|

Loading ...
|
|
|
|
pages views since 05/19/2016 : 142924
· Members : 7
· News : 806
· Downloads : 0
· Links : 0
|
|
|
|
AI Innovation Solves Supercomputer Math on Desktop Computers in a Matter of Seconds
|
|
|
Posted by Okachinepa on 01/07/2025 @


Courtesy of SynEvol
Credit: Johns Hopkins University
The ability to model complicated systems, such as how vehicles deform in collisions, how spacecraft handle harsh circumstances, or how bridges bear stress, at thousands of times faster than previously feasible is being made possible by a breakthrough in artificial intelligence. Massive mathematical problems that formerly required the power of supercomputers can now be solved by personal computers because to this invention.
The new AI framework provides a flexible and effective way to forecast answers to difficult mathematical problems. These formulas are essential for simulating phenomena that are frequently seen in engineering and design tests, such as fluid flow or the behavior of electrical current in different geometries.
The system, known as DIMON (Diffeomorphic Mapping Operator Learning), resolves partial differential equations, which are common mathematical problems found in almost all engineering and scientific study. Researchers can convert real-world systems or processes into mathematical depictions of how environments or items will change across time and space by using these equations.
"This is a solution that we think will generally have a massive impact on various fields of engineering because it's very generic and scalable," said Natalia Trayanova, a professor of biomedical engineering and medicine at Johns Hopkins University who co-led the study, although the idea to develop it came from our own work. "It can essentially solve partial differential equations on multiple geometries in any problem, in any field of science or engineering, such as in crash testing, orthopedics research, or other complex problems where shapes, forces, and materials change."
Trayanova's team evaluated the new AI on more than 1,000 cardiac "digital twins," which are extremely precise computer models of actual patients' hearts, in addition to showcasing DIMON's suitability for resolving other engineering issues. The platform achieved great prognosis accuracy by predicting the propagation of electrical signals through each distinct heart shape.
In order to investigate cardiac arrhythmia—an electrical impulse disturbance in the heart that results in irregular beating—Trayanova's team employs partial differential equations. Researchers can determine whether patients are at risk of developing the frequently fatal illness and provide treatment options using their cardiac digital twins.
Trayanova, who leads the Johns Hopkins Alliance for Cardiovascular Diagnostic and Treatment Innovation, stated, "We're bringing novel technology into the clinic, but a lot of our solutions are so slow that it takes us about a week from when we scan a patient's heart and solve the partial differential equations to predict if the patient is at high risk for sudden cardiac death and what is the best treatment plan." The speed at which we can find a solution with this new AI method is astounding. By using a desktop computer instead of a supercomputer, the time required to compute the prediction of a heart digital twin will drop from many hours to 30 seconds, enabling us to make it a component of the regular clinical process.
In order to solve partial differential equations, complicated objects, such as body organs or airplane wings, are typically divided into grids or meshes composed of tiny components. Each simple piece's difficulty is then resolved and reassembled. However, the grids must be updated and the solutions computed if their shapes change, as in crashes or deformations, which can be costly and computationally slow.
This issue is resolved by DIMON, which uses AI to comprehend the behavior of physical systems in a variety of shapes without requiring a complete recalculation for every new shape. The AI is far quicker and more effective at activities like designing or modeling shape-specific scenarios because it uses learned patterns to forecast how variables like heat, stress, or motion will react rather than breaking forms into grids and repeatedly calculating equations.
The group is adding cardiac pathology that causes arrhythmia to the DIMON framework. According to Minglang Yin, a Johns Hopkins Biomedical Engineering Postdoctoral Fellow who created the platform, the technology's adaptability allows it to be used for shape optimization and numerous other engineering activities where solving partial differential equations on novel shapes is frequently required.
"DIMON maps the solution to several new shapes after first solving the partial differential equations on a single shape for each problem. Its amazing adaptability is highlighted by its shape-shifting capacity, according to Yin. "We are eager to use it to solve a variety of issues and to make it available to the larger community so they can expedite their engineering design solutions."
|
|
|