New optimisation trick speeds boundary-layer mathematics, with trade-offs

Air moving over a wing forms a thin layer that affects drag and safety.

Scientists use difficult equations to understand this layer.

Ziya Uddin created a method called PI-OHAM to solve these equations more quickly.

It combines an older mathematical technique with computer-based optimisation.

The computer chooses settings that make the proposed answer obey the full physics problem.

Unlike some neural networks, the steps of the answer can still be inspected mathematically.

In a test involving the Blasius equation, PI-OHAM was much faster than two comparison methods.

It was about as accurate as a physics-informed neural network in the reported comparison.

However, the older HAM method was slightly more accurate in the most demanding test, so the new method is faster but not always more precise.