The APL Orchard

« first day (2748 days earlier) ← previous day next day → last day (40 days later) »

B. Wilson

7:08 AM

       a←?1e6⍴2
       x←0.57721
       cmpx '+⌿a×x*⍳≢a' '{⍺+⍵×x}⌿a'
   +⌿a×x*⍳≢a → 8.9E¯2 |    0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
   {⍺+⍵×x}⌿a → 2.7E¯1 | +203% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
       a←?1e6⍴10
       cmpx '+⌿a×x*⍳≢a' '{⍺+⍵×x}⌿a'
   +⌿a×x*⍳≢a → 8.9E¯2 |    0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
   {⍺+⍵×x}⌿a → 2.6E¯1 | +196% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕

Horner's method slower than evaluating polynomial naively.

Mostly as expected, but is the gain straightforwardly due to parallelism in the Multiply compared to the linear fold?

Does the interpreter assume commutativity for the plus-fold? I guess that could speed things up, at the risk of permutation-unstable results.

5 hours later…