Supporting the future of Mars exploration with supercomputers – EurekAlert

You may have flown a flight simulator in a computer game or at a science museum. Landing without crashing is always the hardest part. But thats nothing compared to the challenge that engineers are facing to develop a flight simulation of the very large vehicles necessary for humans to explore the surface of Mars. The Red Planet poses innumerable challenges to astronauts, not the least of which is getting there. Thats where the Department of Energy Office of Sciences user facility supercomputers come in. Researchers at DOEs Oak Ridge Leadership Computing Facility (OLCF) are working with NASA engineers and scientists tosimulate the process of slowing down a huge spacecraft as it moves towards Mars surface.

Landing spacecraft on Mars isnt new to NASA. The agency ran its first missions to the planet in 1976 with the Viking project. Since then, NASA has successfully carried out eight additional Mars landings.

What makes this goal different is the fact that its much more difficult to land the huge spacecraft required for human exploration than those for robotic missions. The robotic vehicles use parachutes to decelerate through Mars atmosphere. But a spacecraft carrying humans will be about 20 to 50 times heavier. A vehicle this large simply cant use parachutes. Instead, NASA will need to rely on retro-propulsion. This technology uses rockets that fire forwards to slow down the vehicle as it approaches the surface.

A number of challenges come with using retropropulsion. The high-energy rocket engine exhaust interacts with both the vehicle and the Martian atmosphere. Those dynamics change how the team needs to guide and control the vehicle. In addition, engineers cant fully replicate how a flight on Mars would go on Earth. While they can test spacecraft in wind tunnels and use other tools, those tools arent a perfect replacement or direct analog for the Martian environment.

To fill in the gaps, NASA turned to the OLCF supercomputers and their expert computer scientists. In theory, programs running on supercomputers could fully simulate the Martian environment and many of the complex physics associated with using retropropulsion.

The project team has relied on FUN3D, a long-standing suite of software tools that models how fluids including air move. Engineers created the first version of the code in the late 1980s and have continually made major improvements since then. Agencies and companies in aeronautics and space technology have used it to tackle major challenges.

The current Mars effort began in 2019 on Summit, OLCFs fastest computer at the time. The initial simulations assumed fixed conditions. They simulated just one point along the vehicles trajectory. Those early versions allowed scientists to evaluate the impacts of flight speeds, engine settings, and more. Further developments enabled engineers to explore real gas effects. They could account for the liquid oxygen-methane rocket engines and the carbon dioxide-heavy Martian atmosphere. Even these early simulations typically resulted in petabyte-sized datasets. It would take about 1,000 powerful home computers to store a single petabyte. But even these werent full simulations that wasnt possible yet.

The next step was to incorporate a whole new piece of software into the simulation the Program to Optimize Simulated Trajectories (POST2). NASA developed POST2 to analyze flight mechanics for a broad range of applications. While initial simulations relied on static conditions, POST2 allowed scientists to dynamically fly the vehicle in the simulation. The team engaged researchers from Georgia Techs Aerospace Systems Design Laboratory. They had previously developed unique strategies to couple POST2 with high-fidelity aerodynamic simulations. Incorporating POST2 also required engineers to change the project workflow. The softwares use was restricted to NASA computing systems for security reasons. As such, the team needed to ensure the NASA systems could communicate smoothly with Summit at OLCF. Resolving issues with firewalls, network interruptions, and other programs required a full year of planning for the cybersecurity and system administration teams at both facilities!

The latest advance involved moving the entire simulation over to the newest and most powerful computer at OLCF Frontier. The firstexascale computerin the world, Frontier is massively more powerful than previous supercomputers. With a series of coordinated runs over a two-week period, the team ran its most elaborate flight simulation to date. It was a 35-second closed-loop descent from 5 miles altitude to approximately 0.6 miles. The simulation slowed the vehicle from 1,200 miles per hour to approximately 450 miles per hour. POST2 was able to autonomously control the vehicle in a stable fashion using its eight main engines and four reaction control system modules.

With the immense power provided by Frontier at OLCF, NASA engineers are moving forward to tackle new frontiers in space travel.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.

See the original post:

Supporting the future of Mars exploration with supercomputers - EurekAlert

The Future of Neuroscience and AI: DeepSouth Supercomputer – Medriva

Stepping into the Future with DeepSouth

The world is on the brink of a significant breakthrough with the development of DeepSouth, the first-ever supercomputer capable of simulating the entire human brain. This trailblazing project, under the leadership of Western Sydney University in Australia, is set to revolutionize neuroscience and artificial intelligence as we know it. With an operational launch date set for 2024, DeepSouth is poised to unlock a myriad of opportunities in medicine, technology, and beyond.

DeepSouth isnt just another supercomputer it is a powerhouse that matches the human brains estimated rate of operations. With the ability to perform 228 trillion synaptic operations per second, it is set to redefine what technology can achieve. Its not just about the staggering numbers; its about how DeepSouth uses this power. The supercomputer is designed to process colossal amounts of information with minimal power, emulating the efficiency of the human brain.

What sets DeepSouth apart is its neuromorphic system. Unlike traditional computer systems, the neuromorphic system is designed to operate like networks of neurons. This unique configuration allows the supercomputer to emulate large networks of spiking neurons efficiently, requiring less power and enabling greater efficiencies. This ingenious design is what makes the simulation of the entire human brain possible.

DeepSouth is expected to be a game-changer for neuroscience and AI. By simulating the human brain, it will provide unprecedented insights into how our brains process information so efficiently. These findings could lead to breakthroughs in understanding both healthy and diseased human brains, opening new avenues for medical research and treatment. The supercomputer will also be an invaluable resource for researchers looking to prototype new engineering solutions in the AI space.

While the implications for neuroscience are profound, the potential applications of DeepSouth dont stop there. The project could lead to advances in smart devices and sensors, further enhancing the technology that forms an integral part of our lives. With AI applications becoming increasingly prevalent, the insights derived from DeepSouths brain simulations could pave the way for more sophisticated AI systems, pushing the boundaries of what technology can achieve.

As we look forward to DeepSouths operational launch in 2024, its clear that were on the cusp of a new era in technology and neuroscience. This supercomputer isnt just simulating the human brain; its unlocking the potential to understand our brains better, develop more advanced AI, and create smarter technology. The future of neuroscience and AI is bright, and DeepSouth is leading the charge.

Read the original post:

The Future of Neuroscience and AI: DeepSouth Supercomputer - Medriva