trigonometry.rs
8.72 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
//
// Some trigonometic functions with Fractions results.
// Currently only sin, cos and tan are implemented.
// As I was unable to find a really good integral approximation for them I
// implement them as a table which is predefined using the floating point
// function f64::sin and then transformed into a fraction of a given
// PRECISION.
// These approximations are quite good and for a few edge cases
// even better than the floating point implementations.
//
// 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::Neg;
use std::marker::Sized;
use crate::{Fractional, Error};
use crate::continuous::Continuous;
pub trait Trig {
fn pi() -> Self;
fn recip(self) -> Self;
fn sqrt(self) -> Result<Self, Error> where Self: Sized;
fn sintab() -> Vec<Self> where Self: Sized;
fn tantab() -> Vec<Self> where Self: Sized;
fn sin(d :i32) -> Self
where Self: Sized + Neg<Output = Self> + Copy {
match d {
0 ..=90 => Self::sintab()[d as usize],
91 ..=180 => Self::sintab()[180 - d as usize],
181..=270 => -Self::sintab()[d as usize - 180],
271..=359 => -Self::sintab()[360 - d as usize],
_ => {
Self::sin(if d < 0 { d % 360 + 360 } else { d % 360 })
},
}
}
fn cos(d :i32) -> Self
where Self: Sized + Neg<Output = Self> + Copy {
match d {
0 ..=90 => Self::sintab()[90 - d as usize],
91 ..=180 => -Self::sintab()[90 - (180 - d as usize)],
181..=270 => -Self::sintab()[90 - (d as usize - 180)],
271..=359 => Self::sintab()[90 - (360 - d as usize)],
_ => {
Self::cos(if d < 0 { d % 360 + 360 } else { d % 360 })
},
}
}
fn tan(d :i32) -> Self where Self: Sized + Copy {
match d {
0 ..=179 => Self::tantab()[d as usize],
180..=359 => Self::tantab()[d as usize - 180],
_ => {
Self::tan(if d < 0 { d % 360 + 360 } else { d % 360 })
},
}
}
}
// Try to keep precision as high as possible while having a denominator
// as small as possible. The values are taken by try and error.
const PRECISION :i64 = 1000000;
const MAX_DENOMINATOR :i64 = 7000;
// This is a really close fractional approximation for pi.
impl Trig for Fractional {
fn pi() -> Self {
Fractional(355, 113)
}
fn recip(self) -> Self {
let Fractional(n, d) = self;
Fractional(d, n)
}
// 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/
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 {
(&Continuous::sqrt(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)
}
fn sintab() -> Vec<Self> {
// hold sin Fractionals from 0 to 89 ...
// luckily with a bit of index tweeking this can also be used for
// cosine values.
lazy_static::lazy_static! {
static ref SINTAB :Vec<Fractional> =
(0..=90).map(|x| _sin(x)).collect();
}
// fractional sin from f64 sin. (From 0° to 90°)
fn _sin(d: u32) -> Fractional {
match d {
0 => Fractional(0, 1),
90 => Fractional(1, 1),
_ => reduce(d, PRECISION, &f64::sin),
}
}
SINTAB.to_vec()
}
fn tantab() -> Vec<Self> {
// This table exists only because the sin(α) / cos(α) method
// yields very large unreducable denominators in a lot of cases.
lazy_static::lazy_static! {
static ref TANTAB :Vec<Fractional> =
(0..180).map(|x| _tan(x)).collect();
}
// fractional tan from f64 tan. (From 0° to 179°)
fn _tan(d: u32) -> Fractional {
match d {
0 => Fractional(0, 1),
45 => Fractional(1, 1),
90 => Fractional(1, 0), // although they are both inf and -inf.
135 => -Fractional(1, 1),
_ => reduce(d, PRECISION, &f64::tan),
}
}
TANTAB.to_vec()
}
}
impl Trig for f64 {
fn pi() -> Self {
std::f64::consts::PI
}
fn recip(self) -> Self {
self.recip()
}
fn sqrt(self) -> Result<Self, Error> {
let x = self.sqrt();
match x.is_nan() {
true => Err("sqrt on negative undefined"),
false => Ok(x),
}
}
fn sintab() -> Vec<Self> {
lazy_static::lazy_static! {
static ref SINTAB :Vec<f64> =
(0..=90).map(|x| _sin(x)).collect();
}
// f64 sin. (From 0° to 90°)
fn _sin(d: u32) -> f64 {
match d {
0 => 0.0,
90 => 1.0,
_ => (d as f64).to_radians().sin(),
}
}
SINTAB.to_vec()
}
fn tantab() -> Vec<Self> {
// This table exists only because the sin(α) / cos(α) method
// yields very large unreducable denominators in a lot of cases.
lazy_static::lazy_static! {
static ref TANTAB :Vec<f64> =
(0..180).map(|x| _tan(x)).collect();
}
// fractional tan from f64 tan. (From 0° to 179°)
fn _tan(d: u32) -> f64 {
match d {
0 => 0.0,
45 => 1.0,
90 => std::f64::INFINITY,
135 => -1.0,
_ => (d as f64).to_radians().tan(),
}
}
TANTAB.to_vec()
}
}
// search for a fraction with a denominator less than MAX_DENOMINATOR that
// provides the minimal PRECISION criteria.
// !! With f = &f64::tan and d close to the inf boundarys of tan
// we get very large numerators because the numerator becomes a
// multiple of the denominator.
fn reduce(d :u32, p :i64, f :&dyn Fn(f64) -> f64) -> Fractional {
// This is undefined behaviour for very large f64, but our f64
// is always between 0.0 and 1000000.0 which should be fine.
let s = (f((d as f64).to_radians()) * p as f64).round() as i64;
let Fractional(n, dn) = Fractional(s, p).reduce();
match dn.abs().cmp(&MAX_DENOMINATOR) {
Ordering::Less => Fractional(n, dn),
_ => reduce(d, p + 1, f),
}
}