Given an error surface,

E$($p$,$q$,$ r$)=3$p$3+3$q$2+4$r$+5$

compute the optimal value of learning rate that will minimize the error surface, the largest learning rate for convergence and smallest learning rate for divergence.