Can I provide an initial guess for the perfect foresight solver other than the initval block?

I want to solve the transition path when the state variable (not predetermined) starts from x in period 1. I do this by setting the initial value of the state variable at 0 and adding a shock equal to x in period 1. But the perfect foresight solver fails. I attach my code below. When the initial value of the state variable is set to zero, Dynare actually found the correct solution but reported a failure. Donâ€™t know why.

It seems that 0 cannot be used as an initial guess for the state variable in my model. When I use other initial values, say y>0, the solver works fine. But now the state variable is not x in period 1.

Is there a way to set the initial value of the state variable at 0, but give another initial guess for the perfect foresight solver?

model.mod (955 Bytes)