_gr_poly_sqrt_series, gr_poly_sqrt_series, _gr_poly_sqrt_series_basecase, gr_poly_sqrt_series_basecase, _gr_poly_sqrt_series_miller, gr_poly_sqrt_series_miller, _gr_poly_sqrt_series_newton, gr_poly_sqrt_series_newton, _gr_poly_rsqrt_series, gr_poly_rsqrt_series, _gr_poly_rsqrt_series_basecase, gr_poly_rsqrt_series_basecase, _gr_poly_rsqrt_series_miller, gr_poly_rsqrt_series_miller, _gr_poly_rsqrt_series_newton, gr_poly_rsqrt_series_newton
#include <flint/gr_poly.h>int _gr_poly_sqrt_series_basecase(gr_ptr res, gr_srcptr f, slong flen, slong len, gr_ctx_t ctx)sourceint _gr_poly_sqrt_series_newton(gr_ptr res, gr_srcptr f, slong flen, slong len, slong cutoff, gr_ctx_t ctx)sourceint gr_poly_sqrt_series_newton(gr_poly_t res, const gr_poly_t f, slong len, slong cutoff, gr_ctx_t ctx)sourceint _gr_poly_rsqrt_series_basecase(gr_ptr res, gr_srcptr f, slong flen, slong len, gr_ctx_t ctx)sourceint _gr_poly_rsqrt_series_miller(gr_ptr res, gr_srcptr f, slong flen, slong len, gr_ctx_t ctx)sourceint _gr_poly_rsqrt_series_newton(gr_ptr res, gr_srcptr f, slong flen, slong len, slong cutoff, gr_ctx_t ctx)sourceThe basecase versions solve for one coefficient at a time with dot products. The miller versions use the recurrence of gr_poly_pow_series_fmpq_recurrence with exponent \pm 1/2, which costs O(\mathit{flen} \cdot \mathit{len}) operations but divides by 1, \ldots, \mathit{len} - 1. The newton versions use the third-order iteration for the reciprocal square root (each step about triples the precision), and obtain the square root from the reciprocal square root to half the length with a final Karp-Markstein step.
The default methods GR_METHOD_POLY_SQRT_SERIES and GR_METHOD_POLY_RSQRT_SERIES use Miller's recurrence for short inputs (\mathit{flen} \le 8 < \mathit{len}) over exact rings; otherwise the square root uses Newton iteration, and the reciprocal square root the basecase for \mathit{len} \le 8 or classical multiplication and Newton iteration otherwise. A GR_DOMAIN failure of the square root is turned into GR_UNABLE unless the ring is a field of characteristic other than 2, since in other rings a square root may exist that the algorithm cannot find. The underscore versions require \mathit{flen} \ge 1, write len coefficients and do not allow aliasing.
The FLINT team, “_gr_poly_sqrt_series, gr_poly_sqrt_series, _gr_poly_sqrt_series_basecase, gr_poly_sqrt_series_basecase, _gr_poly_sqrt_series_miller, gr_poly_sqrt_series_miller, _gr_poly_sqrt_series_newton, gr_poly_sqrt_series_newton, _gr_poly_rsqrt_series, gr_poly_rsqrt_series, _gr_poly_rsqrt_series_basecase, gr_poly_rsqrt_series_basecase, _gr_poly_rsqrt_series_miller, gr_poly_rsqrt_series_miller, _gr_poly_rsqrt_series_newton, gr_poly_rsqrt_series_newton (gr_poly.h),” FLINT: Fast Library for Number Theory, documentation, version 3.7.0-dev, 2026. https://flintlib.org/doc/gr_poly/_gr_poly_sqrt_series.html (generated 2026-10-20).
@misc{flint-doc-gr-poly-gr-poly-sqrt-series,
author = {{The FLINT team}},
title = {{\_gr\_poly\_sqrt\_series, gr\_poly\_sqrt\_series, \_gr\_poly\_sqrt\_series\_basecase, gr\_poly\_sqrt\_series\_basecase, \_gr\_poly\_sqrt\_series\_miller, gr\_poly\_sqrt\_series\_miller, \_gr\_poly\_sqrt\_series\_newton, gr\_poly\_sqrt\_series\_newton, \_gr\_poly\_rsqrt\_series, gr\_poly\_rsqrt\_series, \_gr\_poly\_rsqrt\_series\_basecase, gr\_poly\_rsqrt\_series\_basecase, \_gr\_poly\_rsqrt\_series\_miller, gr\_poly\_rsqrt\_series\_miller, \_gr\_poly\_rsqrt\_series\_newton, gr\_poly\_rsqrt\_series\_newton (gr\_poly.h)}},
howpublished = {FLINT: Fast Library for Number Theory, documentation, version 3.7.0-dev},
year = {2026},
url = {https://flintlib.org/doc/gr_poly/_gr_poly_sqrt_series.html},
note = {Generated 2026-10-20}
}