Gradient method problems - steepest conjugate etc.

trante · #1 $Old$ January 1st, 2009, 04:01 AM

I have three questions as a take home exam in my course.

Do you know the answers?

(Problems in the jpeg file)

mr fantastic · #2 $Old$ January 1st, 2009, 04:18 AM

Quote:

Originally Posted by trante $View Post$

I have three questions as a take home exam in my course.

Do you know the answers?

(Problems in the jpeg file)

Has your instructor said you can get help with this take home exam? The thread is closed pending further information (you can pm me).

mr fantastic · #3 $Old$ January 1st, 2009, 05:48 AM

Quote:

Originally Posted by trante $View Post$

I have three questions as a take home exam in my course.

Do you know the answers?

(Problems in the jpeg file)

Received via pm:

Quote:

Originally Posted by trante

She noted that we can even ask to our graduate friends for help. This work will effect our grades %5.
Thanks for your interst.

[snip]

I have re-opened the thread.

TheMasterMind · #4 $Old$ January 1st, 2009, 07:23 AM

Quote:

Originally Posted by trante $View Post$

I have three questions as a take home exam in my course.

Do you know the answers?

(Problems in the jpeg file)

2. get $x_{k+1}$ from $x_k$ you must use $Q$ to evaluate $g_k$ and $Qg_k$

trante · #5 $Old$ January 2nd, 2009, 02:41 PM

*Q1 Show the direction vectors are perpendicular to each other in the Steepest Descent Method;
that is $g_{k+1}^t * g_k=0$

I know that in steepest descent, the algorithm makes zigzags.
If one vector is A and the other is B.
And the angle between them is $\theta$ .
In vector multiplication $A*B=AB*cos \theta$
Because of $cos(90)=0$ the answer is 0.

But how can i relate it with steepest descent??