Struct Split 

pub struct Split { /* private fields */ }

An exact sum of two Real values split at an absolute binary digit.

Implementations

impl Split

pub fn new<T: Real>(num: &T, max_p: Option<usize>, n: isize) -> Self

Splits a Real at binary digit n into two RFloat values:

• all significant digits above position n
• all significant digits at or below position n

The sum of the resulting values will be exactly the input number, that is, it “splits” a number.

pub fn num(&self) -> &RFloat

Extracts the upper value of the split.

pub fn lost(&self) -> &RFloat

Extracts the lower value of the split.

pub fn max_p(&self) -> Option<usize>

The precision of the upper value of the split.

pub fn split_pos(&self) -> isize

The position of the first digit lost in the split.

pub fn rgs(&self) -> (bool, bool, bool)

Extracts the round, guard, and sticky bit (“RGS”) from lost digits.

pub fn rs(&self) -> (bool, bool)

Extracts the round and sticky bit (“RGS”) from lost digits.

pub fn is_exact(&self) -> bool

Returns true if the lost digits are all zero.

Trait Implementations

impl Clone for Split

fn clone(&self) -> Split

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Split

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Real for Split

fn radix() -> usize

Radix of a number. It must be strictly positive.

fn sign(&self) -> Option<bool>

The sign bit. This is not always well-defined, so the result is an Option. This is distinct from is_negative (e.g. -0.0 is not negative).

fn exp(&self) -> Option<isize>

The exponent of this number when viewed as (-1)^s * c * b^exp where c is an integer integer. Only well-defined for finite, non-zero numbers.

fn e(&self) -> Option<isize>

The exponent of this number when viewed as (-1)^s * f * b^e where f is a fraction between 1 and 2. This is the preferred IEEE 754 interpretation of an exponent. Only well-defined for finite, non-zero numbers.

fn n(&self) -> Option<isize>

The “absolute digit”, the place below the least significant digit of the mantissa. Always equal to self.exp() - 1. For integer formats, this is just -1. Only well-defined for finite, non-zero numbers.

fn c(&self) -> Option<Integer>

The _unsigned“ integer significand of this number when viewed as (-1)^s * c * b^exp. Only well-defined for finite, non-zero numbers. Only well-defined for finite, non-zero numbers.

fn m(&self) -> Option<Integer>

The signed integer significand of this number when viewed as (-1)^s * c * b^exp. Only well-defined for finite, non-zero numbers. Only well-defined for finite, non-zero numbers.

fn prec(&self) -> Option<usize>

Precision of the significand. This is just floor(logb(c)) where b is the radix and c is the integer significand. Only well-defined for finite, non-zero numbers.

fn is_nar(&self) -> bool

Returns true if this number is not a real number. Example: NaN or +/-Inf from the IEEE 754 standard.

fn is_finite(&self) -> bool

Returns true if this number is finite. For values that do not encode numbers, intervals, or even limiting behavior, the result is false.

fn is_infinite(&self) -> bool

Returns true if this number if infinite. For values that do not encode numbers, intervals, or even limiting behavior, the result is false.

fn is_zero(&self) -> bool

Returns true if this number is zero.

fn is_negative(&self) -> Option<bool>

Returns true if this number is negative. This is not always well-defined, so the result is an Option. This is not necessarily the same as the sign bit (the IEEE 754 standard differentiates between -0.0 and +0.0).

fn is_numerical(&self) -> bool

Returns true if this number represents a numerical value: either a finite number, interval, or some limiting value.

fn split(&self, n: isize) -> (RFloat, RFloat)

Splits this value at the nth binary digit, returning two RFloat values.