Quadrature routines for pre-sampled data

[IlianPihlajamaa]

Hello,

thanks for a great package!

I don't know if this is the right place to ask this question, but:

Are there plans for SciML to offer quadrature routines for pre-sampled data (where one either does not have the luxury of choosing the grid points at all, or at least not dynamically/adaptively? Right now this functionality is scattered throughout the Julia ecosystem (see eg NumericalIntegration.jl, Romberg.jl, Trapz.jl), even though I think this is a fairly common use case. For reference, there is an excellent python package QuadPy, that inspiration might be drawn from, even though it does not offer functionality for arbitrary grids. SciPy of course does offer that.

I am asking this, because right now, I am typically rolling my own (eg irregular Simpson's) rules, and I thought that if many people do the same, it might be a waste of effort.

[ChrisRackauckas]

I think it would be great to add these kinds of methods, even non-adaptive, to here. They are widely used methods to approximate integrals, so there is no reason to leave them out.

[IlianPihlajamaa]

Okay, I might try to do that (for some simple rules like trapezoidal/Simpson's). No promises when though. I may also need some guidance adhering to the code structure and API that this package, together with SciML, dictates.

Would you like me to close this issue in the meantime?

[ChrisRackauckas]

Let's leave this open until they are implemented.