From monitoring concentrations of greenhouse gases to detecting COVID in the breath, laser systems known as frequency combs can identify specific molecules as simple as carbon dioxide and as complex as monoclonal antibodies with unprecedented accuracy and sensitivity. Amazing as they are, however, frequency combs have been limited in how fast they can capture a high-speed process such as hypersonic propulsion or the folding of proteins into their final three-dimensional shapes.

Now, researchers at the National Institute of Standards and Technology (NIST), Toptica Photonics AG and the University of Colorado Boulder have developed a frequency comb system that can detect the presence of specific molecules in a sample every 20 nanoseconds, or billionths of a second. With this new capability, researchers can potentially use frequency combs to better understand the split-second intermediate steps in fast-moving processes ranging from the workings of hypersonic jet engines to the chemical reactions between enzymes that regulate cell growth. The research team announced its results in a paper published in Nature Photonics.

In their experiment, the researchers used the now-common dual-frequency comb setup, which contains two laser beams that work together to detect the spectrum of colors that a molecule absorbs. Most dual-frequency comb setups involve two femtosecond lasers, which send out a pair of ultrafast pulses in lockstep.

In this new experiment, the researchers used a simpler and cheaper setup known as “electro-optic combs,” in which a single continuous beam of light first gets split into two beams. Then, an electronic modulator produces electric fields that alter each light beam, shaping them into the individual “teeth” of a frequency comb. Each tooth is a specific color or frequency of light that can then be absorbed by a molecule of interest.

Whereas conventional frequency combs can have thousands or even millions of teeth, the researchers’ electro-optic comb only had 14 in a typical experimental run. However, as a result, each tooth had much higher optical power, and was far apart from others in frequency, resulting in a clear, strong signal that enabled the researchers to detect changes in the absorption of light at the 20-nanosecond time scale.

In their demonstration, the researchers used the instrument to measure supersonic pulses of CO₂ emerging from a small nozzle in an air-filled chamber. They measured the CO₂ mixing ratio, the proportion of carbon dioxide in the air. The changing concentration of CO₂ told researchers about the motion of the pulse. The researchers saw how the CO₂ interacted with the air and created oscillations of air pressure in its wake. Such details are often hard to accurately obtain even with the most sophisticated computer simulations.

“In a more complicated system like an aircraft engine we could use this approach to look at a particular species of interest, such as water or fuel or CO₂, to observe the chemistry. We can also use this approach to measure things such as pressure, temperature or velocity by looking at changes in the signal,” said NIST research chemist David Long. The information from these experiments could provide insights that could lead to design improvements in combustion engines, or a better understanding of how greenhouse gases interact with the atmosphere.

A special component in the setup, known as an optical parametric oscillator, was used to shift the comb teeth from the near-infrared to the mid-infrared colors absorbed by CO₂. But the optical parametric oscillator can also be tuned to other regions of the mid-infrared so that the combs can detect other molecules that absorb light in those regions.

The paper includes information that other researchers can use to build a similar system in the lab, making this new technique widely available across many research fields and industries.

“What is truly special about this work is that it substantially lowers the barrier to entry for researchers who would like to use frequency combs to study fast processes,” said co-author Greg Rieker, a professor at the University of Colorado Boulder and former NIST research associate.

“With this setup, you can generate any comb you want. The tunability, flexibility and speed of this method open the door to lots of different types of measurements,” Long said.

This work was supported in part by the Air Force Office of Scientific Research.

Choosing the best design for these components, known as deep neural network accelerators, is challenging because they can have an enormous range of design options. This difficult problem becomes even thornier when a designer seeks to add cryptographic operations to keep data safe from attackers.

Now, MIT researchers have developed a search engine that can efficiently identify optimal designs for deep neural network accelerators, that preserve data security while boosting performance.

Their search tool, known as SecureLoop, is designed to consider how the addition of data encryption and authentication measures will impact the performance and energy usage of the accelerator chip. An engineer could use this tool to obtain the optimal design of an accelerator tailored to their neural network and machine-learning task.

When compared to conventional scheduling techniques that don’t consider security, SecureLoop can improve performance of accelerator designs while keeping data protected.

Using SecureLoop could help a user improve the speed and performance of demanding AI applications, such as autonomous driving or medical image classification, while ensuring sensitive user data remains safe from some types of attacks.

“If you are interested in doing a computation where you are going to preserve the security of the data, the rules that we used before for finding the optimal design are now broken. So all of that optimization needs to be customized for this new, more complicated set of constraints. And that is what [lead author] Kyungmi has done in this paper,” says Joel Emer, an MIT professor of the practice in computer science and electrical engineering and co-author of a paper on SecureLoop.

Emer is joined on the paper by lead author Kyungmi Lee, an electrical engineering and computer science graduate student; Mengjia Yan, the Homer A. Burnell Career Development Assistant Professor of Electrical Engineering and Computer Science and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL); and senior author Anantha Chandrakasan, dean of the MIT School of Engineering and the Vannevar Bush Professor of Electrical Engineering and Computer Science. The research will be presented at the IEEE/ACM International Symposium on Microarchitecture.

“The community passively accepted that adding cryptographic operations to an accelerator will introduce overhead. They thought it would introduce only a small variance in the design trade-off space. But, this is a misconception. In fact, cryptographic operations can significantly distort the design space of energy-efficient accelerators. Kyungmi did a fantastic job identifying this issue,” Yan adds.

Secure acceleration

A deep neural network consists of many layers of interconnected nodes that process data. Typically, the output of one layer becomes the input of the next layer. Data are grouped into units called tiles for processing and transfer between off-chip memory and the accelerator. Each layer of the neural network can have its own data tiling configuration.

A deep neural network accelerator is a processor with an array of computational units that parallelizes operations, like multiplication, in each layer of the network. The accelerator schedule describes how data are moved and processed.

Since space on an accelerator chip is at a premium, most data are stored in off-chip memory and fetched by the accelerator when needed. But because data are stored off-chip, they are vulnerable to an attacker who could steal information or change some values, causing the neural network to malfunction.

“As a chip manufacturer, you can’t guarantee the security of external devices or the overall operating system,” Lee explains.

Manufacturers can protect data by adding authenticated encryption to the accelerator. Encryption scrambles the data using a secret key. Then authentication cuts the data into uniform chunks and assigns a cryptographic hash to each chunk of data, which is stored along with the data chunk in off-chip memory.

When the accelerator fetches an encrypted chunk of data, known as an authentication block, it uses a secret key to recover and verify the original data before processing it.

But the sizes of authentication blocks and tiles of data don’t match up, so there could be multiple tiles in one block, or a tile could be split between two blocks. The accelerator can’t arbitrarily grab a fraction of an authentication block, so it may end up grabbing extra data, which uses additional energy and slows down computation.

Plus, the accelerator still must run the cryptographic operation on each authentication block, adding even more computational cost.

An efficient search engine

With SecureLoop, the MIT researchers sought a method that could identify the fastest and most energy efficient accelerator schedule — one that minimizes the number of times the device needs to access off-chip memory to grab extra blocks of data because of encryption and authentication.

They began by augmenting an existing search engine Emer and his collaborators previously developed, called Timeloop. First, they added a model that could account for the additional computation needed for encryption and authentication.

Then, they reformulated the search problem into a simple mathematical expression, which enables SecureLoop to find the ideal authentical block size in a much more efficient manner than searching through all possible options.

“Depending on how you assign this block, the amount of unnecessary traffic might increase or decrease. If you assign the cryptographic block cleverly, then you can just fetch a small amount of additional data,” Lee says.

Finally, they incorporated a heuristic technique that ensures SecureLoop identifies a schedule which maximizes the performance of the entire deep neural network, rather than only a single layer.

At the end, the search engine outputs an accelerator schedule, which includes the data tiling strategy and the size of the authentication blocks, that provides the best possible speed and energy efficiency for a specific neural network.

“The design spaces for these accelerators are huge. What Kyungmi did was figure out some very pragmatic ways to make that search tractable so she could find good solutions without needing to exhaustively search the space,” says Emer.

When tested in a simulator, SecureLoop identified schedules that were up to 33.2 percent faster and exhibited 50.2 percent better energy delay product (a metric related to energy efficiency) than other methods that didn’t consider security.

The researchers also used SecureLoop to explore how the design space for accelerators changes when security is considered. They learned that allocating a bit more of the chip’s area for the cryptographic engine and sacrificing some space for on-chip memory can lead to better performance, Lee says.

In the future, the researchers want to use SecureLoop to find accelerator designs that are resilient to side-channel attacks, which occur when an attacker has access to physical hardware. For instance, an attacker could monitor the power consumption pattern of a device to obtain secret information, even if the data have been encrypted. They are also extending SecureLoop so it could be applied to other kinds of computation.

This work is funded, in part, by Samsung Electronics and the Korea Foundation for Advanced Studies.

The study, published today in Nature Climate Change, is the most up-to-date and comprehensive analysis of the global carbon budget. The carbon budget is an estimate of the amount of carbon dioxide emissions that can be emitted while keeping global warming below certain temperature limits.

The Paris Agreement aims to limit global temperature increase to well below 2°C above preindustrial levels and pursue efforts to limit it to 1.5°C. The remaining carbon budget is commonly used to assess global progress against these targets.

The new study estimates that for a 50% chance of limiting warming to 1.5°C, there are less than 250 gigatonnes of carbon dioxide left in the global carbon budget.

The researchers warn that if carbon dioxide emissions remain at 2022 levels of about 40 gigatonnes per year, the carbon budget will be exhausted by around 2029, committing the world to warming of 1.5°C above preindustrial levels.

The finding means the budget is less than previously calculated and has approximately halved since 2020 due to the continued increase of global greenhouse gas emissions, caused primarily from the burning of fossil fuels as well as an improved estimate of the cooling effect of aerosols, which are decreasing globally due to measures to improve air quality and reduce emissions.

Dr Robin Lamboll, research fellow at the Centre for Environmental Policy at Imperial College London, and the lead author of the study, said: “Our finding confirms what we already know — we’re not doing nearly enough to keep warming below 1.5°C.

“The remaining budget is now so small that minor changes in our understanding of the world can result in large proportional changes to the budget. However, estimates point to less than a decade of emissions at current levels.

“The lack of progress on emissions reduction means that we can be ever more certain that the window for keeping warming to safe levels is rapidly closing.”

Dr Joeri Rogelj, Director of Research at the Grantham Institute and Professor of Climate Science & Policy at the Centre for Environmental Policy at Imperial College London, said: “This carbon budget update is both expected and fully consistent with the latest UN Climate Report.

“That report from 2021 already highlighted that there was a one in three chance that the remaining carbon budget for 1.5°C could be as small as our study now reports.

“This shows the importance of not simply looking at central estimates, but also considering the uncertainty surrounding them.”

The study also found that the carbon budget for a 50% chance of limiting warming to 2°C is approximately 1,200 gigatonnes, meaning that if carbon dioxide emissions continue at current levels, the central 2°C budget will be exhausted by 2046.

There has been much uncertainty in calculating the remaining carbon budget, due to the influence of other factors, including warming from gasses other than carbon dioxide and the ongoing effects of emissions that are not accounted for in models.

The new researchused an updated dataset and improved climate modelling compared to other recent estimates, published in June, characterising these uncertainties and increasing confidence around the remaining carbon budget estimates.

The strengthened methodology also gave new insights into the importance of the potential responses of the climate system to achieving net zero.

‘Net zero’ refers to achieving an overall balance between global emissions produced and emissions removed from the atmosphere.

According to the modelling results in the study, there are still large uncertainties in the way various parts of the climate system will respond in the years just before net zero is achieved.

It is possible that the climate will continue warming due to effects such as melting ice, the release of methane, and changes in ocean circulation.

However, carbon sinks such as increased vegetation growth could also absorb large amounts of carbon dioxide leading to a cooling of global temperatures before net zero is achieved.

Dr Lamboll says these uncertainties further highlight the urgent need to rapidly cut emissions. “At this stage, our best guess is that the opposing warming and cooling will approximately cancel each other out after we reach net zero.

“However, it’s only when we only when we cut emissions and get closer to net zero that we will be able to see what the longer-term heating and cooling adjustments will look like.

“Every fraction of a degree of warming will make life harder for people and ecosystems. This study is yet another warning from the scientific community. Now it is up to governments to act.”

Now, researchers from McGill University, UCLA and Princeton have found in a new study an inexpensive, sustainable alternative to mechanical cooling with refrigerants in hot and arid climates, and a way to mitigate dangerous heat waves during electricity blackouts.

The researchers set out to answer how to achieve a new benchmark in passive cooling inside naturally conditioned buildings in hot climates such as Southern California. They examined the use of roof materials that radiate heat into the cold universe, even under direct sunlight, and how to combine them with temperature-driven ventilation. These cool radiator materials and coatings are often used to stop roofs overheating. Researchers have also used them to improve heat rejection from chillers. But there is untapped potential for integrating them into architectural design more fully, so they can not only reject indoor heat to outer space in a passive way, but also drive regular and healthy air changes.

“We found we could maintain air temperatures several degrees below the prevailing ambient temperature, and several degrees more below a reference ‘gold standard’ for passive cooling,” said Remy Fortin, lead author and PhD candidate at the Peter Guo-hua Fu School of Architecture “We did this without sacrificing healthy ventilation air changes.” This was a considerable challenge, considering air exchanges are a source of heating when the aim is to keep a room cooler than the exterior.

The researchers hope the findings will be used to positively impact communities suffering from dangerous climate heating and heat waves. “We hope that materials scientists, architects, and engineers will be interested in these results, and that our work will inspire more holistic thinking for how to integrate breakthroughs in radiative cooling materials with simple but effective architectural solutions,” said Salmaan Craig, Principal Investigator for the project and Assistant Professor at the Peter Guo-hua Fu School of Architecture.