Scala reduce on Point class

Question

I cannot get the Point class to search for the x coordinate maximum with one reduce operation (the following I use map then reduce)

case class Point(x : Double, y : Double)
val pts = List(Point(1,3), Point(-3,2), Point(5,3))

pts.map(p => p.x).reduce( _ max _ )

This will return

res22: Double = 5.0

How to use a single reduce operation to get the max of x out? I tried the following, but didn't get what I expect.

pts.reduce( _.x max _.x )
<console>:11: error: type mismatch;
 found   : Double
 required: Point

The Archetypal Paul · Answer 1 · 2015-09-12T08:48:58.457

3

All these are over-complicated. Use the standard library function that does precisely what is needed here:

 pts.maxBy(_.x).x   //> res0: Double = 5.0

Or, if the example is just a toy one and the real use case is more complex, you can use aggregate

pts.aggregate(Double.MinValue)( _ max _.x , _ max _)

edited Sep 12 '15 at 08:48

answered Sep 11 '15 at 07:07

score 2 · Accepted Answer · answered Sep 11 '15 at 02:23

2

You can either use foldLeft to get to a double:

pts.foldLeft(Double.MinValue)(_ max _.x)

Or you can use reduce to keep passing a Point around and pull its x out afterwards:

pts.reduce((a, b) => if (a.x > b.x) a else b).x

answered Sep 11 '15 at 02:23

Sean Vieira

Thanks it works. May you elaborate how `pts.foldLeft(Double.MinValue)(_ max _.x)` this work? Reading the code, the `MinValue` is counterintuitive as we are trying to take the max. – Carson Pun Sep 11 '15 at 20:52
We want the smallest value possible to "seed" our reduce (`fold` is a `reduce` but with a default in case the collection is empty). We start with the smallest entry possible and then take the max of the smallest value possible and our first `Point.x` entry. Then we compare the max of that comparison with our next `Point.x` entry. If we started with a larger value we might wind up with garbage data. (For example, if we started out with 0 and all of our points were `-x` our "max" would be 0, not the smallest negative number in the list.) See also: http://stackoverflow.com/q/17408880/135978 – Sean Vieira Sep 11 '15 at 21:38

elm · Answer 3 · 2015-09-12T18:25:47.510

1

Using reduce and max

pts.map(_.x).reduce( (a,v) => a max v)

Other approaches to find the maximum in a List[Point] includes

pts.map(_.x).max

edited Sep 12 '15 at 18:25

answered Sep 11 '15 at 05:54

elm

1

`pts.map(_.x).sorted.last` O(NlogN) with an extra unnecessary O(N) at the end for good measure. Really not a good alternative... – The Archetypal Paul Sep 11 '15 at 07:04

3 Answers3