- Study sheds light on turbulence in astrophysical plasmas
- Building the hardware for the next generation of artificial intelligence
- Harnessing a “meritocracy of great ideas”
- MIT issues statement about early-morning fire near Central Square
- Vivienne Sze receives Engineering Emmy Award
- MIT's AIM Photonics Academy looks to expand
- Letter regarding MIT's efforts in combating sexual harassment
Posted: 30 Nov 2017 09:00 PM PST
Plasmas, gas-like collections of ions and electrons, make up an estimated 99 percent of the visible matter in the universe, including the sun, the stars, and the gaseous medium that permeates the space in between. Most of these plasmas, including the solar wind that constantly flows out from the sun and sweeps through the solar system, exist in a turbulent state. How this turbulence works remains a mystery; it's one of the most dynamic research areas in plasma physics.
Now, two researchers have proposed a new model to explain these dynamic turbulent processes.
The findings, by Nuno Loureiro, an associate professor of nuclear science and engineering and of physics at MIT, and Stanislav Boldyrev, a professor of physics at the University of Wisconsin at Madison, are reported today in the Astrophysical Journal. The paper is the third in a series this year explaining key aspects of how these turbulent collections of charged particles behave.
"Naturally occurring plasmas in space and astrophysical environments are threaded by magnetic fields and exist in a turbulent state," Loureiro says. "That is, their structure is highly disordered at all scales: If you zoom in to look more and more closely at the wisps and eddies that make up these materials, you'll see similar signs of disordered structure at every size level." And while turbulence is a common and widely studied phenomenon that occurs in all kinds of fluids, the turbulence that happens in plasmas is more difficult to predict because of the added factors of electrical currents and magnetic fields.
"Magnetized plasma turbulence is fascinatingly complex and remarkably challenging," he says.
Simulation conducted by MIT student Daniel Groselj.
Magnetic reconnection is a complicated phenomenon that Loureiro has been studying in detail for more than a decade. To explain the process, he gives a well-studied example: "If you watch a video of a solar flare" as it arches outward and then collapses back onto the sun's surface, "that's magnetic reconnection in action. It's something that happens on the surface of the sun that leads to explosive releases of energy." Loureiro's understanding of this process of magnetic reconnection has provided the basis for the new analysis that can now explain some aspects of turbulence in plasmas.
Loureiro and Boldyrev found that magnetic reconnection must play a crucial role in the dynamics of plasma turbulence, an insight that they say fundamentally changes the understanding of the dynamics and properties of space and astrophysical plasmas and "is indeed a conceptual shift in how one thinks about turbulence," Loureiro says.
Existing hypotheses about the dynamics of plasma turbulence "can correctly predict some aspects of what is observed," he says, but they "lead to inconsistencies."
Loureiro worked with Boldyrev, a leading theorist on plasma turbulence, and the two realized "we can fix this by essentially merging the existing theoretical descriptions of turbulence and magnetic reconnection," Loureiro explains. As a result, "the picture of turbulence gets conceptually modified and leads to results that more closely match what has been observed by satellites that monitor the solar wind, and many numerical simulations."
Loureiro hastens to add that these results do not prove that the model is correct, but show that it is consistent with existing data. "Further research is definitely needed," Loureiro says. "The theory makes specific, testable predictions, but these are difficult to check with current simulations and observations."
He adds, "The theory is quite universal, which increases the possibilities for direct tests." For example, there is some hope that a new NASA mission, the Parker Solar Probe, which is planned for launch next year and will be observing the sun's corona (the hot ring of plasma around the sun that is only visible from Earth during a total eclipse), could provide the needed evidence. That probe, Loureiro says, will be going closer to the sun than any previous spacecraft, and it should provide the most accurate data on turbulence in the corona so far.
Collecting this information is well worth the effort, Loureiro says: "Turbulence plays a critical role in a variety of astrophysical phenomena," including the flows of matter in the core of planets and stars that generate magnetic fields via a dynamo effect, the transport of material in accretion disks around massive central objects such as black holes, the heating of stellar coronae and winds (the gases constantly blown away from the surfaces of stars), and the generation of structures in the interstellar medium that fills the vast spaces between the stars. "A solid understanding of how turbulence works in a plasma is key to solving these longstanding problems," he says.
"This important study represents a significant step forward toward a deeper physical understanding of magnetized plasma turbulence," says Dmitri Uzdensky, an associate professor of physics at the University of Colorado, who was not involved in this work. "By elucidating deep connections and interactions between two ubiquitous and fundamental plasma processes — magnetohydrodynamic turbulence and magnetic reconnection — this analysis changes our theoretical picture of how the energy of turbulent plasma motions cascades from large down to small scales."
He adds, "This work builds on a previous pioneering study published by these authors earlier this year and extends it into a broader realm of collisionless plasmas. This makes the resulting theory directly applicable to more realistic plasma environments found in nature. At the same time, this paper leads to new tantalizing questions about plasma turbulence and reconnection and thus opens new directions of research, hence stimulating future research efforts in space physics and plasma astrophysics."
The research was supported by a CAREER award from the National Science Foundation and the U.S. Department of Energy through the Partnership in Basic Plasma Science and Engineering.
Posted: 30 Nov 2017 08:59 PM PST
On a recent Monday morning, Vivienne Sze, an associate professor of electrical engineering and computer science at MIT, spoke with enthusiasm about network architecture design. Her students nodded slowly, as if on the verge of comprehension. When the material clicked, the nods grew in speed and confidence. "Everything crystal clear?" she asked with a brief pause and a return nod before diving back in.
This new course, 6.S082/6.888 (Hardware Architecture for Deep Learning), is modest in size — capped at 25 for now — compared to the bursting lecture halls characteristic of other MIT classes focused on machine learning and artificial intelligence. But this course is a little different. With a long list of prerequisites and a heavy base of assumed knowledge, students are jumping into deep water quickly. They blaze through algorithmic design in a few weeks, cover the terrain of computer hardware design in a similar period, then get down to the real work: how to think about making these two fields work together.
The goal of the class is to teach students the interplay between two traditionally separate disciplines, Sze says. "How can you write algorithms that map well onto hardware so they can run faster? And how can you design hardware to better support the algorithm?" she asks rhetorically. "It's one thing to design algorithms, but to deploy them in the real world you have to consider speed and energy consumption."
"We are beginning to see tremendous student interest in the hardware side of deep learning," says Joel Emer, who co-teaches the course with Sze. A professor of the practice in MIT's Department of Electrical Engineering and Computer Science, and a senior distinguished research scientist at the chip manufacturer NVidia, Emer has partnered with Sze before. Together they wrote a journal article that provides a comprehensive tutorial and survey coverage of recent advances toward enabling efficient processing of deep neural networks. It is used as the main reference for the course.
In 2016, their group unveiled a new, energy-efficient computer chip optimized for neural networks, which could enable powerful artificial-intelligence systems to run locally on mobile devices. The groundbreaking chip, called "Eyeriss," could also help usher in the internet of things.
"I've been in this field for more than four decades. I've never seen an area with so much excitement and promise in all that time," Emer says. "The opportunity to have an original impact through building important and specialized architecture is larger than anything I've seen before."
Hardware at the heart of deep learning
Deep learning is a new name for an approach to artificial intelligence called neural networks, a means of doing machine learning in which a computer learns to perform some tasks by analyzing training examples. Today, popular applications of deep learning are everywhere, Emer says. The technique drives image recognition, self-driving cars, medical image analysis, surveillance and transportation systems, and language translation, for instance.
The value of the hardware at the heart of deep learning is often overlooked, says Emer. Practical and efficient neural networks, which computer scientists have researched off and on for 60 years, were infeasible without hardware to support deep learning algorithms. "Many AI accomplishments were made possible because of advances in hardware," he says. "Hardware is the foundation of everything you can do in software."
Deep learning techniques are evolving very rapidly, Emer says. "There is a direct need for this sort of hardware. Some of the students coming out of the class might be able to contribute to that hardware revolution."
Meanwhile, traditional software companies like Google and Microsoft are taking notice, and investing in more custom hardware to speed up the processing for deep learning, according to Sze.
"People are recognizing the importance of having efficient hardware to support deep learning," she says. "And specialized hardware to drive the research forward. One of the greatest limitations of progress in deep learning is the amount of computation available."
New hardware architectures
Real-world deployment is key for Skanda Koppula, a graduate student in electrical engineering and computer science. He is a member of the MIT Formula SAE Race Car Electronics Team.
"We plan to apply some of these ideas in building the perception systems for a driverless Formula student race car," he says. "And in the longer term, I see myself working toward a doctorate in related fields."
Valerie Sarge, also a graduate student in electrical engineering and computer science, is taking the course in prepration for a career that involves creating hardware for machine learning applications.
"Deep learning is a quickly growing field, and better hardware architectures have the potential to make a big impact on researchers' ability to effectively train networks," she says. "Through this class, I'm gaining some of the skills I need to contribute to designing these architectures."
Posted: 30 Nov 2017 02:55 PM PST
A team of 25 students, faculty, and staff from across the Institute has been working to develop a new design subject with an ambitious goal: to propose and rigorously evaluate potential new curriculum, pedagogies, and policies to advance the educational experience of first-year undergraduate students at MIT.
The subject, Designing the First Year at MIT, will be offered in spring 2018 for MIT undergraduate and graduate students, as well as Harvard University graduate students. "The class will be tasked with leading a community-based design effort that conceptualizes the first year experience (FYE) as a complex system," says Ian A. Waitz, vice chancellor of the Institute. "It's a neat opportunity to shape education at MIT, while learning about design methodologies and applying them to a very challenging problem. The subject is being developed and taught by faculty and staff from all five schools. That may be a first."
The idea bubbled up during conversations last spring between Waitz (then dean of the School of Engineering) and a group of students from the MIT Undergraduate Association (UA). Both he and the students share an interest in improving the undergraduate learning experience.
Enhancing the first-year experience was among the initial charges identified by Chancellor Cynthia Barnhart when she created the Office of the Vice Chancellor in July. Since then, Waitz has met with hundreds of faculty, administrators, and staff to seek advice on the best approach. In each session, he always asks two questions: What are the objectives of the first year? How well are we meeting them? Such intelligence is helping to guide the development of the subject in real-time.
"The course is a really interesting and effective way of getting the entire community involved in this conversation," says junior Alexa Martin, UA vice president. "It's more likely to result in actual change than surveys, presentations, and other things that have been done in the past."
Martin is one of four students on the team planning the subject, along with sophomore Kathryn Jiang, UA secretary; sophomore Noah McDaniel, who chairs the UA Committee on Education; and sophomore Edward Fan, who serves on the Institute's Committee on Curricula (CoC). This fall, they fanned out across campus to tell students about the class and ask their perceptions of the first year. What they heard revolved around a few themes, such as major exploration, first year advising, and preparing students for the second-year and beyond.
Because community engagement is fundamental for success, students in the class will continue to gather and analyze extensive input and data from stakeholders across campus. "We'd like to put the MIT culture to work — a meritocracy of great ideas," says Bruce Cameron, director of the System Architecture Lab and one of the subject instructors.
The other instructors are Bryan Moser, academic director and senior lecturer in the System Design and Management program; Maria Yang, an associate professor of mechanical engineering and engineering systems; Glen Urban, the David Austin Professor in Management (emeritus); and Justin Reich, an assistant professor in comparative media studies and writing.
Using a project-focused approach, the teaching team will lead weekly lectures on ways to think about, frame, research, and design solutions to the problem, along with lab-based workshops, readings, and field work. Groups of students will tackle different aspects of the first year, such as the GIRs or the residential experience, and along the way, develop skills in design, learning science, and communications.
"They've done an exceptional job of involving students in all of this," Martin says about the development of the subject. "I've felt like my voice has been heard along the entire way. In almost every decision they make, they're getting input from students, which is really nice to see and be a part of."
In that sense, the approach the planning team has taken bodes well for the desired outcome. "We hope to get a lot of 'user-centered innovation,'" Cameron says. "Who better to guide the first year experience into the 21st century than a group of MIT students who lived it?"
Waitz notes that having experience as an undergraduate at MIT, while valuable, is not required. "I hope graduate students will also take the course, as they are able to look back on their undergraduate experience, and also bring perspectives from other college programs."
Posted: 30 Nov 2017 02:51 PM PST
MIT issued the following statement today, following a fire near Central Square in Cambridge, Massachusetts. The four-alarm fire broke out at approximately 1:00 a.m. and damaged several buildings, including 22-24 Magazine Street. All residents were safely evacuated.
Posted: 30 Nov 2017 02:10 PM PST
Vivienne Sze, an associate professor of electrical engineering and computer science, was a member of the Joint Collaborative Team on Video Coding (JCT-VC), which developed the acclaimed High Efficiency Video Coding (HEVC) standard. For its work, the team received an Engineering Emmy Award during the Television Academy's recent 69th Engineering Emmy Awards ceremony in Hollywood.
In a statement about the award, the Television Academy said HEVC "has enabled efficient delivery in ultra-high-definition (UHD) content over multiple distribution channels."
"This new compression coding has been adopted, or selected for adoption, by all UHD television distribution channels, including terrestrial, satellite, cable, fiber, and wireless, as well as all UHD viewing devices, including traditional televisions, tablets, and mobile phones," the Academy stated.
The JCT-VC's award was one of seven Emmy's given to individuals, companies, or organizations for engineering innovations that significantly improve television transmission, recording, or reception.
"HEVC provides higher compression than previous standards," says Sze, who also co-edited a 2014 book on the subject. "At the same time, it can operate at the high processing speed necessary for UHD video and at the low power consumption necessary for mobile devices. It was such an honor for the whole team to receive an Emmy from the Television Academy."
The JCT-VC — which Sze describes as "a group of world-renowned video-coding experts" — consists of engineers from the Video Coding Experts Group (VCEG) of the International Telecommunication Union (ITU) and the Moving Pictures Expert Group (MPEG) of the International Organization for Standardization (ISO), as well as the International Electrotechnical Commission (IEC). She served as the primary coordinator of the team's core experiment on coefficient scanning and coding, and chaired ad hoc groups on topics related to entropy coding and parallel processing.
Sze received a bachelor's degree in electrical engineering from the University of Toronto and a master's degree and PhD from MIT. During her PhD studies, she worked on the design of energy-efficient video-coding hardware under the guidance of Anantha Chandrakasan, now Vannevar Bush Professor of Electrical Engineering and Computer Science and dean of the School of Engineering.
She soon realized that the video-coding algorithms limited the amount of energy reduction that could be achieved by the hardware. Accordingly, she started to investigate ways to jointly design the algorithms and hardware to improve the energy efficiency of next-generation video coding systems. She published her results at the 2011 International Solid-State Circuit Conference (ISSCC).
Toward the end of Sze's PhD work, she participated in VCEG meetings as the group was starting to discuss developing a new video-compression standard. After graduating, she joined the video coding team at Texas Instruments, which had sponsored her PhD research, and actively participated in the development of HEVC.
Once the HEVC standard was finalized, Sze joined the faculty of the Department of Electrical Engineering and Computer Science (EECS). She heads the Energy-Efficient Multimedia Systems Group at MIT's Research Laboratory of Electronics (RLE). Her research involves applying the algorithm and hardware co-design approach to a broad set of applications including machine learning, computer vision, robotics, image processing and, of course, video coding. Recent results include energy-efficient algorithms and hardware for deep learning and autonomous navigation for miniature drones. She is also co-teaching a new class at MIT that focuses on the co-design of algorithms and hardware for deep learning.
Her work has earned numerous other awards and honors. She received the EECS Jin-Au Kong Outstanding Doctoral Thesis Prize in 2011 for her thesis on "Parallel Algorithms and Architectures for Low-Power Video Decoding." She also received the 2017 Qualcomm Faculty Award, the 2016 Google Faculty Research Award, the 2016 Air Force Office of Scientific Research Young Investigator Research Program Award, the 2016 3M Non-Tenured Faculty Award, the 2014 DARPA Young Faculty Award, and the 2007 Design Automation Conference/ISSCC Student Design Contest Award. She was also a co-recipient of the 2016 MICRO Top Picks Award and the 2008 Asian-SSCC Outstanding Design Award.
Posted: 30 Nov 2017 10:00 AM PST
MIT's AIM Photonics Academy helped organize a gathering of more than 60 people at Stonehill College in Easton, Massachusetts, earlier this month to explore opportunities in integrated photonics, and discuss possibilities for a large investment to create a Lab for Education and Application Prototypes (LEAP) in integrated photonics there.
Attendees included representatives from companies, colleges, and universities, the Massachusetts Manufacturing Extension Program, the Massachusetts Technology Collaborative, and aides to U.S. Rep. Joseph P. Kennedy III.
Integrated photonics uses complex optical circuits to process and transmit signals of light, similar to the routing of electrical signals in a computer microchip. In contrast to the electrical transmission in a microchip, a photonic integrated circuit can transmit multiple information channels simultaneously using different wavelengths of light with minimal interference and energy loss to enable high-bandwidth, low-power communications.
"Students need to be prepared for the jobs that are coming," Cheryl Schnitzer, an associate professor of chemistry at Stonehill College, said at the Nov. 14 event. "It's our obligation to teach them about the exploding field of photonics and integrated photonics."
MIT's AIM Photonics Academy is the education and workforce development arm of the AIM Photonics Institute, one of 14 Manufacturing USA institutes launched as part of a federal initiative to revitalize American manufacturing. The federal government has committed $110 million to the AIM Photonics Institute over five years. At the same time, the Commonwealth of Massachusetts will spend $100 million on projects related to colleges and industry within the state, including $28 million to help launch AIM Photonics projects such as LEAP facilities.
MIT received funding for the first LEAP facility, with a focus on packaging. The MIT Lab for Education and Application Prototypes is currently housed in Building 35, and will relocate to the fifth floor of MIT.nano in June 2018.
A second LEAP site in its final stages of planning will be located at Worcester Polytechnic Institute, and will also serve nearby Quinsigamond Community College. AIM Photonics Academy and the Commonwealth of Massachusetts are also in discussions to build four more LEAP Labs, including one at Stonehill College, which would serve the southeastern corner of the state.
Once up and running, these labs will form a training network that helps Massachusetts become a major hub for photonics technology.
The meeting at Stonehill College, which also included the NextFlex Flexible Hybrid Electronics manufacturing innovation institute, generated many plans. The college has already connected with Bridgewater State and Bristol Community Colleges about creating photonic tracks in their programs. A team from AIM Photonics Academy, Stonehill College, and MassTech will begin visiting companies to follow up on how they might get engaged in a LEAP Lab at Stonehill.
Companies were enthusiastic about the opportunity to expand in these areas as well.
"Any time you add high-tech education to an area, you are going to incubate high-tech companies," noted John Lescinskas of Brockton Electro-Optics. "You're planting a seed. It can lead to a tree, or even a forest."
Because integrated photonics "is a technology that originated in Massachusetts, at MIT," said Lionel Kimerling, AIM Photonics Academy executive director and professor in the MIT Department of Materials Science and Engineering, the state is an optimal location for this initiative to take place.
"With the help of the state, Massachusetts can be the Silicon Valley for the growth of ultra-high performance communications systems using integrated photonics," Kimerling said.
Posted: 30 Nov 2017 08:30 AM PST
The following email was sent yesterday to the MIT community by President L. Rafael Reif.
To the members of the MIT community,
In the last several weeks, the nation has once again seen evidence that sexual harassment is pervasive. I am deeply disturbed by the revelations of misconduct elsewhere — and I know it also happens at MIT.
On this question, our community is not an oasis of safety. When it comes to sexual harassment, assault and related misconduct, a community like ours presents a particular set of risks: a 24/7 environment that brings together people across a broad range of ages, incomes and backgrounds, many of whom have power over others — power to make being at MIT miserable, power enough to make or break a career.
I want to use this moment of heightened attention to be clear about why this abuse of power is so disturbing in the context of our community — and to highlight what we must do and are doing about it. I expect we do not yet know the full extent of the problem at MIT. But the fact it exists here at all demands our serious attention.
The MIT community is built on collaboration and mutual respect. Sexual harassment is an act of aggression that belittles, unnerves and controls. It violates our fundamental expectations of respect and equality, and it violates the humanity of the person being harassed.
For many who suffer sexual harassment, the experience seriously damages their lives, their aspirations, their confidence and their careers. In some cases, the "remedy" can be damaging too. It grieves me to know that some of you reading this may have endured sexual misconduct at MIT, sought to take action and felt thwarted, silenced or ignored. As a community and as an administration, we must make sure that seeking help actually helps.
So, what are we doing?
In the end, the most important work is up to all of us. We need to actively build a culture that treats sexual harassment, coercion and assault as taboo, absurdly out of bounds — unthinkable for anyone, of any age, in any context. Let me now state the obvious. Most harassers are men. As a result, the men in our community must play a particularly important role in leading and driving the necessary change in culture.
Every member of our community is valuable, and harm to one is harm to all. As long as sexual harassment and assault persist in our community, we fail to live up to our shared potential and to fulfill our aspiration to make a better world.
* * *
If the problem seems daunting, we can take inspiration from two MIT giants we lost this year, President Emeritus Paul Gray and Institute Professor Millie Dresselhaus.
In the 1970s and '80s, Paul and Millie both saw that the MIT they lived and worked in should be better — more fair, more open and more welcoming to talent from every background — and they took deliberate, concerted, strategic action. Their leadership helped to reshape the MIT community — and helped deliver "the future" we inhabit today.
Their progress proves something important and hopeful: that in the life of a community, cultural change and moral growth are possible. Today, the responsibility to sustain that momentum falls to us. So I close with a challenge: that we each strive to define what we can do to invent a better MIT community for those who are here today, and for those who will follow us tomorrow.
I look forward to joining you in this vital work.
|You are subscribed to email updates from MIT News. |
To stop receiving these emails, you may unsubscribe now.
|Email delivery powered by Google|
|Google, 1600 Amphitheatre Parkway, Mountain View, CA 94043, United States|