fminsearch question double logistic function

Hello, I have the following double logistic function with 7 parameters.

Y = vb+((k/(1+exp(-c*(t-p))))-((k+vb-ve)/(1+exp(-d*(t-q)))));

where: Y=predicted variable and t=time

the parameters are: k = asymptotic value of Y; vb = Y value at the beginning of the season; ve = Y value at the end of the season; c = slope at the first inflection point; d = slope at the second inflection point; p = date of the first inflection point; q = date of the second inflection point;

the t variable ranges from 1 to 365 with an interval of 1. I have observations of Y at 7-10 different time points for each of my datasets. What I would like to do is to use fminsearch to find the 7 parameters listed above. I also would like to know the error, possibly using a least squares algorithm. However, I am terribly confused on the usage of fminsearch. I know that I have to create the function m file for the double logistic function and call it within fminsearch but I don't know how to call the function nor how to request the parameters to be found. Any help on this would be much appreciated. Thank you. Mutlu..

[John D'Errico]

Sigh. You want to estimate 7 parameters using only 7-10 points? This alone will likely produce complete crap for a result, especially if you have only 7 points. And in that case, knowing the error? You are posing 7 nonlinear equations, in 7 unknowns. A solution if it exists, will exactly interpolate the points. So the "error" would be meaningless in that case.

Worse, of course, you want to do this all using fminsearch. 7 paremeters estimated using fminsearch is near the limits of that algorithms capabilities. It will be somewhat slowly convergent, and require good starting values. Do you have good starting values always?

And, finally, you want to do this with no knowledge or experience at estimation, of how to use an optimizer. And it seems, you barely know how to define a function.

Hey. I wish you good luck. Perhaps what you need to do is spend some time to learn how to use tools like fminsearch to estimate a set of parameters. For example, can you use fminsearch to fit a straight line through a set of points? Learn the simple first. Then proceed to the slightly more difficult. Learn to walk before you try to run.