forum.alglib.net

Hi,

I'm a newbie at ALGLIB and also to optimization problems.

I was trying to use Levenberg-Marquardt algortihm with the ALGLIB VBA modules to determine the coeficient of the following expression:

Y=sum(1/X(J)*A), being J=1 to N

I have a bunch of points wiht (A(i),Y(i)) i=1 to M , and I want do determine all the X(J) terms that fit the data.

The following code was written by me (for the first time), to fit the equation with (N=10 and M=401), but somehow it's not working.

Can anyone give me some help!?

Public Const M = 401
Public Const N = 10

Public Sub DemoRoutine()
Dim State As MinLMState
Dim Rep As MinLMReport
Dim S(N - 1) As Double
Dim X(N - 1) As Double

Dim t(M - 1) As Double ' array com as idades : eixo x
Dim t0 As Double ' variavel com idade referencia
t0 = Cells(2, 2)

Dim tau(N - 1) As Double

Dim C(M - 1) As Double 'array com os valores da fun??o compliance C(t,t0) :eixo y

Dim soma_c As Double ' variavel do somat?rio dos valores de C(t,t0)
soma_c = 0

' ler arrays da folha aberta
For I = 0 To M - 1
t(I) = Cells(I + 4, 2) ' gravar em array valores de t() come?am na linha 4, coluna 2
C(I) = Cells(I + 4, 15) 'gravar em array valores de C(t,t0) come?am na linha 4, coluna 15
soma_c = soma_c + C(I) 'soma de todos os valores de C(t,t0) para os intervalo t0 a tfinal
Next

For I = 0 To N - 1
S(I) = (I + 1) ^ 3 'valores inicias da solu??o (formula escolhida por mim)

tau(I) = Cells(2, 3 + I) ' gravar em array valores de tau (i)
Next

Call MinLMCreateFJ(N, M, S, State)
Call MinLMSetCond(State, 0, 0, 0.1, 1000)
Do While MinLMIteration(State)
For I = 0 To N - 1
X(I) = State.X(I)
Cells(I + 4, 16) = X(I)
Next

If State.NeedF Then
State.F = 0
For I = 0 To M - 1
For J = 0 To N - 1
State.F = State.F + (1 / X(J) * (1 - Exp(-(t(I) - t0) / tau(J))))
Next
Next
State.F = Square(State.F - soma_c)

Cells(4, 17) = State.F
End If


If State.NeedFiJ Then
For I = 0 To M - 1
State.FI(I) = 0
Next

For I = 0 To M - 1
For J = 0 To N - 1
State.FI(I) = State.FI(I) + 1 / X(J) * (1 - Exp(-(t(I) - t0) / tau(J)))
Next
State.FI(I) = State.FI(I) - C(I)
Next
For I = 0 To M - 1
For J = 0 To N - 1
State.J(I, J) = -(1 - Exp(-(t(I) - t0) / tau(J))) * 1 / X(J) * 1 / X(J)
Next
Next

End If
Loop
Call MinLMResults(State, S, Rep)

For I = 0 To N - 1
Cells(I + 4, 16) = S(I)
Next
End Sub