Re: convolution with "corrections"

[Gilles Hug <Gilles.Hug@onera.fr>]

Dear John, Alex, Matt & others,

I enclose a tar file of my calculations. Feff version used is 8.00. 
It was compiled on the mac G4 with (nclusx=500) and (lx=2).

I guess I agree with Matt's way of explanations & interpretation. I 
am not aware eiter of convolutions with non-evenly spaced data. 
Should it involve some kind of interpolation somewhere?

I would like to add the following comments:

1/ since the calculation lasted 3 days, I am not happy with re-doing 
the whole stuff with the exchange card. Then it is faster and more 
efficient to use an external procedure. This was the reason why 
didn't used the exchange.

2/ I am NOT using any shift, but only a broadening. Apart from a 
technical point of vue, I don't see the reason for limiting the 
broadening to 1 eV. Some instruments have 1.5-2eV resolution. It is 
also possible to make deconvolution with instrumental response 
function but this is delicate because signal-to-noise pr

3/ The broadening is used to take into account an instrumental 
resolution which I expect to be of gaussian form and not 
energy-dependent. I expect also the broadening due to life-time to be 
of lorentzian shape and energy-dependant.

I think that instrumental broadening should not be modelized through 
potentials.

Best regards,
Gilles

sik.tar