Message Boards Message Boards

GROUPS:

Units are cute but with come with high overhead

Posted 7 years ago
3433 Views
|
0 Replies
|
4 Total Likes
|

Scenario - a series of distances between GPS points. (Some overhead in getting there too). Fold the list for accumulated distances along a path. How costly are units? About 3 orders of magnitude after realizing that 0.0 is converted on every function call.

(* series of distances between points, will accumulate the path length *)

points = 30000;
distances = 
  Table[Quantity[RandomReal[{10., 100.}], "Meters"], {points}];
Timing[
 path = FoldList[Plus, 0.0, distances];
 ]

(* Lets not forget to put the same units on that zero - could be the  source of a gotcha *)

distances = 
  Table[Quantity[RandomReal[{10., 100.}], "Meters"], {points}];
Timing[
 path = FoldList[Plus, Quantity[ 0.0, "Meters"], distances];
 ]

(* what if I stick to unit-less data? *)

rawdistances = Table[RandomReal[{10., 100.}], {points}];
Timing[
 path = FoldList[Plus, 0.0, rawdistances];
 ]

{28.984986, Null}

{7.597249, Null}

{0.015600, Null}

What about Accumulate? (Only 1000 points - far worse)

points = 1000;
distances = 
  Table[Quantity[RandomReal[{10., 100.}], "Meters"], {points}];
Timing[
 path = Accumulate[distances];
 ]


{23.868153, Null}
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract