many Asian currencies, and perhaps the Euro as well.

Here is a very nice comment on the soft scenario post I linked to above:

[Here’s how we worked our way out of the last current-account crisis (1984-88). The figures are what would be required in 2005, 2006 and 2007, to match the cumulative change in 1984-88.

Exports of Goods +9% p.a.
Export of Services +9.7% p.a.
Income Receipts –0.8% p.a.

Import of Goods +1.5% p.a.
Import of Services +3.7% p.a.
Income Payments +6.9% p.a.

Net Transfers –7.4% p.a.

Current Account Balance –9.4% p.a.
Nominal GDP Growth +3.9% p.a.

Net results: Current-Account Balance falls from 5.7% of GDP to 3.8% of GDP in 2007. Not quite as good (3.4%) as in 1988, but the Dubious Administration got us in a lot worse trouble this time around.

Other interesting things going on at the same (1984-88) time: The US dollar lost 60.7% of value against the Swiss Franc between 1984 and 1988, 51% against an artificial Euro, 85.5% against the Yen, and 38.5% against the Korean Won…]

See here, here and here for related comments and historical data.

Advertisements

Written by infoproc

March 31, 2005 at 3:46 pm

Posted in Uncategorized

Tech gap

with 2 comments

Even Tech Execs Can’t Get Kids To Be Engineers (WSJ)  

Vinod Dham is among a growing number of technology executives warning that the U.S. faces an engineer shortage. To stay globally competitive, he says, the nation must do better at steering its youth toward engineering careers. Mr. Dham knows how hard that is: He can’t persuade his own kids to go into engineering.

The 54-year-old Mr. Dham would seem to be a prime role model. His engineering degree lifted him from his humble origins in India into a 16-year career at Intel Corp., where he became well-known for helping create the Pentium chip. His older son, 22-year-old Ankush, is studying economics, and that’s fine with Mr. Dham, who says he couldn’t get him interested enough to develop the rigor required for engineering. But ever since his younger son, 19-year-old Rajeev, was a boy, Mr. Dham has been urging him to pursue engineering — and he, too, is going into economics. Rajeev “doesn’t want to do electrical engineering,” the elder Mr. Dham laments. “He tells me the job will be outsourced.”

Silicon Valley is doing a lot of hand-wringing these days about a coming engineer shortage. Tech leaders such as Cisco Systems Inc.’s John Chambers and Stanford University President John Hennessey warn that the U.S. will lose its edge without homegrown talent. The U.S. now ranks 17th world-wide in the number of undergraduate engineers and natural scientists it produces, they point out; that’s down from 1975, when the U.S. was No. 3 (after Japan and Finland).

But some of the nation’s tech elite — including many immigrants who benefited greatly from engineering careers — are finding even their own children shun engineering. One oft-cited reason: concern that dad and his contemporaries will ship such jobs overseas.

Venture capitalist Promod Haque, for example, is in an ironic bind when it comes to advising his own kids. Like many other Silicon Valley financiers, Mr. Haque has recently begun funding tech start-ups in India and urging U.S. tech entrepreneurs to outsource from the start by forming companies that split operations between the U.S. and India. Mr. Haque chuckles about a recent dinner conversation with his college-age daughter, who he hoped would go into engineering just as he did. “She said, ‘Dad, I’m not going to take any more computer-science classes,’ ” he recalls. “I asked her why. She looked at me straight and said, ‘I don’t want to go to India to get a job.’ “

Experts cite a variety of other reasons for the U.S.’s engineer shortage, including poor math and science curricula in public schools. And there is also a persistent image problem. A recent study of 2,800 of Silicon Valley’s youth by consultants A.T. Kearney found that 73% were familiar with high-tech careers but only 32% wanted to pursue them. In describing tech careers, students in the study used a variety of unflattering terms, including “intimidating” and “uninteresting.” Others said they considered engineers to be “socially awkward” or “obsessed with work.” Some female respondents linked computer engineering with work that is “tedious” or “antisocial.”

That was the case for Susan Mason’s two stepdaughters, Alexandra and Joanna. Ms. Mason, a Silicon Valley venture capitalist with a background in computer engineering, says she urged the girls to consider engineering when they were in high school. They ignored her advice: Alexandra became an audiologist and Joanna went into nursing.

“They felt that engineering was too solitary, even if they were working in a team environment,” Ms. Mason recalls. “They wanted to have more interactions with people on a ‘human’ level,” she says…

Written by infoproc

March 29, 2005 at 6:34 am

Posted in Uncategorized

How much information in the universe?

with 2 comments

Suppose the universe is described by a single wavefunction which evolves in time according to Schrodinger dynamics. In this framework remarkably little information is required to describe the universe as a whole.

Consider the sub-volume V, just after the big bang, which evolves into our observable 15 Gy universe. Now suppose there is an ultraviolet cutoff (or minimum resolvable length) given by the Planck length. Then the entropy (log of number of degrees of freedom) is of order V in Planck units. To neglect quantum gravity effects, we need to wait sufficiently long after the big bang that curvatures are small in Planck units, so this entropy is a large number, but still much smaller than the observed entropy of our current universe. Given the wavefunction over this large but finite Hilbert space, and the Hamiltonian, we can evolve this system forward in time until today. Neither the informational complexity (number of bits required to specify the initial state) nor the algorithmic complexity (length of program required to evolve the system) is very large. (Note that memory requirements could grow quite rapidly – especially in an expanding universe.)

If this confuses you, just imagine you had to write a computer program to evolve this finite system (remember, we have both UV and IR cutoffs) forward in time. How much input would your program need, and how long would the code be? Then compare the answer to what would be required, e.g., to simulate to just a small part of planet earth today.

So what is the origin of the apparent complexity of our world? The answer is that the wavefunction described above contains all branches of Everett’s many worlds (see previous post). In order to locate your particular branch (the one on which your consciousness resides), you have to specify the outcomes of the branchings in your past (or, at least, the important ones – I think Gell-Mann and Hartle refer to this as decoherent histories). The amount of information required to specify a particular history is related to the number of possible present universes, and is mainly responsible for the complexity we observe.

Written by infoproc

March 27, 2005 at 7:15 pm

Posted in Uncategorized

Everett and many-worlds

with 3 comments

I recently found this wonderful biography of Hugh Everett III, the discoverer of the many-worlds interpretation of quantum mechanics. While its name suggests science fiction, one is naturally led to this interpretation by requiring completeness and economy of quantum mechanics.

In the usual formulation, an isolated system is described by a wavefunction which evolves unitarily in time according to the Schrodinger equation. When an external observer makes a “measurement” on the system, its state “collapses” into a particular eigenstate associated with the quantity (operator) that is measured, such as position, momentum, spin, etc. However, an astute thinker (like Everett) is led to ask why one has to introduce the rather mysterious concepts of measurement and collapse. Shouldn’t the universe as a whole be described by a single wavefunction, which evolves according to Schrodinger dynamics? In that case, a measurement is a process by which the state of the observer’s brain (or apparatus) becomes correlated with the state of the observed system. All possible outcomes are described by different “branches” of the universal wavefunction (hence the many-worlds description). Each of our consciousnesses is a particular semi-classical branch of the part of the wavefunction describing the atoms and molecules in our brains.

Everett’s interpretation doesn’t (as far as I know) make any new experimental predictions – it predicts the same probabilities as the old Copenhagen measurement-collapse description. (There is some debate over whether it gives a more satisfying explanation of how probability arises in quantum mechanics.) However, I and many others find it more intellectually satisfying and complete. The real argument starts over interpretation of the “other” branches of the wavefunction. Are they real, or just a mathematical abstraction? Can we ever detect their existence?

The biography I linked to fills in a number of the historical questions I had about many-worlds. Why is it that so many younger physicists (perhaps without thinking too deeply about it) accept many-worlds, while many (but certainly not all) of our elders regard it as nutty? Apparently the idea languished in obscurity for 20 years before being re-introduced into the physics community by Bryce DeWitt and others. Everett himself, obviously a first-rate intellect, left physics and worked in the defense and software industries, where he made important contributions and became a wealthy man. This story should be a cautionary note to academic scientists about ignoring the many creative and original thinkers who continue doing interesting things outside the academy.*

See here for a nice FAQ on many-worlds. See here for a letter to Physics Today from Murray Gell-Mann in which both he and Feynman are identified as believers in many-worlds.

* Only a small fraction of our students manage to stay in academic research – a quick calculation confirms this: we have zero population growth in academic science, but each professor produces several if not many PhD students in their career. One might argue that it is the best and brightest among our students who stay in academia, but even if true on average, that leaves many talented former-academicians in the real world!

Written by infoproc

March 27, 2005 at 6:28 pm

Posted in Uncategorized

Times on hedge funds

with 3 comments

NYTimes: [The numbers are mind-boggling: 15 years ago, hedge funds managed less than $40 billion. Today, the figure is approaching $1 trillion. By contrast, assets in mutual funds grew at an impressive but much slower rate, to $8.1 trillion from $1 trillion, during the same period. The number of hedge fund firms has also grown – to 3,307 last year, up 74 percent from 1,903 in 1999. During the same period, the number of funds created – a manager can start more than one fund at a time – has surged 209 percent, with 1,406 funds introduced in 2004, according to Hedge Fund Research, based in Chicago…

“Hedge funds are an innovation of compensation,” said one fund-of-funds executive. “It’s a compensation system, not an asset class.”

…In the same way that there is no quelling the bulls, there will be no quieting of the critics. The Horvitz family of Cleveland, which made its fortune in road construction, media and real estate, started investing in hedge funds in the 1990’s. A decade or so later, it has virtually no money in such funds, said Jeffrey Horvitz, who oversees his family’s investments. Too often, he said, the funds produced disappointing returns.

“Hedge funds are no longer attractive,” Mr. Horvitz said, noting the influx of start-ups. “I see no relief in sight, especially for taxable investors like us.”]

See previous posts here and here, and many others 😉

Written by infoproc

March 27, 2005 at 7:05 am

Posted in Uncategorized

Gizmo report

leave a comment »

I promised I would report on the Aiptek DV5300 I ordered. So far I would have to say it performs well, considering how inexpensive it was (note, I seem to have gotten a good deal, as the typical street price is around $120 now).

I’ve posted some video here (mostly of our physics building and the Institute of Theoretical Science – if your player has trouble connecting to the stream, just right-click and download the whole file). It is shot in ASF, 352 x 288 pixels and 30 frames per second. You can see some pixellation when I move the camera too fast. VGA (= TV) quality is possible at 640 x 480, but only 11 FPS, so is jerkier. The next generation of cameras at this price point will do 30 FPS at VGA resolution. Note that ASF is a Microsoft MPEG4 variant, so you may have trouble viewing it on OS X or a unix machine. I think you’ll see manufacturers supporting more open MPEG4 formats like divX before long.

Ergonomically the camera is very easy to operate, with separate buttons for still photos and video capture. Still photos are not bad, but the optics are not as good as in my Sony Cybershot. I think digital video cameras using MPEG4 are the future – 1 GB of SD flash will store hours of fairly high quality footage, and units with 20-40GB hard drives aren’t far away. Image files upload to a PC using a USB cable.

Core technologies in these devices: the CMOS (or CCD, in higher end models) chips for image capture, image processing CPU, some simple software controlling camera functions, and flash memory. Aiptek, a small Taiwanese company, designed the camera, but it was assembled in China. The market for these devices is obviously very competitive and fast-moving. The next generation of cameras will be out within a year. Higher end devices are typically made by Japanese companies (Sony, Sanyo, Panasonic), but probably also assembled in China. I don’t see a big technology gap here – given slightly bigger R&D budgets and better distribution the little guys could easily out-innovate the bigger players.

Written by infoproc

March 25, 2005 at 5:38 am

Posted in Uncategorized

China Development Forum

with 2 comments

Stephen Roach of Morgan Stanley on latest China Development Forum. He is not expecting any near-term drastic changes in RMB policy.

“My bottom line for the markets: With China’s risk-reward calculus acutely sensitive to the all-important stability constraint, I read the message from this year’s China Development Forum as pretty much a green light on the growth front. 

As always, the highlight of this conference is a private session with the Premier in the Great Hall of the People.   Not surprisingly, this year’s discussions were framed around the hot topic du jour — Chinese currency policy.  Our group of outside experts laid out both sides of the debate to Wen Jiabao.  He said nothing to tip his hand and simply reiterated that China continues to actively study the issue and is “now trying to select both the proper plan and timing for RMB reform.”  Here, as well, I suspect it will all boil down to stability.  In my own speech to the Forum, I stressed that the coming global rebalancing could have surprisingly important implications for Chinese stability — especially for a dollar-pegged currency that gets caught up in America’s long overdue current account adjustment (see “China’s Rebalancing Imperatives,” presented to the China Development Forum, 20 March 2005).  Because of its rapid integration into the global economy and world financial markets, China may well find that its stability anchor is less secure than widely perceived.

At the end of the meeting, the Premier shook his head and exclaimed, “I cannot sleep well at night with this issue of the RMB.”   He then glanced at his watch and politely excused himself.  As we left the Great Hall, a noisy motorcade rolled up to the official entrance.  We went out one door, and US Secretary of State Condoleezza Rice literally went in the other door.  There was a lot on Premier Wen Jiabao’s plate that day.  The growth and currency debates are only one piece in the big Chinese puzzle.  But in the end, always think stability when it comes to China — whether the issue is economic or geopolitical.  For that reason alone, and based on what I picked up at this year’s China Development Forum, I have little reason to doubt that China is once again going for growth.”

Written by infoproc

March 23, 2005 at 6:16 pm

Posted in globalization