FEMLAB-Exporting Data from Random Distribution of Points?

[WashU]

I am by no means an expert with FEM or FEMLAB. I am running some simple 2D axial symmetric models and would like to export displacement data for each of my subdomains. At the moment I am exporting in cooridnate & data form using a regular grid of points. However for my application the regular grid is not ideal and infact presents a problem. Is there a way to instruct FEMLAB to export coordinates & data from a uniform distribution of random points? Can I also make FEMLAB export more data from in areas with a more refined mesh and less data from areas where the mesh is less defined? Any ideas or help would be greatly appreciated!

 

[PerennialII]

file->export->post processing data should provide you the means you require, for example for subdomains you can use nodal points of a post-processing mesh of what order you wish or then read the coordinates you want output from an external file, or then specify a regular grid of desired density. Defining new output variables and outputting them with desired number of points is a good way as well, and if you use the scripting interface you can do pretty much what you want.

 

[WashU]

Is there a place I can go for information on FEMLAB's scripting interface? I have the FEMLAB manual in paper form but it really doesn't provide a lot of information on data exporting.

 

[PerennialII]

...if I remember about right the improved export feature was part of release 3.2 and generalized the data export features quite a bit, before it was "somewhat" clumsy. Don't know of any online sources for the scripting interface, the scripting manual is probably the best source, although the Yahoo - Comsol group does from time to time treat problems related to Comsol scripts and the archive may contain some examples.

 

[WashU]

Ok I worked out this problem and now have a new one. Is there any method of importing position and displacement data and having femlab determine the applied strain (essentially working the problem in reverse)?

 

[PerennialII]

...yeah, the scripting interface provides you pretty much the same abilities for data I/O as matlab, and then you can use whatever operations to the data you see fit (using the set of build-in function for differentiation for example comes handy ... if you read in the data, fix it to spatial domains and work from there you can use the application mode to do the work for you [like computing the strain state from a displacement field]). A more straightforward way can be to import data from file to for example "tabular-variables" and then work on them straight on if you want to work within the GUI.