Vincent Punabantu

icon location PhD in Mechanical Engineering student @ UCT (Univesity of Cape Town) | Machine Learning | Computational Fluid Dynamics | Ardunio & Electronics Enthusiast

Results 4 comments of Vincent Punabantu

Lack of convergence in Sobol indices and standard deviation using Gauss quadrature in chaospy

> The scale of `C` is many magnitudes smaller than `R1` and `R2`, making for bad condition number for the underlying methods. > > Try switching to [generalized polynomial chaos](https://chaospy.readthedocs.io/en/master/user_guide/advanced_topics/generalized_polynomial_chaos.html)...

Lack of convergence in Sobol indices and standard deviation using Gauss quadrature in chaospy

Forgot to attach the image to my earlier comment. Above are the plots that I was referring to. Thank you for your quick response earlier.

Lack of convergence in Sobol indices and standard deviation using Gauss quadrature in chaospy

+ The `gauss_evals` are generated using the nodes from `gauss_quads`. The simulations are run separately using OpenFOAM by evaluating the CFD model on the set of gauss nodes for polynomial...

Lack of convergence in Sobol indices and standard deviation using Gauss quadrature in chaospy

Hello Jonathan, Hope your week is off to a good start. Thank you for the suggestion. I tried running simulations for 3rd and 4th order for only 2 uncertain inputs...