Simulation of the real world, particularly in fields like automotive and aerospace design, has always been a challenging task, often hampered by the limitations of traditional computational techniques. However, Beyond Math, a startup led by co-founder Darren Garvey, is taking on this challenge with a groundbreaking approach that leverages AI to dramatically accelerate the simulation process, potentially saving design teams days or even weeks of waiting.
Beyond Math is initially focusing on computational fluid dynamics (CFD), a complex field that involves simulating how objects interact with air or water. CFD has traditionally required massive computational power, often limiting it to supercomputers and GPU clusters. This has made the design process for cars, planes, and boats a time-consuming endeavor, with significant wait times between iterations.
“For designers, the typical process involves running a simulation, waiting overnight for results, and then refining their designs based on those results,” explained Garvey. “This iterative loop can be time-consuming, and even after simulations, the results might not match real-world tests, necessitating further adjustments.”
Beyond Math's AI-powered approach aims to change this by shortening the feedback loop between design iterations. Instead of relying solely on traditional simulations, the company’s first product, a "digital wind tunnel," uses machine learning to simulate airflow over complex surfaces in near-real time. This allows designers to make rapid adjustments and explore far more design variations within the same timeframe.
The innovation lies in Beyond Math’s method of training their AI models. Unlike typical machine learning models that rely on large datasets, Beyond Math focuses on understanding the underlying physics and real-world data. “We’re not just approximating simulations—we’re approximating the real world,” Garvey noted. This approach enables their models to provide insights that are more aligned with real-world outcomes, making the tool incredibly valuable for designers.
One of Beyond Math's early markets is Formula 1 racing, where teams are exploring the software to enhance their vehicle design processes. “F1 teams are some of the heaviest users of CFD, and they’re always on the lookout for technologies that can give them an edge,” Garvey said. “We’re working closely with a couple of teams, and we’re optimistic that our platform will soon demonstrate tangible performance improvements.”
The startup recently secured $8.5 million in seed funding, led by UP.Partners, with participation from Insight Partners and InMotion Ventures. This funding will enable Beyond Math to double its team size and scale up its computing capabilities, including acquiring Nvidia DGX 200 systems.
While the current focus is on top-tier customers like F1 teams, Garvey envisions expanding Beyond Math’s technology to other industries. “Our immediate goal is to refine our models and demonstrate their value in specific domains like automotive and aerospace. However, we’re aware that different fields, such as medical simulations, will require further adaptation of our models.”
In the coming months, Beyond Math aims to transition from research and proofs of concept to delivering real-world impact, potentially transforming the way design teams approach complex simulations across various industries.
