@mmcky and @HumphreyYang , could you please do a first round review of this?

The aim is that swapping to NumPy's interp does not change results or degrade performance.

Note that this is related to QuantEcon/lecture-python.myst#387

Thanks @kp992.

The last dependency on interpolations is in amss

lectures/amss.md:from interpolation.splines import eval_linear, UCGrid, nodes

@kp992 any scope to be able to remove interpolations as a dependency -- otherwise maybe we need to solve EconForge/interpolation.py#110 re: generated_jit to make use of this package in amss

Thanks for the review @mmcky.

The last dependency on interpolations is in amss

Yeah, I checked that. Its the only one left with interpolation.py.

@kp992 any scope to be able to remove interpolations as a dependency -- otherwise maybe we need to solve EconForge/interpolation.py#110 re: generated_jit to make use of this package in amss

Hmm, I will check with some scipy functions and if possible remove it in a follow-up PR.

@kp992 the lecture opt_tax_recur is failing due to issues with the nelder_mead function from quantecon. I am not sure why this is now failing as we haven't upgraded numba in this PR.

@mmcky Thanks for pointing out the issue. The CI is green.

thanks @kp992

Just documenting this is now the only use case of interpolations left

lectures/amss.md:from interpolation.splines import eval_linear, UCGrid, nodes