sensitivity during optimisation, with fun argument

Hi

when optimising with fminbnd I find that the y value may vary due to experimental sensitivity. On reading I find the 'fun' argument may reduce plot sensitivity. Am I correct and will someone please instruct me on how to accomplish this?

Gavin.

[Torsten]

What do you mean by "plot sensitivity" ? What are you trying to plot ?

Best wishes

Torsten.

[Gavin Seddon]

Hello, my X data is time and my Y data is structure deviation. Using fminbnd it appears the maximum (using -f) will be infinite and I must define a deviation limit.

[Gavin Seddon]

Is it possible rather than to find a optimum point use the function equation by finding its derivative, setting this equal to 0 and solving for X? My problem would now be to define a position on the curve at which Y variations become negligible.

Answer 1

[Gavin Seddon]

Thank you for your input, during re-reading the manual it appears that 'smoothing' in chapter 2 will accomplish this. My original data has minor fluctuations. Thanks again.

Gavin.