Hi guys I am learning about line search algorithms. In general the idea is that we want to solve the minimization problem

$$\min_\alpha f(x_k+ \alpha p_k)$$

Then we use alpha as our step-size. In practice this is expensive to solve so we get a close enough alpha not nessesary the best. One of the most used conditions to ensure we do decrease the fuchtion are the Wolfe conditions

$$f(\mathbf{x}_k+\alpha_k\mathbf{p}_k)\leq f(\mathbf{x}_k)+c_1\alpha_k\mathbf{p}_k^{\mathrm T}\nabla f(\mathbf{x}_k)$$ and $$-\mathbf{p}_k^{\mathrm T}\nabla f(\mathbf{x}_k+\alpha_k\mathbf{p}_k) \leq -c_2\mathbf{p}_k^{\mathrm T}\nabla f(\mathbf{x}_k)$$

My question is are we guaranteed that among the alphas that the Wolfe conditions we have the actual solution to the minimization problem? To me there is not an obvious answer.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.