FLINT3.7.0-dev

arb_pow

Synopsis#include <flint/arb.h>
void arb_pow(arb_t z, const arb_t x, const arb_t y, slong prec)source

Sets z to xyx^y: for an exact integer y by binary exponentiation (small exponents) or as exp⁡(ylog⁡x)\exp(y \log x), with xy=±(−x)yx^y = \pm(-x)^y for negative xx; as (x1/2)2y(x^{1/2})^{2y} if y is a small exact half-integer; as exp⁡(ylog⁡x)\exp(y \log x) otherwise (exact arguments in one evaluation). For a wide base and an exact integer exponent, the result is the hull of the range of xyx^y over the ball, from the powers of its endpoints, which is tighter than the ball arithmetic of binary exponentiation (the base is wide when its radius exceeds ∣m∣/(4∣y∣)|m| / (4 |y|) for the midpoint mm, or when it contains 0).

x0=1x^0 = 1 for every xx, and 0y0^y is 0 for positive yy and indeterminate otherwise. A ball containing 0 raised to a negative integer power is indeterminate. For a ball xx containing 0 and positive numbers and a nonnegative y (containing no integer if xx also contains negative numbers, whose integer powers are real), the result is the finite ball [u/2±u/2][u/2 \pm u/2] with u≥Ryu \ge R^y over the ball for the upper endpoint RR of xx (the range of xyx^y over the nonnegative part of xx, where it is real), rather than an indeterminate value. Otherwise a base containing nonpositive numbers gives an indeterminate result unless yy is an exact integer.

Guides#

Cite this page

The FLINT team, “arb_pow (arb.h),” FLINT: Fast Library for Number Theory, documentation, version 3.7.0-dev, 2026. https://flintlib.org/doc/arb/arb_pow.html (generated 2026-11-20).