In scipy, the function scipy.interpolate.make_interp_spline() can be batched since its x argument must be one-dimensional with shape (m,) and its y argument can have shape (m, ...). However, the function scipy.interpolate.make_smoothing_spline() only accepts a y argument of shape (m,). Is there a simple way to batch the behavior of make_smoothing_spline() so it has the same behavior as make_interp_spline()? I was thinking of using numpy.vectorize(), but here I'm not batching operations on an array, I need a single function as output. I guess I could just implement a loop and make a nested list of splines, but I was wondering if there would be a neater way. Probably some combination of decorators but I'm twisting my brain in knots... EDIT: Developers seem to be aware of this issue here. Continue reading...
