Understanding the Derivation of Dual Geometric Programming Problem

106 Views Asked by At

Enthusiastic CS major interested in Optimization Theory here. Pardon me for overlooking something obvious.

I'm referring to this nice tutorial/ebook: http://faculty.uml.edu/cbyrne/optfirst0.pdf

In this, I'm specifically looking at Chapter 3, Section 3.3 (pages 26, 27). I understand the part till where the author applies GAGM (Generalized Arithmetic-Geometric Mean Inequality) to the posynomial. However, I did not understand the reduction from step (3.6) to (3.7). Can anyone help explaining this?

Thanks for reading!

1

There are 1 best solutions below

3
On BEST ANSWER

The statement $(3\cdot 7)$ says: $$g(t)\geq \prod_{j=1}^m\left(\frac{u_j(t)}{\delta_j}\right)^{\delta_j}\tag{3.6}$$

And we know that from $(3\cdot 3)$ in the book: $$u_j(t)=c_i\prod_{i=1}^{n}t_i^{a_i,j} \tag{3.3}$$

if we replace $(3.3)$ in $(3.6)$ we obtain: $$g(t)\geq \prod_{j=1}^m\left(\frac{u_j(t)}{\delta_j}\right)^{\delta_j}= \prod_{j=1}^m\left(\frac{c_i\prod_{i=1}^{n}t_i^{a_i,j} }{\delta_j}\right)^{\delta_j}=\prod_{j=1}^m\left(\frac{c_j}{\delta_j}\right)^{\delta_j}\left(\prod_{j=1}^m\prod_{i=1}^nt_i^{a_{i,j}\delta_j}\right)$$

which gives you $(3.7)$