Commit Graph

831 Commits

Author SHA1 Message Date
Josh Stone 6161f1ade1 impl 128-bit MulAdd and MulAddAssign 2018-05-07 12:28:35 -07:00
Josh Stone fe53805550 impl 128-bit CheckedRem and CheckedNeg 2018-05-07 12:28:15 -07:00
Josh Stone 428e0107d2 Add imports for 128-bit Bounded 2018-05-07 12:27:42 -07:00
Josh Stone d2107ae005 There's no u64 exponent impl for i128/u128 Pow 2018-05-07 12:26:44 -07:00
Josh Stone 08ad9b1642 i128 is not Unsigned 2018-05-07 12:26:00 -07:00
Josh Stone 261efafe0b Merge branch 'master' into regexident-i128 2018-05-04 12:28:48 -07:00
bors[bot] a49013e338 Merge #59
59: Added `MulAdd` and `MulAddAssign` traits r=cuviper a=regexident

Both `f32` and `f64` implement fused multiply-add, which computes `(self * a) + b` with only one rounding error. This produces a more accurate result with better performance than a separate multiplication operation followed by an add:

```rust
fn mul_add(self, a: f32, b: f32) -> f32[src]
```

It is however not possible to make use of this in a generic context by abstracting over a trait.

My concrete use-case is machine learning, [gradient descent](https://en.wikipedia.org/wiki/Gradient_descent) to be specific,  
where the core operation of updating the gradient could make use of `mul_add` for both its `weights: Vector` as well as its `bias: f32`:

```rust
struct Perceptron {
  weights: Vector,
  bias: f32,
}

impl MulAdd<f32, Self> for Vector {
  // ...
}

impl Perceptron {
  fn learn(&mut self, example: Vector, expected: f32, learning_rate: f32) {
    let alpha = self.error(example, expected, learning_rate);
    self.weights = example.mul_add(alpha, self.weights);
    self.bias = self.bias.mul_add(alpha, self.bias)
  }
}
```

(The actual impl of `Vector` would be generic over its value type: `Vector<T>`, thus requiring the trait.)

Co-authored-by: Vincent Esche <regexident@gmail.com>
Co-authored-by: Josh Stone <cuviper@gmail.com>
2018-05-04 19:12:41 +00:00
Josh Stone 0d358034d9 Test MulAdd explicitly, guarded by std for floats 2018-05-04 12:09:02 -07:00
Vincent Esche 157efc5a26 Adjusted CI testing scripts 2018-04-19 09:26:02 +02:00
Vincent Esche 28be885481 Moved impl of `MulAdd`/`MulAddAssign` for `f32`/`f64` behind feature guard 2018-04-18 10:31:37 +02:00
Vincent Esche 257917f3f2 Removed inferrable value suffixes 2018-04-18 10:19:30 +02:00
Josh Stone 4fb749a401 typo: taht -> that 2018-04-13 14:29:00 -07:00
bors[bot] 4195043240 Merge #63
63: Add CheckedRem and CheckedNeg r=cuviper a=LEXUGE

Continue from #58 
I've alreadyremoved all the formats.


Co-authored-by: LEXUGE <lexugeyky@outlook.com>
Co-authored-by: Josh Stone <cuviper@gmail.com>
2018-04-13 21:25:30 +00:00
Josh Stone aa21fba9fc re-export CheckedRem and CheckedNeg at the root 2018-04-13 14:14:49 -07:00
Josh Stone b1c4074cc4 Document CheckedRem and CheckedNeg 2018-04-13 14:14:22 -07:00
Josh Stone 5fb3724b69 rename checked_impl_one_param to checked_impl_unary 2018-04-13 14:13:42 -07:00
LEXUGE f74de249c8
remove formats 2018-04-13 16:04:56 +08:00
bors[bot] 97f3892bd1 Merge #62
62: Update outdated FIXME r=cuviper a=vks



Co-authored-by: Vinzent Steinberg <vinzent.steinberg@gmail.com>
2018-04-10 17:54:20 +00:00
Vinzent Steinberg 2836cfc9ab Update outdated FIXME 2018-04-10 19:51:03 +02:00
Vincent Esche dd5b107c56 Added mention of `i128` feature to ‘README.md’ 2018-04-10 10:42:35 +02:00
Vincent Esche 746db74dac Added impls of `Unsigned` for `i128` and `u128` 2018-04-10 10:39:57 +02:00
Vincent Esche 152b38e03f Added impls of `Num` for `i128` and `u128` 2018-04-10 10:39:54 +02:00
Vincent Esche 6d3b55030f Added ‘i128’ feature and unit test invocation 2018-04-10 10:35:55 +02:00
Vincent Esche 830363024b Added `MulAdd` and `MulAddAssign` traits 2018-04-10 10:08:55 +02:00
Vincent Esche d1334bf903 Added impls of `Signed` for `i128` and `u128` 2018-04-09 12:58:17 +02:00
Vincent Esche f69af180cc Added impls of `Pow` for `i128` and `u128` 2018-04-09 11:11:15 +02:00
Vincent Esche 5ee2570618 Added impls of `Wrapping…` for `i128` and `u128` 2018-04-09 11:11:05 +02:00
Vincent Esche 234706fb97 Added impls of `Saturating…` for `i128` and `u128` 2018-04-09 11:10:57 +02:00
Vincent Esche b44666183d Added impls of `Checked…` for `i128` and `u128` 2018-04-09 11:10:51 +02:00
Vincent Esche bc19c34934 Added impls of `PrimInt` for `i128` and `u128` 2018-04-09 11:10:45 +02:00
Vincent Esche 62723f6f3a Added impls of `Zero` and `One` for `i128` and `u128` 2018-04-09 11:10:36 +02:00
Vincent Esche f8d1896c6c Added impls of `Bounded` for `i128` and `u128` 2018-04-09 11:10:23 +02:00
bors[bot] bb67a3d03a Merge #53
53: Release 0.2.2 r=cuviper a=cuviper
2018-03-18 23:32:18 +00:00
Josh Stone 058a6004f0 Release 0.2.2 2018-03-18 16:27:31 -07:00
bors[bot] fcc33a3577 Merge #52
52: Refactor ToPrimitive range checks r=cuviper a=cuviper

This is a rebase and continuation of PR #28.  The primary benefit is that
floats finally check for overflow before casting to integers, avoiding
undefined behavior.  Fixes #12.

The inter-integer conversions and all of the macros for these have also been
tweaked, hopefully improving readability.  Exhaustive tests have been added for
good and bad conversions around the target MIN and MAX values.
2018-03-13 21:10:04 +00:00
Josh Stone a4d234c253 Further simplify float-to-int range checks
We don't actually need to compute the `trunc()` value, as long as we can
figure out the right values for the exclusive range `(MIN-1, MAX+1)` to
measure the same truncation effect.
2018-03-13 13:38:17 -07:00
Josh Stone f0ed42b3bc Test edge cases of ToPrimitive with ints 2018-03-11 01:37:27 -08:00
Josh Stone 50868c60d2 Refactor to_primitive_int/uint macros 2018-03-11 01:37:00 -08:00
Josh Stone 6d7bbb1b53 Mask debug prints no-std mode 2018-03-11 01:36:17 -08:00
Josh Stone d195eafbe2 Simplify the to_primitive_float macros 2018-03-10 23:33:47 -08:00
Josh Stone f6dc4d29a4 Add thorough tests of float to int edge cases 2018-03-10 23:05:02 -08:00
Josh Stone b025c273c7 Rewrite range checks in float ToPrimitive macros 2018-03-10 23:01:30 -08:00
Dan Barella 8e27c7327d Rename some tests. 2018-03-10 15:14:23 -08:00
Dan Barella aab7098acd Reformat macros. 2018-03-10 15:14:23 -08:00
Alexander Popiak c32cb5c65b Patch in apopiak@'s changes from github.com/rust-num/num/pull/135/. 2018-03-10 15:14:23 -08:00
Dan Barella ab8fda7654 Change assert form. 2018-03-10 15:14:23 -08:00
Dan Barella ecb0816c83 Remove an unneeded assert. 2018-03-10 15:14:23 -08:00
Dan Barella 3534a89858 Don't use assert_ne!
`num` is tested against `rust 1.8.0`, which doesn't include
`assert_ne!` -- so we use a plain ol' `assert` instead.
2018-03-10 15:14:23 -08:00
Dan Barella f99aa0e181 Check overflow when casting floats to integers.
This change adds some new macro rules used when converting from floats
to integers. There are two macro rule variants, one for signed ints, one
for unsigned ints.

Among other things, this change specifically addresses the overflow case
documented in https://github.com/rust-num/num-traits/issues/12
2018-03-10 15:14:23 -08:00
bors[bot] aa36cdb206 Merge #42
42: Release 0.2.1 r=cuviper a=cuviper
2018-03-01 23:29:38 +00:00