A demo application using easel3d to draw in an HTML5 canvas element in a web page.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

352 lines
10 KiB

//
// Some code to support fractional numbers for full precision rational number
// calculations. (At least for the standard operations.)
// This also implements a sqrt on fractional numbers, which can not be precise
// because of the irrational nature of most sqare roots.
// Fractions can only represent rational numbers precise.
//
// Georg Hopp <georg@steffers.org>
//
// Copyright © 2019 Georg Hopp
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
//
use std::cmp::Ordering;
use std::ops::{Add,Sub,Neg,Mul,Div};
use std::fmt;
use std::convert::{TryFrom, TryInto};
use std::num::TryFromIntError;
#[derive(Debug, Eq, Clone, Copy)]
pub struct Fractional (pub i64, pub i64);
pub type Continuous = Vec<i64>;
pub type Error = &'static str;
#[inline]
fn hcf(x :i64, y :i64) -> i64 {
match y {
0 => x,
_ => hcf(y, x % y),
}
}
// calculate a sqrt as continued fraction sequence. Taken from:
// https://en.wikipedia.org/wiki/Methods_of_computing_square_roots#Continued_fraction_expansion
fn sqrt_cfrac(x :i64, a0 :i64) -> Continuous {
let v :Continuous = Vec::new();
fn inner(mut v :Continuous,
x :i64,
a0 :i64,
mn :i64,
dn :i64,
an :i64) -> Continuous {
let mn_1 = dn * an - mn;
let dn_1 = (x - mn_1 * mn_1) / dn;
let an_1 = (a0 + mn_1) / dn_1;
v.push(an);
match v.len() {
10 => v,
_ => inner(v, x, a0, mn_1, dn_1, an_1),
}
// This convergence criterium is not good for very small x
// thus I decided to break the iteration at constant time.
// match an_1 == 2 * a0 {
// true => v,
// _ => inner(v, x, a0, mn_1, dn_1, an_1),
// }
}
inner(v, x, a0, 0, 1, a0)
}
impl Fractional {
#[inline]
pub fn gcd(self, other: Self) -> i64 {
let Fractional(_, d1) = self;
let Fractional(_, d2) = other;
(d1 * d2) / hcf(d1, d2)
}
#[inline]
pub fn reduce(self) -> Self {
let Fractional(n, d) = self;
let (_n, _d) = if n > d { (n, d) } else { (d, n) };
// if the difference from _n % _d to _n is very big we are close to
// a whole number and can ignore the fractional part... this reduces
// the precision but ensures smaller numbers for numerator and
// denominator.
if _d > 1 && (_n % _d) * 10000000 < _n {
Self(_n / _d, 1)
} else {
Self(n / hcf(n, d), d / hcf(n, d))
}
}
#[inline]
pub fn numerator(self) -> i64 {
self.0
}
#[inline]
pub fn denominator(self) -> i64 {
self.1
}
// This is a really bad approximation of sqrt for a fractional...
// for (9/3) it will result 3 which if way to far from the truth,
// which is ~1.7320508075
// BUT we can use this value as starting guess for creating a
// continous fraction for the sqrt... and create a much better
// fractional representation of the sqrt.
// So, if inner converges, but is not a perfect square (does not
// end up in an Ordering::Equal - which is the l > h case)
// we use the l - 1 as starting guess for sqrt_cfrac.
// taken from:
// https://www.geeksforgeeks.org/square-root-of-an-integer/
pub fn sqrt(self) -> Result<Self, Error> {
// find the sqrt of x in O(log x/2).
// This stops if a perfect sqare was found. Else it passes
// the found value as starting guess to the continous fraction
// sqrt function.
fn floor_sqrt(x :i64) -> Fractional {
fn inner(l :i64, h :i64, x :i64) -> Fractional {
if l > h {
(&sqrt_cfrac(x, l - 1)).into()
} else {
let m = (l + h) / 2;
match x.cmp(&(m * m)) {
Ordering::Equal => m.into(),
Ordering::Less => inner(l, m - 1, x),
Ordering::Greater => inner(m + 1, h, x),
}
}
}
match x {
0 => 0.into(),
1 => 1.into(),
_ => inner(1, x / 2, x),
}
}
let Fractional(n, d) = self;
let n = match n.cmp(&0) {
Ordering::Equal => 0.into(),
Ordering::Less => return Err("sqrt on negative undefined"),
Ordering::Greater => floor_sqrt(n),
};
let d = match d.cmp(&0) {
Ordering::Equal => 0.into(),
Ordering::Less => return Err("sqrt on negative undefined"),
Ordering::Greater => floor_sqrt(d),
};
Ok(n / d)
}
}
impl fmt::Display for Fractional {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "({}/{})", self.0, self.1)
}
}
impl From<i64> for Fractional {
fn from(x: i64) -> Self {
Self(x, 1)
}
}
impl TryFrom<usize> for Fractional {
type Error = &'static str;
fn try_from(x: usize) -> Result<Self, Self::Error> {
let v = i64::try_from(x);
match v {
Err(_) => Err("Conversion from usize to i32 failed"),
Ok(_v) => Ok(Self(_v, 1)),
}
}
}
pub fn from_vector(xs: &Vec<i64>) -> Vec<Fractional> {
xs.iter().map(|x| Fractional(*x, 1)).collect()
}
impl TryInto<f64> for Fractional {
type Error = TryFromIntError;
fn try_into(self) -> Result<f64, Self::Error> {
let n :i32 = self.0.try_into()?;
let d :i32 = self.1.try_into()?;
Ok(f64::from(n) / f64::from(d))
}
}
impl TryInto<(i32, i32)> for Fractional {
type Error = TryFromIntError;
fn try_into(self) -> Result<(i32, i32), Self::Error> {
let a :i32 = (self.0 / self.1).try_into()?;
let b :i32 = (self.0 % self.1).try_into()?;
Ok((a, b))
}
}
impl Into<Continuous> for Fractional {
// general continous fraction form of a fractional...
fn into(self) -> Continuous {
let v :Continuous = Vec::new();
fn inner(mut v :Continuous, f :Fractional) -> Continuous {
let Fractional(n, d) = f;
let a = n / d;
let Fractional(_n, _d) = f - a.into();
v.push(a);
match _n {
1 => { v.push(_d); v },
_ => inner(v, Fractional(_d, _n)),
}
}
inner(v, self)
}
}
impl From<&Continuous> for Fractional {
fn from(x :&Continuous) -> Self {
let Self(n, d) = x.iter().rev().fold(Fractional(0, 1), |acc, x| {
let Self(an, ad) = acc + (*x).into();
Self(ad, an)
});
Self(d, n)
}
}
impl PartialEq for Fractional {
fn eq(&self, other: &Self) -> bool {
let Fractional(n1, d1) = self;
let Fractional(n2, d2) = other;
n1 * (self.gcd(*other) / d1) == n2 * (self.gcd(*other) / d2)
}
}
impl PartialOrd for Fractional {
fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
Some(self.cmp(other))
}
}
impl Ord for Fractional {
fn cmp(&self, other: &Self) -> Ordering {
let Fractional(n1, d1) = self;
let Fractional(n2, d2) = other;
let x = n1 * (self.gcd(*other) / d1);
let y = n2 * (self.gcd(*other) / d2);
x.cmp(&y)
}
}
impl Add for Fractional {
type Output = Self;
fn add(self, other: Self) -> Self {
let Fractional(n1, d1) = self;
let Fractional(n2, d2) = other;
let n = n1 * (self.gcd(other) / d1) + n2 * (self.gcd(other) / d2);
Self(n, self.gcd(other)).reduce()
}
}
impl Sub for Fractional {
type Output = Self;
fn sub(self, other: Self) -> Self {
self + -other
}
}
impl Neg for Fractional {
type Output = Self;
fn neg(self) -> Self {
let Fractional(n, d) = self;
Self(-n, d).reduce()
}
}
impl Mul for Fractional {
type Output = Self;
fn mul(self, other :Self) -> Self {
let Fractional(n1, d1) = self;
let Fractional(n2, d2) = other;
Self(n1 * n2, d1 * d2).reduce()
}
}
impl Div for Fractional {
type Output = Self;
fn div(self, other: Self) -> Self {
let Fractional(n, d) = other;
self * Fractional(d, n)
}
}
/* some stuff that could be tested...
let x = Fractional(1, 3);
let y = Fractional(1, 6);
println!(
"Greatest common denominator of {} and {}: {}", x, y, x.gcd(y));
println!("Numerator of {}: {}", x, x.numerator());
println!("Denominator of {}: {}", x, x.denominator());
assert_eq!(Fractional(1, 3), Fractional(2, 6));
assert_eq!(Fractional(1, 3), Fractional(1, 3));
assert_eq!(y < x, true);
assert_eq!(y > x, false);
assert_eq!(x == y, false);
assert_eq!(x == x, true);
assert_eq!(x + y, Fractional(1, 2));
println!("{} + {} = {}", x, y, x + y);
assert_eq!(x - y, Fractional(1, 6));
println!("{} - {} = {}", x, y, x - y);
assert_eq!(y - x, Fractional(-1, 6));
println!("{} - {} = {}", y, x, y - x);
assert_eq!(-x, Fractional(-1, 3));
println!("-{} = {}", x, -x);
assert_eq!(x * y, Fractional(1, 18));
println!("{} * {} = {}", x, y, x * y);
assert_eq!(x / y, Fractional(2, 1));
println!("{} / {} = {}", x, y, x / y);
assert_eq!(y / x, Fractional(1, 2));
println!("{} / {} = {}", y, x, y / x);
println!("Fractional from 3: {}", Fractional::from(3));
let z :f64 = Fractional::into(x);
println!("Floating point of {}: {}", x, z);
let (d, r) = Fractional::into(x);
println!("(div, rest) of {}: ({}, {})", x, d, r);
*/