id,node_id,number,title,user,state,locked,assignee,milestone,comments,created_at,updated_at,closed_at,author_association,active_lock_reason,draft,pull_request,body,reactions,performed_via_github_app,state_reason,repo,type 1024582327,I_kwDOAMm_X849EeK3,5861,Xarray's interpolator behavior compared to scipy with a numpy array: new keyword behavior requested,49281118,open,0,,,0,2021-10-12T23:08:34Z,2021-10-12T23:09:40Z,,NONE,,,,"I'm having trouble with the differing behavior between giving an xarray object to scipy's interpolating functions (particularly the RegularGridInterpolator and the one xarray's interpn is based on) versus giving a numpy array. When giving the interpolator a numpy array, I get a 1D array returned with one value for every point given. When an xarray object is given instead, I get an N dimensional array, as if a np.meshgrid statement is executed on the given points. I have provided more detail at the link below. This differing return behavior and the additional demand for the calculation for a grid made from the points (rather than the points themselves) is much slower than the numpy approach, but I can't use numpy arrays for medium data (because it won't all fit in my memory). Can a feature be added, maybe a 'numpy-like' keyword, to xarray's version of the scipy interpolator to only execute for the points given rather than a grid made from the points? Such a keyword would enable backwards-compatibility and reduce the computational demand for those interested in interpolating along a curved trajectory (such as in my case). Note: the same differing behavior occurs when I give scipy's RegularGridInterpolator an xarray object. https://github.com/scipy/scipy/issues/14824#issue-1021424672","{""url"": ""https://api.github.com/repos/pydata/xarray/issues/5861/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,,13221727,issue