variable declaration for optimization

hello,
I am writing a code where I need to declare two variables which will be used for optimization in further, i.e,
a,b -----> variable(a=complex, b=real)
x=h1*a^2+h2*b^2+h3;------>h1,h2,h3 are constants
x=objective of optimization problem
what will be the code to declare 'a' and 'b';?

4 comentarios

Walter Roberson
Walter Roberson el 31 de Oct. de 2020
Are you doing Problem Based Optimization?
Soumili Sen
Soumili Sen el 31 de Oct. de 2020
no,solver based...
Walter Roberson
Walter Roberson el 31 de Oct. de 2020
You do not define variables for solver based optimization: you define a function that accepts a vector of values.
Soumili Sen
Soumili Sen el 12 de Nov. de 2020
okk.. thanks

Iniciar sesión para comentar.

 Respuesta aceptada

Ameer Hamza
Ameer Hamza el 31 de Oct. de 2020
Only specify them as input of function handle. Normal procedure is something like this
h1 = 2;
h2 = 3;
h3 = 1;
x = @(a, b) h1*a^2+h2*b^2+h3;
objFun = @(p) x(p(1),p(2));
MATLAB optimization functions expect that the function handle only accepts one multi-dimensional input. Therefore, we converted 'x' with two scalar inputs to objFun with a single 2-dimensional input vector. Then use an optimization solver, such as fmincon()
sol = fmincon(objFun, rand(2,1))

10 comentarios

Soumili Sen
Soumili Sen el 31 de Oct. de 2020
Thanks... But what about the complex value of the variable?
Ameer Hamza
Ameer Hamza el 31 de Oct. de 2020
What do you mean by the complex value of variable? Is one of the variable complex number?
Soumili Sen
Soumili Sen el 31 de Oct. de 2020
yes,one veriable is complex number
Ameer Hamza
Ameer Hamza el 31 de Oct. de 2020
Can you show your optimization problem?
Soumili Sen
Soumili Sen el 31 de Oct. de 2020
This is my code-
hr = sqrt ( .5/ 2 ) .* (randn(1)+j*randn(1));
hr = sqrt ( .3/ 2 ) .* (randn(1)+j*randn(1));
n1 = sqrt ( .1/ 2 ) .* (randn(1)+j*randn(1));
constellation = sqrt(1/2)*[1+1i*1 1-1i*1 -1-1i*1 -1+1i*1];
cons_size = size(constellation);
x = constellation(randi([1 4],1,100));
r=@(b) (hr*x+hr1*b+n1); % output ------> b is optimization variable with real value
r1=@(a, b) a*r;------> a is optimization variable with complex value
rr =r1;
rr(1:2:end) = 0; % odd data of the symbol-----> objective
Ameer Hamza
Ameer Hamza el 31 de Oct. de 2020
The objective function must return a scalar number for the minimization algorithm to work. Instead of code, can you attach an image showing the mathematical formulation of your problem?
Soumili Sen
Soumili Sen el 31 de Oct. de 2020
yahh u are right,matlab shows an error like.............. "Supplied objective function must return a scalar value." how I can solve that.
Ameer Hamza
Ameer Hamza el 31 de Oct. de 2020
Can you show the mathematical formulation of your objective function. Currently I don't understand how is it defined.
Walter Roberson
Walter Roberson el 31 de Oct. de 2020
Separate the real part of the complex variables from the imaginary part. Return the norm of the expression.
Soumili Sen
Soumili Sen el 12 de Nov. de 2020
Thank you both of you.

Iniciar sesión para comentar.

Más respuestas (0)

Categorías

Etiquetas

Preguntada:

el 31 de Oct. de 2020

Comentada:

el 12 de Nov. de 2020

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by