Dear SciPy developers, the section 'Notes' of documentation page on Brents's minimization method ( https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.brent.ht... ) states that it ''Uses inverse parabolic interpolation when possible to speed up convergence of golden section method". I belive this statement is not correct. There are actually two Brent's methods: one is to find a root (i.e., zero) of a function, and another one is to search a minima of a function (see a description in https://maths-people.anu.edu.au/~brent/pd/rpb011i.pdf  ). According to the original description, Brent's optimization method is a combination of golden section search and successive parabolic interpolation (see, e.g., page 79, Section 8 of https://maths-people.anu.edu.au/~brent/pd/rpb011i.pdf ) As far as I know, inverse parabolic interpolation is used to find a root of a function. Because Brent's optimization method is a derivative-free method, the inverse parabolic interpolation can not be used 'under-the-hood' of Brent's method  to find a minima / maxima of a function. Regards, Oleg V. Zolotov ⁣Получить BlueMail для Android ​