lsqnonlin results getting out of bound. Any solution?

I'm using lsqnonlin for function minimization, the syntax being:

[y,resnorm] = lsqnonlin(@solver21,k0,lb,ub,options)

k0 is the desired result which should be positive.

Hence, I've given lower bound lb as a zero matrix. However some of the values of the output matrix k0 are negative. Any explanations/suggestions?

[Matt J]

The output matrix in the code you've shown is "y", not "k0". Perhaps you're looking at the wrong variable :-)

Aside from that, show what options you're using, in particular which algorithm you've selected.

[Shashank Prasanna]

Also to add, k0 is your initial starting point, it would not make sense to provide a starting point which outside you bounds. You are starting with an infeasible point.