Hessian problem in solnp when starting from solution

Question

I stumble on this problem when I try to restart solnp from the best solution of a previous iteration, where the hessian is incorrect at the issue of the second call.

Here