wz 发帖数: 1 | 1 There are three parameters, alpha, L1, L2, to be determined optimally, so that
the following quantities (n = 1, 2, 3, …N, where N is a given natural number)
S_n = S_n(n, alpha, L1, L2)
R_n = R_n(n, alpha, L1, L2)
H_n = H_n(n, alpha, L1, L2)
Satisfy
1. R_n > R (R is a given value, n = 1,2,3,…N), and the larger the better for
R_n;
2. S_n > 0 (n = 1,2,3,…N);
3. dh_n = h_(n+1) – h_n > dh (dh is a given value), and the larger the better
for dh_n;
4. R_N+S_N is minimized, and the shorter the better
Th | f**n 发帖数: 401 | 2 My 2 cents:
First of all, your problem is not clearly stated. You need to formalize your
problem into a nonlinear optimization problem. That is, you want to minimize
certain objective function under certain constraints, rather than vaguely
state "the larger the better" or something like that. If there are more than
one objectives, for example you may prefer larger R_n, you
should make some tradeoff among them. For instance you can set up the
objective function as R_N+S_N-\sum c_n*R_n, in which
【在 wz 的大作中提到】![](/moin_static193/solenoid/img/up.png) : There are three parameters, alpha, L1, L2, to be determined optimally, so that : the following quantities (n = 1, 2, 3, …N, where N is a given natural number) : S_n = S_n(n, alpha, L1, L2) : R_n = R_n(n, alpha, L1, L2) : H_n = H_n(n, alpha, L1, L2) : Satisfy : 1. R_n > R (R is a given value, n = 1,2,3,…N), and the larger the better for : R_n; : 2. S_n > 0 (n = 1,2,3,…N); : 3. dh_n = h_(n+1) – h_n > dh (dh is a given value), and the larger the better
| m******k 发帖数: 139 | 3 Use fmincon() in Matlab.
For C/C++:
CFSQP from U. of Maryland.
Opt++ from a national lab.
than
decided
that
number)
for
better
and
【在 f**n 的大作中提到】![](/moin_static193/solenoid/img/up.png) : My 2 cents: : First of all, your problem is not clearly stated. You need to formalize your : problem into a nonlinear optimization problem. That is, you want to minimize : certain objective function under certain constraints, rather than vaguely : state "the larger the better" or something like that. If there are more than : one objectives, for example you may prefer larger R_n, you : should make some tradeoff among them. For instance you can set up the : objective function as R_N+S_N-\sum c_n*R_n, in which
|
|