optimal derivative position through optimization

54 Views Asked by At

So I have the following optimization problem:

min. $-E^Q[u(h(x))]$

s.t $\int h(x)q(x)dx \leq \frac{V_0}{B_0}$

Where $Q$ is the subjective probability which then gives:

$E^Q[u(h(x))]=\int u(h(x))p(x)dx$

Thus the lagrangian becomes:

$L(X,\lambda)=-\int u(h(x))p(x)dx+\lambda \int h(x)q(x)dx + \lambda \frac{V_0}{B_0}$

So If I remember correctly one takes the derivative of, in this case, $X$ and $\lambda$ then equal the derivative to $0$.

I would obviously get $2$ equations and from there solve the optimization problem however I do get one equation correctly after taking the derivative but I don't manage with the derivative with respect to $X$.

$\frac{dL}{d\lambda}=\int h(x)q(x)dx +\frac{V_0}{B_0}=0$

The derivative with respect to $X$ however I don't get it to be correct. The correct outcome should be, according to the book.

$\frac{dL}{dX}=\int(u'(h(x))p(x)-\lambda q(x))(g(x)-h(x))dx=0$

The $g(x)$ isn't defined which kinda makes me think that I might have missed something...can't see what though.