estimation parameters of equation

Hi all!

I have this relation:

d = M*tao + 1/[x1*(1+exp(G1-G2))+x2(exp(G1-G2))]

where:

d is an array with size Nx1

M is a matrix with size N*64, in each row I have 63 times the number 0 and only once the number 1

G1 has the same size of d

tao is an array with size 64x1

x1,x2,G2 are parameters with size 1x1

I know d,M,G1 and I would like to estimate x1,x2,G2 and tao.

I think that maybe I can use lsqnonlin, but I don't understand how I can do it.

Thank you in advance!

Answer 1

[John D'Errico]

In general, not possible to do.

Suppose that you knew the perfect values for x1, x2, G2?

Then your problem reduces to this:

M*tao = d - 1/[x1*(1+exp(G1-G2))+x2(exp(G1-G2))]

where the right hand side is entirely known.

M*tao simply extracts ONE element of tao for each row of M. If for example, M was a permutation matrix, it would simply permute the entries in tao. But if N is larger than 64, you could never solve the problem exactly. If N is less than 64, there are some values of that can never be known.

So I'm not sure what you are trying to do here, but your problem is poorly posed.

[noemi91d]

Thank you for your reply. N is larger than 64, but I am not looking for the exact solution, I would like to estimate the parameters. Is it possible?