Fix `sum()` accuracy

`[1e20, 1.0, -1e20].sum()` returns `0.0`. This happens because during
the summation, `1.0` is too small relative to `1e20`, making it
negligible.

I have tried Kahan summation but it hasn't fixed the problem.
Therefore, I've used Python's `fsum()` implementation with some
help from Jason Fager and Huon Wilson.
For more details, read:
www.cs.cmu.edu/~quake-papers/robust-arithmetic.ps

Moreover, benchmark and unit tests were added.

Note: `Status.sum` is still not fully fixed. It doesn't handle
NaNs, infinities and overflow correctly. See issue 11059:
https://github.com/mozilla/rust/issues/11059

1 files changed, 66 insertions, 1 deletions

diff --git a/src/libextra/stats.rs b/src/libextra/stats.rs
index 44c399c89da..f79ec51a9f7 100644
--- a/src/libextra/stats.rs
+++ b/src/libextra/stats.rs
@@ -15,6 +15,7 @@ use std::cmp;
 use std::hashmap;
 use std::io;
 use std::num;
+use std::util;
 
 // NB: this can probably be rewritten in terms of num::Num
 // to be less f64-specific.
@@ -23,6 +24,12 @@ use std::num;
 pub trait Stats {
 
     /// Sum of the samples.
+    ///
+    /// Note: this method sacrifices performance at the altar of accuracy
+    /// Depends on IEEE-754 arithmetic guarantees. See proof of correctness at:
+    /// ["Adaptive Precision Floating-Point Arithmetic and Fast Robust Geometric Predicates"]
+    /// (http://www.cs.cmu.edu/~quake-papers/robust-arithmetic.ps)
+    /// *Discrete & Computational Geometry 18*, 3 (Oct 1997), 305-363, Shewchuk J.R.
     fn sum(self) -> f64;
 
     /// Minimum value of the samples.
@@ -147,8 +154,37 @@ impl Summary {
 
 impl<'self> Stats for &'self [f64] {
 
+    // FIXME #11059 handle NaN, inf and overflow
     fn sum(self) -> f64 {
-        self.iter().fold(0.0, |p,q| p + *q)
+        let mut partials : ~[f64] = ~[];
+
+        for &mut x in self.iter() {
+            let mut j = 0;
+            // This inner loop applies `hi`/`lo` summation to each
+            // partial so that the list of partial sums remains exact.
+            for i in range(0, partials.len()) {
+                let mut y = partials[i];
+                if num::abs(x) < num::abs(y) {
+                    util::swap(&mut x, &mut y);
+                }
+                // Rounded `x+y` is stored in `hi` with round-off stored in
+                // `lo`. Together `hi+lo` are exactly equal to `x+y`.
+                let hi = x + y;
+                let lo = y - (hi - x);
+                if lo != 0f64 {
+                    partials[j] = lo;
+                    j += 1;
+                }
+                x = hi;
+            }
+            if j >= partials.len() {
+                partials.push(x);
+            } else {
+                partials[j] = x;
+                partials.truncate(j+1);
+            }
+        }
+        partials.iter().fold(0.0, |p, q| p + *q)
     }
 
     fn min(self) -> f64 {
@@ -955,5 +991,34 @@ mod tests {
         t(&Summary::new([-2.0, 0.0]), ~"-2 |[------******#******---]| 0");
 
     }
+    #[test]
+    fn test_sum_f64s() {
+        assert_eq!([0.5, 3.2321, 1.5678].sum(), 5.2999);
+    }
+    #[test]
+    fn test_sum_f64_between_ints_that_sum_to_0() {
+        assert_eq!([1e30, 1.2, -1e30].sum(), 1.2);
+    }
+}
 
+#[cfg(test)]
+mod bench {
+    use extra::test::BenchHarness;
+    use std::vec;
+
+    #[bench]
+    fn sum_three_items(bh: &mut BenchHarness) {
+        bh.iter(|| {
+            [1e20, 1.5, -1e20].sum();
+        })
+    }
+    #[bench]
+    fn sum_many_f64(bh: &mut BenchHarness) {
+        let nums = [-1e30, 1e60, 1e30, 1.0, -1e60];
+        let v = vec::from_fn(500, |i| nums[i%5]);
+
+        bh.iter(|| {
+            v.sum();
+        })
+    }
 }