You are given a function in n-dimensional space,
Which might have a lot of arguments.
You know that this function has a physical sense,
You should maximize it subject to given constraints.
Refrain
You will take the derivative
For any choice of the cost functional
And you will find the maximum
You’ve been searching for a long-long time.
If constraints are specified
On the set of admissible arguments,
Then just write the Lagrangian
And find zeroes of its gradient.
You know that in any maximum point
The derivative has a zero value
But you need to check that the Hessian
Is negative definite, you’ll do it, you can.
Refrain.
The gradient method requires to compute
The gradient value, and it will be so cute.
And Newton needs to linearize,
Then the speed of convergence will substantially rise.
Refrain.
Слушать песню: http://www.realmusic.ru/songs/1437826