Search This Blog

Monday, November 30, 2020

GPS Rules Everything. And it's Getting a Big upgrade

It's a lot more than just driving directions. GPS, managed by the US Space Force, is embedded throughout the high-tech world we live in.

 

On Nov. 5, a SpaceX rocket roared into the heavens from Cape Canaveral, Florida, carrying a boxy, 5,000-pound, antenna-studded satellite toward its destination 12,500 miles away, up in what's known as medium Earth orbit. From that distant vantage point, it'll soon beam signals that will help you find your way to a friend's new house out in the suburbs or a vacation destination six hours down the coast.

If you stop at an ATM along the way to grab some cash, those signals will also help the bank know your withdrawal happened after your direct deposit paycheck refreshed your finances. They'll be a factor, too, in whether your cellphone call to your friend, or the rental agent, goes through without garbling or fading.

Those signals will be coming from a GPS III satellite, the newest member of a constellation of satellites that have become a constant and intimate presence in our daily lives. With GPS III, we're getting not just new boxes in the sky, but a series of upgrades that'll help make the system better for all of us here on Earth. And we'll need it.

The Global Positioning System has become vital to nearly all sectors of the country's critical infrastructure, with much of its work happening behind the scenes, and likely to a much greater extent than you realize. GPS tells us where we are and helps us get where we're going, but a core aspect of the technology is when -- the timing of, well, more or less a zillion things. It plays a critical role in financial transactions and stock trades, forecasting the weather, monitoring earthquakes and keeping the power grid humming.


"It's so much more than just driving directions," says Tonya Ladwig, acting vice president of space navigation systems at Lockheed Martin, which built that satellite.

According to a study last year commissioned by the National Institute of Standards and Technology, GPS has about $1 billion a day in economic impact in the US. Its reach is, simply, mind-blowing.

"Gauging the overall value of GPS is nearly impossible," writes Greg Milner in Pinpoint, a 2016 book about how the space-based system came to be and the effect it's having on the world. "It has become difficult to untangle the worth of GPS from the worth of everything."

That's a lot to put on the shoulders of not much more than a couple dozen satellites and what turns out to be a wisp of signal by the time it reaches your phone or an airplane coming in for a safe landing. Which is why experts and lawmakers have long fretted over GPS' susceptibility to jamming and spoofing and the possibility that this invaluable resource could become a single point of massive failure.

See also: Space has become a junkyard, and it's getting worse

GPS is the premier service among just a handful of global navigation satellite systems, or GNSS, which include the European Union's Galileo, Russia's Glonass and China's BeiDou. It's in the midst of a long-running modernization intended to deliver better signals to folks on the ground and to make the satellites more robust in space. That's good news not just for Uber drivers, pilots, bankers, geologists, farmers doing precision agriculture, and users of drones and self-driving cars but also for the sector that got the whole GPS ball rolling in the first place: the US military.

And the military isn't just a heavy-duty GPS user. It also runs the service, for all of us around the world.

How GPS works

What makes GPS an always-on resource -- every bit as much a reliable utility as the electricity and water in your house -- is the coverage the satellites provide.

There are 31 satellites in the GPS constellation, and 24 are considered the minimum for the core constellation to work as it's supposed to. Those two dozen are spread out in six orbital planes, so you should always be within view of at least four at any given moment. The remaining seven are essentially spares, to be rotated in as necessary. Though they're continuously beaming signals down to Earth that you'll pick up in your phone, fitness tracker or boating sat-nav device, they don't know where you are. They just broadcast, like a radio station in space.

GPS III satellite nears the end of production at the Lockheed Martin facility.

A GPS III satellite stands tall at Lockheed Martin's Littleton, Colorado, facility, in May 2018, a year and a half before its launch into orbit. The striped elements on top are antennas, which will be pointed earthward when the satellite is on orbit.

Lockheed Martin

"The GPS satellites are actually just highly precise atomic clocks, hooked to a radio transmitting a time signal," says Dana Goward, president of the Resilient Navigation and Timing Foundation, a Washington, DC-based nonprofit.

On the ground, your GPS receiver -- which is what your mobile phone is, thanks to a GPS chip -- picks up the signals from four or more satellites. By measuring slight differences in the signals' time of arrival, all the way down to nanoseconds, it can calculate where you are and whether you're in motion.

"[Location] is a byproduct of how the system works," says Scott Burgett, director of GNSS and software engineering at Garmin, which makes devices including fitness trackers and smartwatches. "All the satellites transmit signals, and they're synchronized pretty accurately, but in order to actually get your position information, you have to solve for time."

The timing data gets translated into highly precise three-dimensional location information -- latitude, longitude and altitude -- as well as speed and direction. That's where Google MapsApple Maps and other geographic information systems come into play. It's how we get to the part where you have a street address and you say, "I'll put that in my GPS," and Waze lets you know to take Exit 27, go 3.5 miles and turn left into the parking lot of the beer and burger joint you've heard good things about.

Or it just gets used as a time stamp, pure and simple. Think financial transactions, for instance.

"The timing aspect of this is probably more widely used than the where-are-you aspect," says Goward.

Space Force reporting for duty

The US Space Force operates and maintains the GPS constellation. Each satellite -- picture a PODS storage container, metallic rather than white, with solar arrays sticking out like a pair of wings -- makes two transits around the planet every day. 

Even as precisely programmed as those orbits are, the satellites still need their flight paths tended to around the clock. 

"Those GPS vehicles are only as accurate as the data we provide them," says 1st Lt. Andrew Johnson, a crew commander in the 2nd Space Operations Squadron, or 2 SOPS. "We get where the satellite thinks it is, we know where the satellite is, and we'll basically bake that into a nice little message, we'll send it up to the vehicle, and the vehicle goes, 'OK I'm actually here,' and that change in information finetunes the signal."

2nd Space Operations Squadron

Two members of the 2nd Space Operations Squadron designate the first GPS III satellite as healthy and active for users on Jan. 13, 2020. To make room for it, 2 SOPS pushed a GPS IIA satellite to a higher, less congested, "disposal orbit." Designed for 7.5 years of service, the older satellite had been operational for 26 years.

US Air Force photo by Staff Sgt. Matthew Coleman-Foster

Johnson and 2 SOPS (pronounced "two sops") keep tabs on the GPS satellites from Schriever Air Force Base, located just east of Colorado Springs, Colorado. There are also 16 tracking stations scattered around the globe.

It's no accident that the US Space Force, spun off a year ago from the Air Force Space Command and carrying on its GPS mission, is wrangling a service that's vital to devices used by millions of civilians and businesses worldwide. The origins of GPS stretch back to secret work by the Department of Defense in the 1970s, in a quest for precision targeting. As Milner recounts it, GPS chief architect Brad Parkinson summed up that goal in the phrase "Drop five bombs in the same hole."

See also: How SpaceX Starlink broadband will envelop Earth and transform the sky

In 1983, after a Korean Air Lines passenger jet strayed into Soviet airspace and got shot down, killing 269 people, President Ronald Reagan declassified GPS to give civilian aircraft access to the navigation signals. Almost a decade later, GPS famously earned its stripes as a military resource during Operation Desert Storm, when it helped guide US and allied forces across desert expanses to a swift victory over Iraq during the Gulf War.

Space Force still has military users top of mind as it carries out its GPS mission. 

"For us, it's to deliver sustained, reliable GPS capabilities to America's warfighters," says Maj. Gen. DeAnna Burt, director of operations and communications at Space Force headquarters in Peterson Air Force Base, Colorado. Space Force also works closely with civilian and commercial partners to keep things running smoothly, she tells me. "We're always looking to improve not only our military capabilities but our civilian capabilities as well."

Though the funding to keep things running goes through the Pentagon -- the Space Force GPS program had a 2020 fiscal year budget of $1.71 billion -- there's civilian oversight as well. The Defense Department and the Transportation Department co-chair the US government's National Executive Committee for Space-Based Positioning, Navigation and Timing, which coordinates GPS-related matters across federal agencies and includes representatives from Boeing, GarminGoogle, Ohio State and Stanford.

Note the keywords in that committee name: positioning, navigation and timing, or PNT. Where you are, where you're going, and when the signals hit a receiver. It's a term that's inescapable when you're talking with folks who live and breathe GPS.

What GPS III brings

Like any technology of a certain vintage -- the Air Force Space Command declared full operational capability for GPS in April 1995 -- the system needs to be regularly updated, and what that means right now is GPS III.

Here's what GPS III promises: The signals will be three times stronger, and they'll have eight times the anti-jamming capability. The satellites are projected to have a 15-year lifespan, double that of those from the early part of the previous generation, though the older ones have tended to stay in business longer than expected. A modular design means it's easier to make timely changes on the assembly line or to send software uploads to the satellites on orbit.

There's also a new civilian frequency, called L1C. Besides helping with signal strength, it's compatible with Galileo, the EU's counterpart to GPS. 

In November 2018, the FCC authorized Galileo signals to be received in the US, which made it that much more likely you'll have multiple satellites in view -- in the double figures even, when technically you only need four to get a good, accurate location. The addition of the L1C signal with GPS III will likely make matters even better. 

"If you have more satellites," says Garmin's Burgett, "you can have more direct line-of-sight signals available to you and you can get a better fix."  

The military, meanwhile, is getting, among other things, the encrypted M Code that's key to the enhanced anti-jamming and anti-spoofing capabilities, as well as spot beam capability for focused signals in combat areas.

A little bit down the road, the addition of a laser retro-reflector array will allow the positioning of satellites to be refined via ground-based laser.

Lockheed Martin GPS satellite production

GPS III satellites at various stages of production in Lockheed Martin's Littleton, Colorado, facility.

Lockheed Martin

The first of the GPS III generation of satellites, all built by Lockheed Martin at its Littleton, Colorado, facility, lifted off in December 2018 and became operational in January of this year. The one that lifted off Nov. 5 is the fourth in the series, and it should be ready for duty before we get too deep into 2021.

Lockheed Martin has a contract to deliver a total of 10 GPS III satellites, at a reported average cost of $529 million apiece, but the company says the last two of them will come in at around $200 million each. When that's done, it'll move ahead with a batch called GPS III F, an additional 22 satellites to continue replacing older models, through the coming decade. 

"It takes a long time to replenish the GPS constellation," Burgett says. "It takes years."

Weak spots

It might seem like GPS is pretty much always there when you need it, but it's more vulnerable than you may realize. If you live in a city with tall buildings, you've probably fumed waiting for an Uber driver to get to where you're standing -- it could be that buildings are blocking the satellite signals in what's known as the urban canyon effect.

That's a line-of-sight issue, and it can often be resolved by moving, if you can, to a spot with a better view of the heavens. The US government says that GPS-enabled smartphones are typically accurate to within a 16-foot (5-meter) radius under the open sky.

Lockheed Martin GPS IIR-M satellite

A Lockheed Martin engineer works on a GPS IIR-M satellite in 2005. According to GPS.gov, seven of this generation of satellites are still operational.

Lockheed Martin

Then there's interference -- other, stronger signals making too much radio "noise" nearby. 

"Because it's such a weak signal, it's very, very easy to block, to jam," Goward says. "Virtually any noise within that frequency is going to keep you from hearing the GPS signal."

Space Force's Burt likens it to a nearby sound system at full blast: "If you were at the dinner table and there was a 500-watt stereo playing at full volume in the kitchen, would you be able to hear the conversation going on around you? You might pick up pieces, parts of it, but not pick up all of it."

The US military has to worry about hostile forces jamming or spoofing GPS signals to hide troop movements or to keep friendly forces from getting where they're supposed to go, or weapons from hitting their targets.

Outside of war zones, some countries use GPS interference to mask the whereabouts of VIPs, while criminals use it to pull off shipping heists. The nonprofit Skytruth, which uses satellite images and data to track polluters and poachers, last year reported on GPS manipulation at oil terminals in China likely intended to hide activities that run afoul of export controls.

The Pentagon and other government agencies, meanwhile, are aghast at the Federal Communications Commission's approval last April of a controversial plan by a company called Ligado to create a nationwide 5G network. The frequencies Ligado would be using are very close to those employed for GPS. Ligado says it's sorted out any interference issues, but Defense Department CIO Dana Deasy said in a Senate hearing in May that "there are too many unknowns and the risks are too great."

The vulnerabilities of the satellite signals are something the GPS community has been thinking about for a long time, along with the need for some sort of backup -- the idea being to provide a ground-based service that might not be as good but that would suffice when somebody's jamming or spoofing or if the satellites aren't available. 

There have been a number of false starts down that road over the years. A new push came at the end of 2018 with the National Timing Resilience and Security Act, which directed the Secretary of Transportation to establish a terrestrial timing system that could serve as a backup for GPS within two years. We're at that mark now, with nothing yet to show for it.

A more limited proposal came in February when President Trump signed an executive order on PNT, which at least got the NIST in October to draft guidance on developing a timing system free of GPS.

In the years ahead, there will be more Space Force rocket launches carrying the latest GPS III satellites from Lockheed Martin. Modernization of the constellation will continue apace, new applications will appear, and as much as we're hooked on GPS services now, we're likely to only get more dependent.

Satellite timing is everywhere on Earth and in everything.

"I think most people don't realize how much they depend on GPS day in and day out," Space Force's Burt says. "It would be a bad day if we didn't have GPS."

Monday, November 9, 2020

China Launches World's First 6G Experimental Satellite

China has launched a 6G experimental satellite to verify the performance of the technology in space as this frequency band will expand from the mmWave frequency to the terahertz frequency.

As reported by ITWireA report by the Yicai Global website said the satellite was one of 13 put into orbit on 6 November.

The 6G satellite, named after the University of Electronic Science and Technology of China, was jointly developed by Chengdu Guoxing Aerospace Technology, UESTC, and Beijing MinoSpace Technology.

The 6G technology is expected to be more than 100 times faster than 5G, making it possible to achieve lossless transmission in space making long-distance communications with lesser power usage possible.

Yicai Global quoted Lu Chuan, head of the UESTC’s Institute of Satellite Industry Technology, as saying that the technology would enable wide use of 6G for satellite Internet.

He said the satellite carried an optical remote sensing load system to monitor crop disasters, prevent forest fires, check forestry resources, and monitor water conservancy and mountain floods, besides providing abundant satellite images and data.

Monday, November 2, 2020

Planes Continue to Fly into a GPS Dark Hole Over the Mediterranean, Puzzling Experts

 


As reported by FortuneGPS, the global positioning system that underpins everything from Google maps to telecommunication networks and aviation navigation, has become unreliable across huge stretches of the Mediterranean, Caucasus and the Middle East, as geopolitical tensions have risen in the region, according to the European agency responsible for safe air traffic management.

Air traffic experts worry that a prolonged disruption to GPS could put the safety of commercial airline passengers at risk.

Reports of GPS outages submitted by pilots from the cockpits of commercial flights show that disruptions to the navigation system, which was created and is maintained by the U.S. government, are now standard occurrence on the flight routes between North America and Europe and the Middle East, according to data from the European Organization for the Safety of Air Navigation, known as Eurocontrol.

In 2019, the regulator received more than 3,500 reports of outages, an all time high, according to Gerhard Berz, the senior expert for navigation systems and radio spectrum coordination at the agency. That worked out to an average of about 10 reports per day throughout 2019 being reported to the agency.

This year, the sharp decline in air traffic due to the global pandemic means that fewer planes are flying, and so there's less visibility into how common disruptions are, Berz says. Even still, the agency is receiving at least one report per day of outages, he says. Because reporting outages is entirely voluntary, the full scale of the disruption is expected to be far higher than the figures suggest.

Although the outages are assumed to be a knock-on effect from conflict zones and militaries on the ground, their sheer scale means it's not clear who is responsible for all of the outages, according to experts. The leading assumptions from experts in the area is they come from military operations from multiple countries in the region, including airbases and vessels—in the East Mediterranean, one of the rare examples where public attribution has been made, such outages have been traced to an airbase on the coast of Syria. Berz says the fact that the outages are intermittent and not continuous, and therefore difficult to predict, makes it even harder to trace the disruptions to their exact source.

In several cases, those outages had strange and potentially dangerous implications, including several instances where confusion to the GPS systems triggered a warning from the plane’s Ground Proximity Warning System, which warns a pilot they are about to hit land or an obstacle and must immediately “pull up” the nose of the plane to avoid crashing. Such warnings are usually worst-case scenarios. In these cases, they were false alarms that were correctly ignored by the pilots, who could see their locations clearly, said Berz.

But requiring pilots to selectively ignore urgent safety warnings goes against their training that such an alarm is a key safety net when ground control has failed, and should never be ignored, he said.

“We don’t like when confidence in those systems is undermined,” Berz says.

Flying blind

In other cases, GPS receivers that were disrupted while flying through the region did not regain a signal again after leaving the disrupted region, said Berz, requiring pilots to fly for hours without the GPS system. 

Commercial planes have multiple systems for navigation and communication—GPS is just one. That means that losing GPS, especially for short periods, is not necessarily dangerous, and pilots can function without it. However, Berz warned that it is the “sheer magnitude” of the disruptions that is worrying, because it raises the risk that a disruption could line up with a failure of other key equipment, or that in dark or stormy weather it may be harder to use other cues to navigate. 

There is also the added risk in areas where geopolitical tension is high. And for the few flights passing between Europe and the Middle East, including enroute to Asia, traveling through such risky region is unavoidable. Because Syrian airspace is closed, planes must fly either south of Syria through the Mediterranean, or north, through Turkey and the Caucasus. Both routes see regular reports of outages, Berz said. The area affected includes spill-over from the Syrian war, an ongoing conflict in Libya, and the recent explosion of tensions between Armenia and Azerbaijan, to name but a few.

Cyprus, in particular, has struggled with persistent outages for years now, affecting both maritime and air traffic, as Fortune reported in January as part of a series on the impact of disrupted GPS on the maritime sector.

The country has repeatedly put out official government aviation warnings of GPS interference throughout 2020, via its department of civil aviation. The U.S. Coast Guard, meanwhile, has now reported disruptions extending into the Mediterranean Sea as far Northwest as Italy.

The risks of such widespread disruption has gradually gained more and more attention in the sector. In August, the UN's aviation agency, the International Civil Aviation Organization (ICAO), passed a proposal recognizing the impact of "harmful interference" to GPS, and suggested a range of actions to address it. That included encouraging members to develop contingency systems for GPS failure, and to support alternatives to GPS; address the sale of illegal "jammers"; and recognize that risks to GPS from conflict zones can spill beyond the affected area.

It is difficult to overstate the ubiquity of GPS in both military and civilian life. On-land GPS receivers transmit precise time and location data to vital infrastructure, providing a level of accuracy that is used to sync everything from global telecommunications networks, to banking transactions and ATMs, to stop lights and energy management systems.

Saboteurs

GPS disruptions can be caused by weather, user equipment or by other innocuous reasons. But the disruptions of the scale seen over parts of the Mediterranean and Middle East are believed to be largely caused by intentional “jamming” and “spoofing,” two techniques used to disrupt or confuse signals sent from GPS satellites to receivers on earth. Though GPS is only one of several satellite global navigation systems—China, Russia and the EU all have comparable networks—GPS is the most widely used globally, and all such systems (collectively known by the acronym GNSS) are vulnerable to disruptions. 

Although regular people, small operations and criminal networks, equipped with illegal jammers bought online, can and do regularly cause chaos to GPS signals, the scale of the disruption in the Mediterranean, central Asia and Middle East is believed to be the work of militaries and state actors, according to analysts and industry experts who specialize in GPS systems. NATO's Shipping Centre, which liaises with commercial shipping, has repeatedly linked such disruptions in the Med to nearby conflict zones.

For militaries and state actors, GPS disruption is a classic form of electro-magnetic warfare—after all, disrupting communications and navigation has always been part of conflict. But it is also part of a new form of wide scale, grey-area disruption that has gained pace in the last five to six years as drones and other GPS-directed surveillance and weaponry has become an increasing part of conflict.

Some of the first large scale, public reports of GPS disruption affecting civilian life were from marine vessels in the Black Sea near Crimea, who saw their GPS locations suddenly diverge from their real location—often putting them at inland airports—around the time when Russia had annexed Crimea.

Russia has been openly blamed for disruption to GPS, including during military exercises in northern Scandinavia, and in the eastern Mediterranean, where the Center for Advanced Defence Studies last year said disruptions in the region appeared to emanate from the Khmeimim Russian air base on the coast of Syria. (Russia has either denied, or ignored the accusations.) The Cypriot government and Eurocontrol have also attributed the disruption in the region to the conflict in Syria.

However, the total extent of the disruptions globally is far too large to be attributed to just one actor. Rings of spoofing, nicknamed “crop circles” have been tracked to ports in China and to bizarre locations off the coast of San Francisco, according to investigations by Bjorn Bergman at the NGO SkyTruth. And such disruptions are a feature of life near the coast of North Korea, where illegal fishing fleets reportedly operate on a huge scale. Such disruptions have also been tracked at a more local level: in late 2019, small circles of GPS disruption affecting mobile phones began to appear in central Tehran, according to a report made to the U.S. Coast Guard and provided to Fortune. 

Friday, October 30, 2020

AI has Cracked a Key Mathematical Puzzle for Understanding our World: Partial Differential Equations (PDEs)

Partial differential equations can describe everything from planetary motion to plate tectonics, but they’re notoriously hard to solve.


As reported by MIT Technology Review: Unless you’re a physicist or an engineer, there really isn’t much reason for you to know about partial differential equations. I know. After years of poring over them in undergrad while studying mechanical engineering, I’ve never used them since in the real world.

But partial differential equations, or PDEs, are also kind of magical. They’re a category of math equations that are really good at describing change over space and time, and thus very handy for describing the physical phenomena in our universe. They can be used to model everything from planetary orbits to plate tectonics to the air turbulence that disturbs a flight, which in turn allows us to do practical things like predict seismic activity and design safe planes.

The catch is PDEs are notoriously hard to solve. And here, the meaning of “solve” is perhaps best illustrated by an example. Say you are trying to simulate air turbulence to test a new plane design. There is a known PDE called Navier-Stokes that is used to describe the motion of any fluid. “Solving” Navier-Stokes allows you to take a snapshot of the air’s motion (a.k.a. wind conditions) at any point in time and model how it will continue to move, or how it was moving before.

These calculations are highly complex and computationally intensive, which is why disciplines that use a lot of PDEs often rely on supercomputers to do the math. It’s also why the AI field has taken a special interest in these equations. If we could use deep learning to speed up the process of solving them, it could do a whole lot of good for scientific inquiry and engineering.

Now researchers at Caltech have introduced a new deep-learning technique for solving PDEs that is dramatically more accurate than deep-learning methods developed previously. It’s also much more generalizable, capable of solving entire families of PDEs—such as the Navier-Stokes equation for any type of fluid—without needing retraining. Finally, it is 1,000 times faster than traditional mathematical formulas, which would ease our reliance on supercomputers and increase our computational capacity to model even bigger problems. That’s right. Bring it on.

Hammer time

Before we dive into how the researchers did this, let’s first appreciate the results. In the gif below, you can see an impressive demonstration. The first column shows two snapshots of a fluid’s motion; the second shows how the fluid continued to move in real life; and the third shows how the neural network predicted the fluid would move. It basically looks identical to the second.

The paper has gotten a lot of buzz on Twitter, and even a shout-out from rapper MC Hammer. Yes, really.

Okay, back to how they did it.

When the function fits

The first thing to understand here is that neural networks are fundamentally function approximators. (Say what?) When they’re training on a data set of paired inputs and outputs, they’re actually calculating the function, or series of math operations, that will transpose one into the other. Think about building a cat detector. You’re training the neural network by feeding it lots of images of cats and things that are not cats (the inputs) and labeling each group with a 1 or 0, respectively (the outputs). The neural network then looks for the best function that can convert each image of a cat into a 1 and each image of everything else into a 0. That’s how it can look at a new image and tell you whether or not it’s a cat. It’s using the function it found to calculate its answer—and if its training was good, it’ll get it right most of the time.

Conveniently, this function approximation process is what we need to solve a PDE. We’re ultimately trying to find a function that best describes, say, the motion of air particles over physical space and time.

Now here’s the crux of the paper. Neural networks are usually trained to approximate functions between inputs and outputs defined in Euclidean space, your classic graph with x, y, and z axes. But this time, the researchers decided to define the inputs and outputs in Fourier space, which is a special type of graph for plotting wave frequencies. The intuition that they drew upon from work in other fields is that something like the motion of air can actually be described as a combination of wave frequencies, says Anima Anandkumar, a Caltech professor who oversaw the research alongside her colleagues, professors Andrew Stuart and Kaushik Bhattacharya. The general direction of the wind at a macro level is like a low frequency with very long, lethargic waves, while the little eddies that form at the micro level are like high frequencies with very short and rapid ones.

Why does this matter? Because it’s far easier to approximate a Fourier function in Fourier space than to wrangle with PDEs in Euclidean space, which greatly simplifies the neural network’s job. Cue major accuracy and efficiency gains: in addition to its huge speed advantage over traditional methods, their technique achieves a 30% lower error rate when solving Navier-Stokes than previous deep-learning methods.

The whole thing is extremely clever, and also makes the method more generalizable. Previous deep-learning methods had to be trained separately for every type of fluid, whereas this one only needs to be trained once to handle all of them, as confirmed by the researchers’ experiments. Though they haven’t yet tried extending this to other examples, it should also be able to handle every earth composition when solving PDEs related to seismic activity, or every material type when solving PDEs related to thermal conductivity.

Super-simulation

The professors and their PhD students didn’t do this research just for the theoretical fun of it. They want to bring AI to more scientific disciplines. It was through talking to various collaborators in climate science, seismology, and materials science that Anandkumar first decided to tackle the PDE challenge with her colleagues and students. They’re now working to put their method into practice with other researchers at Caltech and the Lawrence Berkeley National Laboratory.

One research topic Anandkumar is particularly excited about: climate change. Navier-Stokes isn’t just good at modeling air turbulence; it’s also used to model weather patterns. “Having good, fine-grained weather predictions on a global scale is such a challenging problem,” she says, “and even on the biggest supercomputers, we can’t do it at a global scale today. So if we can use these methods to speed up the entire pipeline, that would be tremendously impactful.”

There are also many, many more applications, she adds. “In that sense, the sky’s the limit, since we have a general way to speed up all these applications.”