is there a small typo in the gradient computation following 11.6.7?
if we expand this, it will come to Qx - c which does not match the derivative of 11.6.6.
I think what is meant on the RHS is Q(x-x*)
Oh, I am really glad to see your comment because I was almost going to crazy. Eqn. 11.6.7 has also similiar sign mistake I guess.