Optimisation without the toolbox!

Hi all,

I am looking to start building a function which goal seeks along a given boundary. I have some solution data from an FEA software with relative velocity data. I want to sort through the data and find the maximum value possible from an boundary box of say [50,2].

How would I begin to start writing a code which sorts through the data and finds the maximum?

Many thanks, chris

Answer 1

[Matt J]

It's not clear to me why you wouldn't simply use the max() function if you're just maximizing over a set of samples that you already have.

If instead you need to find the max. over the continuous region of the boundary box then you could use fminsearch or fminbnd which don't require more than basic MATLAB. However, you would need to decide on an objective function and in particular how it will interpolate your FEA samples over the box boundary. The interpolation code will depend on whether the FEA data is gridded or scattered.

[Christopher Hall]

Hi Matt,

Thanks for taking the time to respond. I haven't seen those functions yet so I will definitely take a look. You are correct in assuming that I am looking at a region of data within a "bounding box". I am working on functions to define the regions but they aren't coming very naturally as the data i am dealing with is scattered node data.

[Matt J]

You might look at scatteredInterpolant() then as well.

[Christopher Hall]

That looks pretty spot on. Any ideas how I could analyse a group of x,y vectors using this function?

[Matt J]

Not without knowing what objective function of those vectors you're trying to optimize

[Christopher Hall]

I want to assign regions to specific nodes within the geometry depending on their global coordinate position. So almost like building a library of sector data which can be queried to find the sector with max(u).

The data has 4 columns [X,Y,u,v].