This is beautiful .....absolutely beautiful Front wing optimized for a specific track ...WOW. All........... that I have been studying starting from Partial differential equation to variational calculus to CFD to Turbo-machinery to Fluid structure interaction and mesh morphing to Aircraft aerodynamics to Multi-objective design optimization , everything coming together to make that beautiful front wing ......no words
You're correct - it can't actually be a true "global" optimum, as the adjoint method calculates gradients and progresses along these - so if you're in a local convex hull, which doesn't contain the global optimum, you wont find it. If your question was more related to how to progress to the ("global") of the convex hull you're in - after the adjoint solver provides the sensitivities / gradients, you simply choose a step size (based on some method - line search or fixed size) and advance along the that gradient vector.
This is beautiful .....absolutely beautiful
Front wing optimized for a specific track ...WOW.
All........... that I have been studying starting from Partial differential equation to variational calculus to CFD to Turbo-machinery to Fluid structure interaction and mesh morphing to Aircraft aerodynamics to Multi-objective design optimization , everything coming together to make that beautiful front wing ......no words
Glad you're impressed!
So beautiful video! The speaker resembles Messi a bit.😀
Great video ! However, i dont really understand how at 16:50 we progress to the global optimum ? Is that possible ?
Did you find the answer to your question?
You're correct - it can't actually be a true "global" optimum, as the adjoint method calculates gradients and progresses along these - so if you're in a local convex hull, which doesn't contain the global optimum, you wont find it. If your question was more related to how to progress to the ("global") of the convex hull you're in - after the adjoint solver provides the sensitivities / gradients, you simply choose a step size (based on some method - line search or fixed size) and advance along the that gradient vector.