Skip to content

about matrices merge and retrieve algorithm.

2 messages · pratik wankhade, Petr Savicky

#
On Sat, Jan 22, 2011 at 12:13:35PM +0530, pratik wankhade wrote:
Hello.

Note that the above operations lose information. For example, if

  A <- matrix(2, nrow=2, ncol=2)
  B <- diag(2)
  C <- B[2:1, ]
  ABC <- A + B + C
  ABC
  #      [,1] [,2]
  # [1,]    3    3
  # [2,]    3    3

then ABC does not contain enough information to reconstruct the
original A, B, C.

If you want to get any three matrices A, B, C, whose sum is ABC,
then choose any two matrices A, B and compute C as ABC - A - B.

If it is known that the original matrices are not arbitrary and
satisfy some property, which allows to find a unique solution, then
the algorithm depends on what is known about the original matrices.

Hope this helps.

Petr Savicky.