1
$\begingroup$

Hi everyone I want to solve this problem with Lagrange multiplier method

This is the problem $$ \operatorname{Min} F(x)=3a x_1+5a x_2 $$

subject to these Constraints : $$ g_1 : (2.16/x_1)+(10/x_2)\leq 1 $$ $$ g_2 : x_1 \geq 11.25 $$ $$ g_3 : x_2 \geq 18.75 $$

  • (Note 1: I only know a little about KKT conditions for solving nonlinear problems that has one inequality and one equality constraints and i only seen that in an example , so i don't know much about method)

    (Note 2: This problem is based on weight optimization of a symmetric truss (engineering design problem) that i formulated that and Objective function F(x) and Constraints came from optimization problem formulation method that uses finite element method and reason i'm telling you this because i think you want to know the origin of my question)

$\endgroup$
  • $\begingroup$ What denotes $a$ here? $\endgroup$ – Dr. Sonnhard Graubner Jun 5 at 13:32
  • $\begingroup$ it's a constant that has a value that changes from one problem to another and it's not defined .....if it helps you can give it like 2 or 3 but i need an answer with those if it's possible $\endgroup$ – Ali Eslami Jun 5 at 13:37
  • $\begingroup$ Can we assume that $a$ is positive? $\endgroup$ – Dr. Sonnhard Graubner Jun 5 at 13:42
  • $\begingroup$ yes definitely it's a positive constant because it's a design parameter $\endgroup$ – Ali Eslami Jun 5 at 13:54
  • $\begingroup$ I get $$x_1=\frac{45}{4},x_2=\frac{75}{4}$$ $\endgroup$ – Dr. Sonnhard Graubner Jun 5 at 13:59
1
$\begingroup$

Because $a$ is a positive constant we can ignore it and minimise $$ F(x)=3 x_1+5 x_2 $$

Rewrite the inequalities with slack variables:

$$ \frac{2.16}{x_1}+\frac{10}{x_2}\leq 1 \implies \frac{2.16}{x_1}+\frac{10}{x_2}+s_1^2= 1\\ g_1=\frac{2.16}{x_1}+\frac{10}{x_2}+s_1^2-1 $$ Now the second one: $$ x_1 \geq 11.25\implies x_1 = 11.25+s_2^2\\ g_2= x_1-11.25-s_2^2 $$ Finally, the third one: $$ x_2 \geq 18.75 \implies x_2 = 18.75+s_3^2\\ g_3 = x_2 - 18.75-s_3^2 $$ Form the Lagrangian:

$$ \begin{split} L(x_1,x_2,\lambda_1,\lambda_2,\lambda_3,s_1,s_2,s_3) &= 3 x_1 + 5 x_2 \\ &- \lambda_1 \left(\frac{2.16}{x_1}+\frac{10}{x_2}+s_1^2-1\right) \\ &- \lambda_2 \left(x_1-11.25-s_2^2\right) \\ &- \lambda_3 \left(x_2 - 18.75-s_3^2\right) \end{split} $$ This implies the following partial derivatives: $$ \begin{split} \frac{\partial L}{\partial x_1} &= 3+\frac{2.16\lambda_1}{x_1^2}-\lambda_2\\ \frac{\partial L}{\partial x_2} &= 5+{{10\lambda_1}\over{x_2^2}}-\lambda_3\\ \frac{\partial L}{\partial s_1} &= -2s_1\lambda_1\\ \frac{\partial L}{\partial s_2} &= 2s_2\lambda_2\\ \frac{\partial L}{\partial s_2} &= 2s_3\lambda_3 \end{split} $$

$\endgroup$

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.