scipy.optimize.fsolve-输出如何取决于初始猜测?

I do not understand why fsolve() output depends on initial guesses.

In a simple case to find roots of f(x) = x**2 - 100 which are [-10, 10] Could you please explain to me why and how the answer depends on initial guesses here, or fsolve(f,[-10,-8]) = [-10. -10.] and fsolve(f,[-10,-7]) = [-10. 10.] why is the difference there? is the answer the closest root to the guess numbers or ...?

enter image description here

谢谢。