{convex} + {concave}
Apparently, there is at least one term in the sum for which beta.*Dk.*Ck*theta0_k is negative, which would make that term concave. And at least one term for which that is positive, making that term convex.
thank you very much. I didn’t notice that there could be negative numbers in Dk or Ck. If I guarantee that Dk, Ck are non-negative, I can guarantee that E_cp2 is convex. I understand this right?
Presuming that beta and theta0_k also are nonnegative, then yes.
All that really matters is the sign of the product of all the items which multiply square(f0) .
