Passing data to function

[fintzij]

Hello,

I have a function that I would like to integrate repeatedly but which depends on data whose values might change. Is there a way to specify a function in Quadrature.jl such that I don't need to create a new QuadratureProblem each time the values of the data change?

For example, is there a way to pass f(x, p, d) = sum(sin.(x * p + d)) so that I could change the value of d in an outer loop and just call solve() instead of creating a new QuadratureProblem each time?

Apologies in advance if this is obvious and there is something trivial I am overlooking.

Thanks,
Jon

[ChrisRackauckas]

So d is the parameter? Why not make the parameters p?

[fintzij]

So, for a bit more context, this is part of a statistical package. d are data in the sense that the values are known, not estimated, like covariates. p are parameters that are being estimated, so, e.g., we would want autodiff to be taking gradients with respect to p but not d. Zooming out a bit, d might be a row in a dataset that we want to pass as an argument to f()

[fintzij]

One more point is that we'd like to handle the data with some abstraction so that we have a general modeling framework. Obviously, it is trivial to hard code the values when they're known ahead of time.

[ChrisRackauckas]

You can use a closure then. f(x, p) = sum(sin.(x * p + d)) just enclose