Asymetric gain profil

[Yoann_CELIA]

Dear Thomas,

I have a question concerning fiberdesk. I am still using the 2006
version and I would like to know if it is possible for you to add a
new gain profile in the gain profile list. Well we are trying to
simulate amplification with gain shaping and gain narrowing and I
would like to put an asymetric function. In fact I would need a
function with the left side (short wavelength) looking like a sech²
and the right side looking like a Lorentzian with a width we can
adjust ... Do you see what I mean ? I already compute this function in
C and here is how it looks like:

if (x<P3)
y = 1/(cosh((x-P1)/(0.45*P2/2.041475)));
else
y = 1/(1+(2*(x-P1)/(0.55*3.5*P2))2);

P1 is the centrale wavelenght
and P2 is the FWHM

Do you think you could do that ??

Regards from Bordeaux

Yoann

[webm...@fiberdesk.com]

Dear Yoann,

the next version will feature this asmmetric gain profile as well as
multiple gain peaks. We already discussed the update of your version
and I will inform you, when it is ready.

Best regards,

Thomas

[webm...@fiberdesk.com]

Dear Yoann,

just one more question about the asymmetric profile: The value for
width and central wavelength in FiberDesk will be given in nm,
however, wouldn't it be better to compute the function internally in
frequency space, as it is done for all other profiles? It has no
effect for small peaks but for broad ones ... what do you think?

BR

Thomas

[Yoann_CELIA]

Dear Thomas,

We agree on the fact that the functions have to be implemented in the
w space (the piecewise
asymetric parametrization can be kept identical to that given in
lambda except that axis is inverted).
However, if the interface can ask for values in lambda (lambda0,
Delta_lambda) even if you translate it
in w (deltaw = 2cPi/(1/lambda_min-1/lambda_max) so that it accounts
for large bandwidth.

So the function to be computed would be:

if (x<P3)
y = P1+P2/(1+(2*(x-P3)/(0.55*3.5*P4))^2);
else
y = P1+P2/(cosh((x-P3)/(0.45*P4/2.041475)));

If you have any other question please feel free to ask.

Greatings to you

Yoann

[Yoann_CELIA]

Dear Thomas,

You probably corrected by yourself that the real function to compute
is:

if (x<P1)
y = 1/(1+(2*(x-P1)/(0.55*3.5*P2))^2);
else
y = 1/(cosh((x-P1)/(0.45*P2/2.041475)));

Thanks again,

Yoann

PS: Claude warmly says hello to Anna-Katherine !!