How to slice a sparse matrix in scala breeze? - scala

Question

I create a sparse matrix in scala breeze, ie using http://www.scalanlp.org/api/breeze/linalg/CSCMatrix.html. Now I want to get a column slice from it. How to do this?
Edit: there are some further requirements:
it is important to me that I can actually do something useful with the slice, eg multiply it by a float:
X(::,n) * 3.
It's also important to me that the resulting structure/matrix/vector remains sparse. Each column might have a dense dimension of several million, but in fact have only 600 entries or so.
I need to be able to use this to mutate the matrix, eg:
X(::,0) = X(::,1)

Answer 1

Slicing works the same as for DenseMatrix, which is discussed in the Quickstart.
val m1 = CSCMatrix((1, 2, 3, 4), (5, 6, 7, 8), (9, 10, 11, 12), (13, 14, 15, 16))
val m2 = m1(1 to 2, 1 to 2)
println(m2)
This prints:
6 7
10 11

Answer 2

I wrote my own slicer method in the end. Use like this:
val col = root.MatrixHelper.colSlice( sparseMatrix, columnIndex )
code:
// Copyright Hugh Perkins 2012
// You can use this under the terms of the Apache Public License 2.0
// http://www.apache.org/licenses/LICENSE-2.0
package root
import breeze.linalg._
object MatrixHelper {
def colSlice( A: CSCMatrix[Double], colIndex: Int ) : SparseVector[Double] = {
val size = A.rows
val rowStartIndex = A.colPtrs(colIndex)
val rowEndIndex = A.colPtrs(colIndex + 1) - 1
val capacity = rowEndIndex - rowStartIndex + 1
val result = SparseVector.zeros[Double](size)
result.reserve(capacity)
var i = 0
while( i < capacity ) {
val thisindex = rowStartIndex + i
val row = A.rowIndices(thisindex)
val value = A.data(thisindex)
result(row) = value
i += 1
}
result
}
}