Re: [femm] More Batch processing/ Interfaceing

[Dave Squires <djsquires@xxxxxxxx>]

Robin,
I would be interested.
I had to do over 100 runs to get a force contour plot over a large
range of linear motion. There are some things I would need that could
be problematic.
1. Moving the high density mesh boundary area along with the moving
elements can lead to some wierd shapes if not careful. The stretching
action can pull sections of stationary pieces out of shape.
2. Sometimes the stretching action can cut off and isolate regions that will
then come up having undefined properties. Proper setup taking this into
account with redundant property definitions can help avoid this.
3. If femmview is invoked to post process force contour integrations the
integration contour line would need to be moved also.

Well, there is something for you to think about. Maybe the two operations
would have to be separated for now. Get the simulations done first and then
worry about measurements later. I would be happy with just getting iterative
simulations done automatically.

Regards,
Dave Squires


Robin Cornelius wrote:

> Dear group,
>
> I know there was a lot of talk a short while ago about batch processing in
> Femm.
>
> I had my own need for batch processing now and decided to do it my self. I
> have implimented the lua programming language into femm(3.0) and have nearly
> completed the work on the pre-processor.
>
> What this means is you write a lua script file (lua can contain control
> loops, variables, counters, IO, maths routines etc) and call my extention
> functions such as addnode(x,y) directly. All my extention functions are
> essentialy wrapper functions for the real femm code but this is transparent
> to the user.
>
> I am planning to also impelment this in the post-processor and have a way of
> getting feedback from the post processor to the pre-processor so that
> self-refining scripts can be generated or the results in the post-processor
> can be dumped to a file for later analysis and so many hundred runs could be
> proformed, dumping the results each time of only the data you are intrested
> in, into an external file which could then be loaded in to an external
> analysis package matlab, excel etc.
>
> The pre-processor code should be finished later this week and the
> post-processor code by next week.
>
> Is anybody intrested in this as I will make it avaiable if there is any
> demand.
>
> Best Regards,
>
> Robin Cornelius
>
> robin.cornelius@xxxxxxxxxxxxxxxxxxx
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/