Physics Embraces AI: A Development for One Percent of the One Percenters
July 22, 2020
Say what you like about Newton. Teachers have made gravity “real” to indifferent students with the apple on the noggin metaphor.
Physics teachers today face a different challenge. The “old school” ideas are not going to win promotions, grants, or — even better — a prize. Cash! Fame! The cash thing may work among those who are work from home dads and colleagues without a tenure track ticket.
What can physicists who are the one percent of the one percenters? The answer is to combine esoteric mathematical concepts with the future forward concept “artificial intelligence.”
AI can do physics and “AI in Physics: Are We Facing a Scientific Revolution?” explains this shift. Now between you and me, there are a number of revolutions underway, but the “real life” stuff is of scant interest to physicists in my experience. Einstein anecdotes notwithstanding, physicists are an interesting chunk of the one percent’s one percenters.
Your homework? Verify the over density equation and show each step. No shortcuts! This is forward leaning physics with “real” representations, simulations, and predicted properties. No apple either.
The write up states with significant seriousness that symbolic regression:
can be used to derive mathematical formulas from the internally represented relationships in the network. Symbolic regression is carried out as a genetic algorithm. Equipped with variables and mathematical operators, the algorithm searches for the simplest mathematical formula with which known data can be reproduced.
Like many helpful mathy statements, this statement illuminates the process:
Their result clearly shows that the mixture of data, neural graph networks and symbolic regression is actually suitable for extracting mathematical formulas – in this case an already known natural law – from data with AI.
I enjoy the “clearly.”
But the future is not the stuff one can see, touch, feel, sniff, or think about in substantive ways. The future is tackling Dark Matter with AI.
I learned:
The researchers used the neural grapheme network again. Each node contains information about a dark matter halo such as position, speed and mass and is connected to other halos at a distance of 50 Mpc / h. The network was trained with data from the Quijote Dark Matter Simulation , a collection of generated dark matter structures.
And there is a payoff. Ready?
After the training, the GNN was able to predict the desired property of the halos more accurately than previous models. Using symbolic regression, the researchers were then able to produce a previously unknown mathematical formula that has a lower error rate than the currently most commonly used human-made formula for the same task. The resulting formula was also better able to deal with previously unknown data. For Cranmer, this is a clear sign that the mathematical formula generalizes much better than the neural graph network from which it was derived. This coincides with our previous experience in physics, says Cranmer: “The language of simple symbolic models describes the universe correctly.”
Forget the apple falling on Newton’s slightly addled brain carrier. Think in terms of this metaphor:
If AI is like Columbus, computing power is Santa Maria
And Big Data? Of course, of course. One percent of one percenters know this.
Stephen E Arnold, July 22, 2020
Jargon Alert: Direct from the Video Game Universe
July 22, 2020
I scanned a write up called “Who Will Win the Epic Battle for Online Meeting Hegemony?” The write up was a rah rah for Microsoft because, you know, it’s Microsoft.
Stepping away from the “epic battle,” the write up contained a word from the video game universe. (It’s a fine place: Courteous, diverse, and welcoming.)
The word is “upleveled” and it was used in this way:
Upleveled security and encryption. Remote work sites, especially home offices, have become a prime target for a surge in cybersecurity attacks due to their less hardened and secure nature.
A “level” in a game produced the phrase “level up” to communicate that one moved from loser level 2 to almost normal level 3. That “jump” is known as a “level up.”
Now the phrase has become an adjective as in “leveled up.”
DarkCyber believes that the phrase will be applied in this way:
That AI program upleveled its accuracy.
Oh, and the article: Go Microsoft Teams. It’s an elephant and one knows what elephants do. If you are near an elephant uplevel your rubber boots. Will natural language processing get the drift?
Stephen E Arnold, July 22, 2020
Optical Character Recognition for Less
July 10, 2020
Optical character recognition software was priced high, low, and in between. Sure, the software mostly worked if you like fixing four or five errors per scanned page with 100 words on it. Oh, you use small sized type. That’s eight to 10 errors per scanned page. Good enough I suppose.
You may want to check out EasyOCR, now available via Github. The information page says:
Ready-to-use OCR with 40+ languages supported including Chinese, Japanese, Korean and Thai.
Worth a look.
Stephen E Arnold, July 10, 2020
Interesting Supercomputer Item: Lenovo
July 2, 2020
“Lenovo, Top of the World Chinese Supercomputer Supplier, Sweeps All Markets” contains an interesting statement:
In the Top 500 list for June 2020, China is shown with a home installed base of 228 machines, whereas 20 years ago, in 2000, the country had just two of the top 500 machines installed. In comparison, the US had 258 machines in place 20 years ago, now it has just 117 supercomputers – of which 44, or 38%, are Chinese Lenovo machines. And to further hammer home China’s success, not a single one of the country’s own huge installed base of 228 machines is an American machine – there are no Crays, no IBMs, no Dells. Plenty of American chips, but no American supplier presence.
But wait. Was Lenovo an IBM unit?
The answer is, “Yes until 2005.”
The question is, “What was Lenovo’s management able to do with a unit IBM deemed surplus?”
Answer: Nose into new markets.
Why? Let’s ask Watson.
Stephen E Arnold, July 2, 2020
The Cancel Culture in Technology: A New Approach to Sustained and Informed Discussion
July 1, 2020
DarkCyber sifts through a range of content. Some of it is becoming repetitive. Acquisition of promising start ups like Google’s devouring of a rival maker of smart glasses. The story? Competitive fear, a desire to make hay after burning the field and most of the equipment barn, or an easy way to get some employees not yet prone to management resistance while doing the WFH thing. More details about this deal appear in “Google Completes Acquisition of Ontario Smart-Glasses Maker North.”
Another repetitive theme is turning off, disconnecting, and cancelling. This is not the wonky folks living in SUVs and converted delivery trucks. This dropping out is not the Timothy Leary thing. The new approach to cancelling embraces throwing $450 million into the bonfire nobody cared about: The Microsoft retail stores. And top experts in smart software leaving Twitter because of a New Age “conversation.”
“Yann LeCun Quits Twitter Amid Acrimonious Exchanges on AI Bias” brings the culture of open range disputes between sheep herders and cattle ranchers into the zippy 24×7 digital era.
The write up explains in Silicon Valley speak that sheep muddy drinking water and cows do not. Sheep ruin the grazing land. Cattle do not. How is the dispute resolved? As I recall one of my addled teachers explaining, the approach involved poisonings, shootings, fencing, and law enforcement. I am not sure that the problem has been eliminated, but I will generalize that most people do not care about muddy streams and sparse grass.
Today we care about smart software.
The write up points out:
Penn State University Associate Professor Brad Wyble tweeted “This image speaks volumes about the dangers of bias in AI.” LeCun responded, “ML systems are biased when data is biased. This face upsampling system makes everyone look white because the network was pretrained on FlickFaceHQ, which mainly contains white people pics. Train the *exact* same system on a dataset from Senegal, and everyone will look African.“
Definitely contentious.
What interested DarkCyber, however, is not the socio-tech discussion. The message seems to be “I can’t talk to you so I am out of here.”
This is a nice way of hitting the cancel button.
Several observational questions:
- Is this a sheep versus cattle argument?
- How does technology’s refinement processes operate when improvement muddies the drinking water?
- How does dropping out, turning off, and tuning out contribute to innovation?
Cancel means more than not tweeting. Cancelling is officially a trend even for allegedly informed and enlightened techno-herders.
Stephen E Arnold, July 1, 2020
Quantum Computing Has Time to Build Some Hype Momentum
June 30, 2020
After artificial intelligence, quantum computing wants to be a leader in the hyperbole arms race. “Value of Quantum Computing Uncertain for at Least 10 Years” makes a case for the quantum cheerleaders to adopt two Stalin five year plans. These can be implemented back to back.
The article quotes an outfit called Lux Research (which I assume connotes expensive information) as learning:
Quantum bits, or qubits, are inherently unstable, thus reducing the accuracy of any computation that relies on them; this is the first major obstacle to commercialization. For this reason, problems that lack clearly defined answers (like machine learning) but still benefit from improved solutions are the best problems to target with quantum computing.
DarkCyber thinks this means that graduate programs and venture capitalists have plenty of time to make their personal and financial investments pay off.
In the meantime, quantum computing cheerleaders can perfect their routine without too much fear of a rival coming up with a show stopper. And conferences? Absolutely.
Stephen E Arnold, June 30, 2020
Big Blue, Number Two
June 24, 2020
In the land of IBM, Watson works and the once towering giant, progenitor of OS2, and big disc drives that made “crash” a popular term is in the news again.
“ARM Based Fugaku Supercomputer Now World’s Fastest Supercomputer” reports that
The Fugaku supercomputer located in Kobe, Japan and developed jointly by RIKEN and Fujitsu Limited recently took the top spot in several supercomputer rankings making it the first time since June 2011 that Japan has held the Top500 supercomputer list crown and the first time ever that a supercomputer has simultaneously hit the HPCG, HPL-AI, and Graph500 world records.
And guess what company is Number 2? IBM and its IBM-built Summit system located at the US Oak Ridge National Laboratory. When not working on Covid, Summit matches possible owners to stray dogs.
Stephen E Arnold, June 24, 2020
Physicists May Be Inventors
June 23, 2020
Physics is a fascinating subject. There’s the high school variety involving steel balls, magnets, and fire. Then there is the world of wonky “things” like quarks, flavors, and wimps. (No, wimp does not mean a polo player afraid of falling off his or her equine dynamical system.)
“CERN Wants to Build a new $23 Billion Super-Collider That’s 100 Kilometers Long” explains that the hadron folks need to up the ante. The idea is that wackos who embrace string theory and the neo-Einsteinians will be outflanked; misguided miscreants who cannot dis-CERN that progress in physics is more than wonk-babble emitted by pundits who are not bounded by the time and space of mere mortals.
The write up reports:
CERN wants to build a successor to the Large Hadron Collider to further study the Higgs boson particle.
I learned:
The Large Hadron Collider took a decade to build and cost around $4.75 billion. Most of that money came from European countries like Germany, the UK, France and Spain. Some believe that countries like the US and Japan might need to pony up for this second collider if it’s actually going to get built.
The money will be found! Physicists have to have a gizmo big enough to permit physicists to make the leap between observing fundamental phenomena to creating objects.
Mother Nature is obviously not performing up to the Ernest Rutherford. Are physicists becoming inorganic chemists with a better understanding of fancy math? Yep. The need to find has may be veering into create via a big, expensive machine hopefully with better reliability than the existing collider.
Stephen E Arnold, June 23, 2020
Free Dissertation? Act Fast or You May Have to Pay Up and a Lot
June 20, 2020
DarkCyber spotted “Discovering Dennis Ritchie’s Lost Dissertation.” The main point of the write up is that a wizard failed to hand over a copy of his dissertation to the institution library. As a result, no PhD and no scanning, indexing, and selling of the good student’s work by University Microfilms. I have no clue what this outfit is called today, but in the 1960s, the outfit zoomed through Kodak film and helped animate environmental controls on photoprocessing chemicals. Silver and all that, of course.
The main point of the write up for me is the link to the aforementioned dissertation. Free and online as June 20, 2020, at Ritchie_dissertation.pdf. Miss this chance and you may have to pony up some hard cash for a professional publishing/database company’s honest work of making money by converting students’ fear and perspiration into an online charge.
Oh, what did the student cook up? The C language.
Stephen E Arnold, June 20, 2020
Springer Free Computer Science Books
June 16, 2020
The list of free Springer computer science books is at this link. More than 40 books are available. Our faves include The Algorithm Design Manual and Introduction to Evolutionary Computing. DarkCyber did not ask, “Why?”
Stephen E Arnold, June 16, 2020
-
- Subscribe to Beyond Search
Feature archive
News archive
-
