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
265
266
267
268
269
270
271
272
273
274
275
276
277
278
use crate::reduced::impl_reduced_binary_pow;
use crate::{ModularUnaryOps, Reducer, Vanilla};

/// Negated modular inverse on binary bases
/// `neginv` calculates `-(m^-1) mod R`, `R = 2^k. If m is odd, then result of m + 1 will be returned.
mod neg_mod_inv {
    // Entry i contains (2i+1)^(-1) mod 256.
    #[rustfmt::skip]
    const BINV_TABLE: [u8; 128] = [
        0x01, 0xAB, 0xCD, 0xB7, 0x39, 0xA3, 0xC5, 0xEF, 0xF1, 0x1B, 0x3D, 0xA7, 0x29, 0x13, 0x35, 0xDF,
        0xE1, 0x8B, 0xAD, 0x97, 0x19, 0x83, 0xA5, 0xCF, 0xD1, 0xFB, 0x1D, 0x87, 0x09, 0xF3, 0x15, 0xBF,
        0xC1, 0x6B, 0x8D, 0x77, 0xF9, 0x63, 0x85, 0xAF, 0xB1, 0xDB, 0xFD, 0x67, 0xE9, 0xD3, 0xF5, 0x9F,
        0xA1, 0x4B, 0x6D, 0x57, 0xD9, 0x43, 0x65, 0x8F, 0x91, 0xBB, 0xDD, 0x47, 0xC9, 0xB3, 0xD5, 0x7F,
        0x81, 0x2B, 0x4D, 0x37, 0xB9, 0x23, 0x45, 0x6F, 0x71, 0x9B, 0xBD, 0x27, 0xA9, 0x93, 0xB5, 0x5F,
        0x61, 0x0B, 0x2D, 0x17, 0x99, 0x03, 0x25, 0x4F, 0x51, 0x7B, 0x9D, 0x07, 0x89, 0x73, 0x95, 0x3F,
        0x41, 0xEB, 0x0D, 0xF7, 0x79, 0xE3, 0x05, 0x2F, 0x31, 0x5B, 0x7D, 0xE7, 0x69, 0x53, 0x75, 0x1F,
        0x21, 0xCB, 0xED, 0xD7, 0x59, 0xC3, 0xE5, 0x0F, 0x11, 0x3B, 0x5D, 0xC7, 0x49, 0x33, 0x55, 0xFF,
    ];

    pub mod u8 {
        use super::*;
        pub const fn neginv(m: u8) -> u8 {
            let i = BINV_TABLE[((m >> 1) & 0x7F) as usize];
            i.wrapping_neg()
        }
    }

    pub mod u16 {
        use super::*;
        pub const fn neginv(m: u16) -> u16 {
            let mut i = BINV_TABLE[((m >> 1) & 0x7F) as usize] as u16;
            // hensel lifting
            i = 2u16.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i.wrapping_neg()
        }
    }

    pub mod u32 {
        use super::*;
        pub const fn neginv(m: u32) -> u32 {
            let mut i = BINV_TABLE[((m >> 1) & 0x7F) as usize] as u32;
            i = 2u32.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u32.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i.wrapping_neg()
        }
    }

    pub mod u64 {
        use super::*;
        pub const fn neginv(m: u64) -> u64 {
            let mut i = BINV_TABLE[((m >> 1) & 0x7F) as usize] as u64;
            i = 2u64.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u64.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u64.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i.wrapping_neg()
        }
    }

    pub mod u128 {
        use super::*;
        pub const fn neginv(m: u128) -> u128 {
            let mut i = BINV_TABLE[((m >> 1) & 0x7F) as usize] as u128;
            i = 2u128.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u128.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u128.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i = 2u128.wrapping_sub(i.wrapping_mul(m)).wrapping_mul(i);
            i.wrapping_neg()
        }
    }

    pub mod usize {
        #[inline]
        pub const fn neginv(m: usize) -> usize {
            #[cfg(target_pointer_width = "16")]
            return super::u16::neginv(m as _) as _;
            #[cfg(target_pointer_width = "32")]
            return super::u32::neginv(m as _) as _;
            #[cfg(target_pointer_width = "64")]
            return super::u64::neginv(m as _) as _;
        }
    }
}

/// A modular reducer based on [Montgomery form](https://en.wikipedia.org/wiki/Montgomery_modular_multiplication#Montgomery_form), only supports odd modulus.
///
/// The generic type T represents the underlying integer representation for modular inverse `-m^-1 mod R`,
/// and `R=2^B` will be used as the auxiliary modulus, where B is automatically selected
/// based on the size of T.
#[derive(Debug, Clone, Copy)]
pub struct Montgomery<T> {
    m: T,   // modulus
    inv: T, // modular inverse of the modulus
}

macro_rules! impl_montgomery_for {
    ($t:ident, $ns:ident) => {
        mod $ns {
            use super::*;
            use crate::word::$t::*;
            use neg_mod_inv::$t::neginv;

            impl Montgomery<$t> {
                pub const fn new(m: $t) -> Self {
                    assert!(
                        m & 1 != 0,
                        "Only odd modulus are supported by the Montgomery form"
                    );
                    Self { m, inv: neginv(m) }
                }
                const fn reduce(&self, monty: DoubleWord) -> $t {
                    debug_assert!(high(monty) < self.m);

                    // REDC algorithm
                    let tm = low(monty).wrapping_mul(self.inv);
                    let (t, overflow) = monty.overflowing_add(wmul(tm, self.m));
                    let t = high(t);

                    if overflow {
                        t + self.m.wrapping_neg()
                    } else if t >= self.m {
                        t - self.m
                    } else {
                        t
                    }
                }
            }

            impl Reducer<$t> for Montgomery<$t> {
                #[inline]
                fn new(m: &$t) -> Self {
                    Self::new(*m)
                }
                #[inline]
                fn transform(&self, target: $t) -> $t {
                    if target == 0 {
                        return 0;
                    }
                    nrem(merge(0, target), self.m)
                }
                #[inline]
                fn check(&self, target: &$t) -> bool {
                    *target < self.m
                }

                #[inline]
                fn residue(&self, target: $t) -> $t {
                    self.reduce(extend(target))
                }
                #[inline(always)]
                fn modulus(&self) -> $t {
                    self.m
                }
                #[inline(always)]
                fn is_zero(&self, target: &$t) -> bool {
                    *target == 0
                }

                #[inline(always)]
                fn add(&self, lhs: &$t, rhs: &$t) -> $t {
                    Vanilla::<$t>::add(&self.m, *lhs, *rhs)
                }

                #[inline(always)]
                fn dbl(&self, target: $t) -> $t {
                    Vanilla::<$t>::dbl(&self.m, target)
                }

                #[inline(always)]
                fn sub(&self, lhs: &$t, rhs: &$t) -> $t {
                    Vanilla::<$t>::sub(&self.m, *lhs, *rhs)
                }

                #[inline(always)]
                fn neg(&self, target: $t) -> $t {
                    Vanilla::<$t>::neg(&self.m, target)
                }

                #[inline]
                fn mul(&self, lhs: &$t, rhs: &$t) -> $t {
                    self.reduce(wmul(*lhs, *rhs))
                }

                #[inline]
                fn sqr(&self, target: $t) -> $t {
                    self.reduce(wsqr(target))
                }

                #[inline(always)]
                fn inv(&self, target: $t) -> Option<$t> {
                    // TODO: support direct montgomery inverse
                    // REF: http://cetinkayakoc.net/docs/j82.pdf
                    self.residue(target)
                        .invm(&self.m)
                        .map(|v| self.transform(v))
                }

                impl_reduced_binary_pow!(Word);
            }
        }
    };
}
impl_montgomery_for!(u8, u8_impl);
impl_montgomery_for!(u16, u16_impl);
impl_montgomery_for!(u32, u32_impl);
impl_montgomery_for!(u64, u64_impl);
impl_montgomery_for!(u128, u128_impl);
impl_montgomery_for!(usize, usize_impl);

// TODO(v0.6.x): accept even numbers by removing 2 factors from m and store the exponent
// Requirement: 1. A separate class to perform modular arithmetics with 2^n as modulus
//              2. Algorithm for construct residue from two components (see http://koclab.cs.ucsb.edu/teaching/cs154/docx/Notes7-Montgomery.pdf)
// Or we can just provide crt function, and let the implementation of monty int with full modulus support as an example code.

#[cfg(test)]
mod tests {
    use super::*;
    use rand::random;

    const NRANDOM: u32 = 10;

    #[test]
    fn creation_test() {
        // a deterministic test case for u128
        let a = (0x81u128 << 120) - 1;
        let m = (0x81u128 << 119) - 1;
        let m = m >> m.trailing_zeros();
        let r = Montgomery::<u128>::new(m);
        assert_eq!(r.residue(r.transform(a)), a % m);

        // is_zero test
        let r = Montgomery::<u8>::new(11u8);
        assert!(r.is_zero(&r.transform(0)));
        let five = r.transform(5u8);
        let six = r.transform(6u8);
        assert!(r.is_zero(&r.add(&five, &six)));

        // random creation test
        for _ in 0..NRANDOM {
            let a = random::<u8>();
            let m = random::<u8>() | 1;
            let r = Montgomery::<u8>::new(m);
            assert_eq!(r.residue(r.transform(a)), a % m);

            let a = random::<u16>();
            let m = random::<u16>() | 1;
            let r = Montgomery::<u16>::new(m);
            assert_eq!(r.residue(r.transform(a)), a % m);

            let a = random::<u32>();
            let m = random::<u32>() | 1;
            let r = Montgomery::<u32>::new(m);
            assert_eq!(r.residue(r.transform(a)), a % m);

            let a = random::<u64>();
            let m = random::<u64>() | 1;
            let r = Montgomery::<u64>::new(m);
            assert_eq!(r.residue(r.transform(a)), a % m);

            let a = random::<u128>();
            let m = random::<u128>() | 1;
            let r = Montgomery::<u128>::new(m);
            assert_eq!(r.residue(r.transform(a)), a % m);
        }
    }

    #[test]
    fn test_against_modops() {
        use crate::reduced::tests::ReducedTester;
        for _ in 0..NRANDOM {
            ReducedTester::<u8>::test_against_modops::<Montgomery<u8>>(true);
            ReducedTester::<u16>::test_against_modops::<Montgomery<u16>>(true);
            ReducedTester::<u32>::test_against_modops::<Montgomery<u32>>(true);
            ReducedTester::<u64>::test_against_modops::<Montgomery<u64>>(true);
            ReducedTester::<u128>::test_against_modops::<Montgomery<u128>>(true);
            ReducedTester::<usize>::test_against_modops::<Montgomery<usize>>(true);
        }
    }
}