Discussion Forum

Hi

For me this means you still have some missing BC's and your system is not fulle defined

--
Good luck
Ivar

Hi

For me this means you still have some missing BC's and your system is not fulle defined

--
Good luck
Ivar

Hi Ivar, Thank you for your answer!
There is only one domain (a square) in my model, and the boundary conditions are all Zero Flux, what do you mean of missing BCs?
And, what do you mean for my system is not fully defined? I have set the BCs for all of the boundaries, given the initial condition to the domain, what did the model lack?

ps: I found the coordinates are on or near the boundaries

Hi

COSMOL is set up to solve only coherent BVP so if there are NaNs mostly it is because your equations or Boundary Conditions are not sufficient to make an unique solution, or to define unambiguously the system

--
Good luck
Ivar

I just had the same problem and after a long and tedious debugging process, I found out that the problem was re-defining a variable using units in Matlab.

After saving the mph file as an m file, I have a variable defined as:

model.param.set('alfa', '3.3e4[ohm^-1*m^-1]', 'atenuation');

Later, in a sub-function on my matlab script, I save the value as oldAlfa and set alfa to zero:

oldAlfa=str2double(model.param.get('alfa'));
model.param.set('alfa',0);

After some process (that doesn't involves alfa neither oldAlfa, I reset the old value for alfa:

model.param.set('alfa',[ num2str(oldAlfa,'%6.3e') '[ohm^-1*m^-1]']);

Everything runs fine until the script passes a second time through this sub-function. Then it shows the same error as Xiaoyi Yu says.

The way I handle it was by removing the physical units ('[ohm^-1*m^-1]') in the previous command.

It will be nice if somebody explain us why is this happening.

Cheers,
--
Felipe

Hi

I have not fiddled COMSOL with Matlab since the arrival of v4, no need averytrhing can be done within COMSOL ;) (for me)
but if I read you correctly you are using alpa as a parameter, but for an equation or a parametric sweep "parameter". If it's for the latter then you should avoid units, in up to 4.2a solver sweep parameters should be unitless.

If this is not the case, I would suspect its the 4.0 issue, get hands on a more up to date version, you will be less frustrated, check with your system admin our your local rep

--
Good luck
Ivar

Hi Ivar,

Thanks for your answer. So far I've only used comsol 4.2a, and now I have installed the current latest version (4.2.1.166).

For my model 'alfa' is the conductivity of my PMLs , so its a parameter and (most of the time) is constant.

Some times, when my script finds out the founded eigenvalue does not represents a guided mode (RF module), it calls a subfunctions that "turns off" the PMLs (by setting alfa to zero), finds the effective refractive index and uses it as a target to search modes around this value after resetting alfa (as I explain in my previous post).

So, in conclusion, its not a parametric variable, it is just a parameter that changes only under very specific circumstances.
--
Felipe

Hi

clear if you define it already as an integer then you do not need any uncertainty limit and the mod(1) is one good way ;)

--
Good luck
Ivar

Re: how can we find pull-in voltage of silicon diaphragm?

Hi

It should be describd in the doc as an example (not sure where I read it now) but one way is to solve for a increasing voltage sweep and see when the system collapses or the solver fails, which is rather close to the pull in point, when you increas the voltage gently

--
Good luck
Ivar