How to solve this partial derivative which includes a summation?

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
6 replies · 6K views · last reply
JERRY-thechuha
Messages
4
Reaction score
0
I was reading a research paper, and I got stuck at this partial differentiation.

Please check the image which I have uploaded.
Typeset equations (11) to (15): total capacity Ctotal as a sum over i of Bi log2 terms, constraints on transmit powers Pi, and second partial derivatives forming a diagonal Hessian matrix with entries A1 to AM.


Now, I got stuck at Equation (13).
How partial derivative was done, where does summation gone?
Is it ok to do derivative wrt Pi where summation also includes Pi?

Please provide me solution or any reference. Its very urgent.
 
Physics news on Phys.org
The partial derivative being with respect to ##P_i##, there is a single term in each sum that depends on ##P_i##.
 
ok, I got this.

Please tell me is this correct way to solve this question?

Handwritten derivation on lined paper: I = Pr·d sum over j of B_j log2(1 + P_j k_j/(N0 B_j)), differentiated with respect to P_i using a Kronecker delta, giving dI/dP_i and d²I/dP_i².

Anyone please help.
 
Your first derivative is correct, but not your second.

Looking at the original text, I suspect that there is a typo in eq. (13). What is written as
$$
\log_2 \left(\frac{P_i k_i}{B_i N_0} + 1 \right)^2
$$
in the denominator should read
$$
\log(2) \left(\frac{P_i k_i}{B_i N_0} + 1 \right)^2
$$
The same goes for the other ##\log_2##.
 
okk, Thanks for the correction.
I also thought that was a typo for log 2.
Please check again the solution.

Handwritten derivation in a notebook: the mutual information I is differentiated with respect to P_i using a Kronecker delta, giving dI/dP_i = P_T x_i k_i / (N_0 ln 2 (1 + P_i k_i / (N_0 B_i))), then the second derivative.
 
That looks good, taking into account that it is a typo.
 
  • Like
Likes   Reactions: JERRY-thechuha