Charles Karney

@pmlandwehr I know this goes beyond the question you asked; but you might find it useful... If the goal is to compute the closest distance of a point P to...

Indeed... The methodology needed is similar to that used for MATLAB.

I've made no progress on this, I'm afraid -- too many other irons in the fire. It might help to know in more-or-less concrete terms what you application is. How...

By the way, [pyproj](https://pyproj4.github.io/pyproj/stable) gives python access to distance calculations via the compiled C implementation. This will be faster that the native python implementation.

> I'm not sure about that. Your statement "users don't know what they want" is a double-way street. If they're not sure about the accuracy that is needed in their...

For sure, 5x or 10x speed-ups are enticing. But users cannot make a decent assessment of the trade-off without some rigorous accounting of the errors in the various methods. For...

@barendgehrels Of course, the definition of "reasonably precise" will depend on the application. However, in my applications, an error in the distance of 68 m for points 10000 km apart...

@danpat In many applications, quantities are "referred to" the reference ellipsoid. Thus latitude and longitudes are usually with reference to the ellipsoid and not the geoid; (otherwise they are the...

Transverse Mercator is also similarly accurate (within 3900 km of the central meridian). Probably other projections are if they can be expressed in closed form. Yes AEQD is the only...

I recommend providing some data on the accuracy of the Andoyer method (error in distance as a function of distance + similar information for the heading).