s-news
[Top] [All Lists]

This must be easy, but...

To: S-News Mail List <s-news@lists.biostat.wustl.edu>
Subject: This must be easy, but...
From: "Kim Elmore" <Kim.Elmore@noaa.gov>
Date: Fri, 23 May 2003 16:55:30 -0500
Dear S'ers,

I'm dabbling with minimal spanning trees. I want the sum of the distances for all branches of the tree in a neat form. As an example, I run mstree with plan=F. This yields a vector of how the objects or cases connect together. Say I have 18 objects, and mstree returns

5 15 11  6 10 12  4  9 13  3 18 11 16 13  4 17  5

Let this be stored in the vector mst. Now assume that I have also built a distance (called dst) matrix for the L2 distances between each of the objects (built with dist and translated to a matrix with dist2full). As I understand this, the integrated distance in the Minimum Spanning Tree is then the sum of dst[1, 5], dst[2, 15], dst[3, 6], ..., dst[17, 5]. I naively thought that I could compute this as

sum(dst[1:17, mst])

but this returns the sum of the whole matrix, because dst[1:17, mst] returns the whole matrix. I can do this in a for loop, but there is surely an elegant, vector way to extract just these elements. What is it?

Kim Elmore


<Prev in Thread] Current Thread [Next in Thread>
  • This must be easy, but..., Kim Elmore <=