I'm reading the book by Nocedal and Wright on numerical optimization - http://www.bioinfo.org.cn/~wangchao/maa/Numerical_Optimization.pdf

On page 322 (the general topic here is optimization with one inequality constraint), right after equation 12.18, the text says that when we are in a region where $ c_1(x) > 0$ ($c_1(x)$ is the constraint), the direction ($f(x)$ is the objective function to be minimized) -

$$d = -c_1(x) \frac{\nabla f(x)}{||\nabla f(x)||}$$

satisfies equation 12.18 which is -

$$c_1(x) + \nabla c_1(x)^T d \geq 0$$

This should hold if -

$$ \frac{\nabla c_1(x)^T \nabla f(x)}{||\nabla f(x)||} < 1$$

But I don't see why this should be true. If there were another $$||\nabla c_1(x)||$$ in the denominator, it might have held. Can someone shed light on what I might be missing?


You are right.

As mentioned on point $43$ of the Errata. The denominator should contain an extra factor $\left\| \nabla c_1(x)\right\|$.

I am attaching the Errata of the book.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.