The function; it is evaluated only inside the interval, never at its ends.
One end of the interval.
The other end.
The absolute width, in the argument's unit, the low is located to; positive.
The most evaluations.
The argument of the least value found and that value.
The argument at which a function of one variable is least on an interval, by Brent's method (Brent 1973, "Algorithms for Minimization without Derivatives", chapter 5,
localmin).Each step fits a parabola through the three best points found and takes its lowest point when that lies inside the bracket, moves less than half the step before last and does not come within the tolerance of an end; otherwise it takes a golden-section step into the larger part. The bracket always holds the least point found, so on a function with one low on the interval (a convex one, for instance) the search closes on it superlinearly where the function is smooth and never slower than golden section where it is not (a kink, a flat stretch).