(11C) Gaussian integration

[Paul Dale]

(06-19-2014 03:13 PM)Dieter Wrote:  The modified Romberg algorithm in the 34C will reliably handle cases that will not work well with the Gaussian method. The WP34s originally used a (much more sophisticated) Gauss quadrature algorithm. This was changed to Romberg's method because of the mentioned reasons.

I would have liked to do an adaptive algorithm that used a Gauss-Kronrod quadrature for the sub steps. The GSL integration uses a variety of Gauss-Kronrod quadratures of increasing point count (but reusing the same points).


- Pauli