Revised development version

[David Meeker <dmeeker@xxxxxxxx>]

I've put up another revision of the development version of femm 3.3. 
This one is called 3.3alpha2. See:
http://femm.berlios.de/femm33src.zip
http://femm.berlios.de/femm33bin.exe

In the 3.3alpha1 release, I had inadvertently introduced a bug into the 
harmonic axisymmetric while changing it to do the nonlinear time 
harmonic solutions which made that solver give the wrong results (oh, 
well--3.3a1 /is/ an alpha version...) If you've been using this part of 
the 3.3a1 solver, you should definitely get the 3.3a2 version instead.

Several modifications have been made since the 3.3a1 version:

/Can now abort running lua scripts/. Any lua script or lua command 
entered from the console can be aborted by hitting the <Break> key. In 
addition, the pre- and post- processor don't "freeze up" anymore when a 
script is running (This means, for example, that any print statements in 
your lua scripts get immediately displayed in the lua console, rather 
than waiting until the end of the run). For those of you that are 
interested in how this works, I implemented a "linehook" function (a 
feature built in to lua) that gets called every time a line in a lua 
script or command gets processed. This linehook function processes any 
messages that might be on the queue, so that the display doesn't freeze 
up. If the linehook function detects that <Break> has been pressed, it 
calls the lua_error function, which aborts the evaluation in progress.

/Improved lamination model for nonlinear time harmonic problems./ 
Version 3.3a1 used a kludge where the permeability at the surface of a 
nonlinear lamination was simply plugged into the linear formula for 
effective lamination permeability. Version 3.3a2 instead solves a series 
of 1D nonlinear time harmonic finite element problems for each 
laminated, nonlinear material (one at each level of H defined in the B-H 
curve) to get the "right" effective BH curve for the laminated material, 
including the effects of hysteresis and eddy currents. If you're 
interested in checking out the 1D solution at each point, it's 
implemented in the function:
CComplex CMaterialProp::LaminatedBH(double w, int i)
at the end of matprop.cpp in the fkn source code.

/Now using the latest version of Triangle./ A new version of the 
Triangle mesh generator came out a couple of months ago. The 3.3a2 
version included this latest version of the Triangle code.

Anyhow, I'd be curious as to comments and bug reports about this latest 
revised development version. Not that since I've been monkeying around 
with the way that the time harmonic solver works, there could again be 
bugs--if you value stability, stick to the 3.2 version for now.
--
David Meeker
email: dmeeker@xxxxxxxx
www: http://femm.berlios.de/dmeeker