TorchLean API

FloatLib.Floats.Formats.DecimalInterchange.Sqrt.Runtime

Decimal square root by exact squared comparisons #

An integer square root brackets the scaled result. Comparing the radicand with the squared midpoint decides nearest rounding exactly, including ties. All decisions use integer and rational arithmetic.

The preferred exponent is the floor of half the operand quantum (IEEE 754-2019 §5.4.1). Negative zero survives, while a negative nonzero operand raises invalid (§§6.3, 7.2).

Decimal scale shift for a square root, computed without evaluating a real square root.

Instances For

    Smallest coefficient grid which fits the exact square root.

    Instances For

      Integer floor of a nonnegative rational's square root.

      Instances For

        Decide the upper adjacent integer by exact comparison with a squared midpoint.

        Instances For

          Round a nonnegative square root on the integer grid.

          Instances For

            Rounded coefficient and quantum; an exact power-of-ten carry loses no value.

            Instances For

              Round the square root of a nonnegative rational, selecting the preferred cohort member when the result is exact.

              Instances For

                Square root in all five rounding modes, including signed zeros and NaN diagnostics.

                Instances For