Package dev.frostlake.functions.aggregate

Class PartialFloatSum

java.lang.Object
dev.frostlake.functions.aggregate.PartialFloatSum
public final class PartialFloatSum extends Object
The double a FLOAT SUM or AVG answers, built from PARTIALS: each partial is a run of terms added with Kahan compensation and read back with its last compensation applied (CompensatedSum.correctedValue()), and the partials are added in turn to a running compensated sum that is read without it (CompensatedSum.value()).

The account's shapes are three uses of that one rule (all live-verified): 

   an aggregate, a whole-partition window     one partial holding every row     the corrected sum
   a running or sliding window frame          every row a partial of its own    the running sum
   a cumulative RANGE frame                   every peer group a partial        corrected within a group

Over 0.3 then 0.7 squared the corrected sum is 0.57999999999999984901 and the running one 0.57999999999999996003: SUM(x * x) and SUM(x * x) OVER () answer the first, SUM(x * x) OVER (ORDER BY i) answers the second on its last row, and OVER (ORDER BY k) answers the first again on both rows when they share k.

  • Constructor Details

    • PartialFloatSum

      public PartialFloatSum()

  • Method Details

    • add

      public void add(double term)
      Add a term to the current partial.
      Parameters:
      term - the next value, in input order

    • endPartial

      public void endPartial()
      Close the current partial: its corrected sum joins the running sum. A partial with no term adds nothing.

    • result

      public double result()
      The running sum of every partial, the current one closed first.
      Returns:
      the sum, or 0.0 before any term