Re: [OctDev] Fwd: Usage of SECS1D (dimensions of input variables)

[Carlo de Falco]

Hi,

>  Hi,
>
>   I want to use the SECS1D package (v0.0.4) provided with octave-3.0.0
>   [under WinXP in the case that matters]. I have serious problems in
>   guessing which parameter dimensions and types the function  
> DDGnlpoisson
>   [1] takes. Currently I run in "nonconformant arguments"-errors when
>   using standard row vectors as inputs [2]:
>
>
>   Is there a way (apart from trial and error) to find out what  
> dimensions
>   the parameters should be?

All vectors in the input should be given as COLUMN vectors

>   Thanks in advance for hints,
>   Johannes Dorfner

>   [2]: octave script file to test DDGnlpoisson()
>   % ------------------------------------------------------------------
>   % geometry and boundary condition
>   L = 200e-9; % length of the device

Physical quantities should be made non-dimensional before running  
DDGnlpoisson, see the example attached

>   U_0 = 2; % potential of contact 2 relative to contact 1

>   N_v = 1e21*1e8; % density of states [m^-3]

Values for such constants are contained in secs1d/Utilities/constants.m

>   N = 100; % number of grid points

>   Vin=linspace(0,U_0,N); % initial guess for the electrostatic  
> potential

You must consider built-in potential at the boundaries

>   D=linspace(0,0,N); % doping profile; here: intrinsic case

This command will create an empty matrix! I think you meant
D = zeros(size(x))

>   l2=0; % scaled electric permittivity (diffusion coefficient)

l2 is the scaled  electric permittivity, setting it to 0 makes no  
sense physically and it will produce singular matrices

Please see the attached usage example.
You can find more at http://www1.mate.polimi.it/CN/MNME/ 
metodi_me.php3 (unfortunately the website is in italian)

c.

MOS.m
Description: Binary data