- 
    Bug 
- 
    Resolution: Fixed
- 
     P3 P3
- 
    8
- 
        b120
- 
        Verified
| Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build | 
|---|---|---|---|---|---|---|
| JDK-8030298 | 8u5 | Joe Darcy | P3 | Resolved | Fixed | b01 | 
Better implementations are required for sequential and parallel evaluation that reduce numerical error.
The implementations could, for example, use the Kahan summation algorithm:
http://en.wikipedia.org/wiki/Kahan_summation_algorithm
That could be applied for sequential evaluation and for parallel evaluation where the leafs are evaluated sequentially and those sums are then added together up the computation tree.
- backported by
- 
                    JDK-8030298 DoubleStream.sum() & DoubleSummaryStats implementations that reduce numerical errors -           
- Resolved
 
-         
- relates to
- 
                    JDK-8024354 Explicitly permit DoubleStream.sum()/average() implementations to use higher precision summation -           
- Resolved
 
-         
- 
                    JDK-8030942 Explicitly state floating-point summation requirements on non-finite inputs -           
- Resolved
 
-         
- 
                    JDK-8030212 Several api.java.util.stream tests got "NaN" value instead of "Infinity" or "-Infinity" -           
- Closed
 
-         
- 
                    JDK-8214761 Bug in parallel Kahan summation implementation -           
- Closed
 
-         
- 
                    JDK-8029372 Improve combination of partial sums/averages in streams API -           
- Closed
 
-