Skip to content
Prev 206542 / 398503 Next

optimization problem

Ravi Varadhan <rvaradhan <at> jhmi.edu> writes:
Dear Ravi,

I thought your solution is ingenious, but after the discussion with 
Erwin Kalvelagen I found two things quite irritating:

(1) Why is solve_LSAP(D) different from solve_LSAP(D^2) in Erwin's
    example? I believed just squaring the distance matrix would make
    no difference to solving the LSAP - except for some numerical
    instability which does not seem to be the case here.

(2) If you change rows and sums in the definition of D, that is

    D[j, i] <- sqrt(sum((B[, j] - A[, i])^2))

    then the solution to Erwin's example comes out right even with
    keeping the square root.

Do you have explanations for these 'phenomena'? Otherwise, I think,
there will remain some doubts about this approach.

Very best
Hans Werner