I want to solve this equation for a and b where A and B is a vector data with dimension (1344x1) A=a(B)^b

Here; A = data1 ; B = data2

y = data1;
Y = log(y);
A = [ones(length(y),1) log(data2)'];

X1 = inv(A'*A);
X2 = A'.*Y;
X = X1*X2;

a = exp(X(1:1,1));
b = X(2,1:1);

this is the method I am using to solve for this equation. But I am not sure this is the right way to solve nonlinear equations. Can someone suggest an alternate way with an example?

Thank you

Answer 1

[Walter Roberson]

p = polyfit(log(B), log(A), 1);

a = exp(p(2));
b = p(1);

This would at least be a good starting point for a search in linear space (it should be close to as good as you can get in log space.)