n_TOF Au data analysis

-- AlbertoMengoni - 15 Jul 2008

Dear Nicola and Cristian,

> Dear Cesar,
> we are very glad that things have improved considerably now.
> After reading your report, there are still a few points that have to
> be worked out before we can make a final comparison. Here are some comments and suggestions:
>
> - for the first saturated resonance, we believe we should still agree
> on how to make the fit. As you can read from a note here attached, the
> kernel may depend on the choice we make for the fit. On the contrary,
> the normalization stays constant no matter what we do. For this
> reason, we suggest to fit the first resonance with whatever parameter
> you like (or with fixed Gg and Gn, but only in the region of the
> saturation), and use the normalization constant instead of the kernel.

That the normalization constant does not depend on how you do the fit, I agree completely (this I showed in Bari already). But that is good, because then you are safe concerning normalization for the rest of higher energy resonances, even if your resonance parameters for that resonance at 4.9eV would not be OK. I agree with you also on that accurate information on this 4.9eV RP could be gained in combination with transmission. But on the other hand, since at this resonance it is quite clear that Gg >> Gn, I think it is quite fair to fix Gg=124meV (the exact value of Gg will not affect the fit anyway), the normalization is fixed also, and the resulting Gn is then quite sensitive to the capture area. Which on the other hand is in agreement with the transmission measurement of Block et al., so that everything then looks consistent for this resonance.

>
> - for all other resonances, we noticed that you have keept fixed the
> Gg from ENDF even in cases in which this is much smaller than Gn (thus
> determining the kernel). This is not the same strategy we are
> following. Clearly in your case the agreement with ENDF is better,
> expecially above 1 keV, but perhaps this is not the best way to proceed.

In the cases that I keep Gg fix is because the fit does not improve noticeably when I let Gg vary instead of Gn or both of them. There are however many cases where one really needs to vary the other parameter (Gg instead of Gn) or both of them (Gg and Gn) if one wants to improve the Chi2 of the fit. And this I also do.

You can see the parameters that I let vary by looking at the uncertainties of the parameters in the table.

Of course I agree that the best in order to make a comparison TAC-C6D6 is that we both try to vary the same parameters in each case. However in the Status report of Cristian (from March 3rd) he gets the same average deviation vs. ENDF when he fixes Gg=124meV (dev.=2.3%) as when he lets both Gg and Gn to vary (dev.=2.3%). (I talk about the deviation of the histogram, and not the deviation of the fitted gaussian distribution, because from one case to another one actually changes the shape of the discrepancies-distribution, but not the discrepancies themselves). It is strange that this discrepancy for TAC is maximum using the "standard" procedure (avg. dev. = 7.3%), which should be more similar to what I have made in the analysis of the C6D6...

>We suggest two possible ways to proceed: i) we leave all parameters
>free for all resonances, and then make the comparison; in alternative
>ii) we exchange the par and input files, and we use the same strategy
>on all our data. Any suggestion from your side would be appreciated.

I rather prefer the approach ii), so let us exchange par and pub files.

Another thing is the number of resonances, in the TAC plots I always see 249 or 250 entries, whereas ENDF has 263 entries, and I always show all these 263 resonances. So we should really compare the same amount of resonances.

>
> We agree on your idea of a video-conference next week (or even
> this week). Perhaps we could first make the test we suggested above,
> so we can converge more quickly.
>

Best regards,

Cesar.

> Ciao
> Cristian and Nicola
>
> ______________________________________________________________________
> ____
> Nicola Colonna
> INFN - Sezione di Baritel: (+39) 080 544 2351 V. Orabona, 4 fax:
> (+39) 080 544 2470
> 70126 Bari (Italy) E-mail: nicola.colonna@baNOSPAMPLEASE.infn.it
>
>

--

Dear Cesar,

we are very glad that things have improved considerably now. After reading your report, there are still a few points that have to be worked out before we can make a final comparison. Here are some comments and suggestions:

* for the first saturated resonance, we believe we should still agree on how to make the fit. As you can read from a note here attached, the kernel may depend on the choice we make for the fit. On the contrary, the normalization stays constant no matter what we do. For this reason, we suggest to fit the first resonance with whatever parameter you like (or with fixed Gg and Gn, but only in the region of the saturation), and use the normalization constant instead of the kernel.

* for all other resonances, we noticed that you have keept fixed the Gg from ENDF even in cases in which this is much smaller than Gn (thus determining the kernel). This is not the same strategy we are following. Clearly in your case the agreement with ENDF is better, expecially above 1 keV, but perhaps this is not the best way to proceed. We suggest two possible ways to proceed: i) we leave all parameters free for all resonances, and then make the comparison; in alternative ii) we exchange the par and input files, and we use the same strategy on all our data. Any suggestion from your side would be appreciated.

We agree on your idea of a video-conference next week (or even this week). Perhaps we could first make the test we suggested above, so we can converge more quickly.

Ciao Cristian and Nicola

Nicola Colonna INFN - Sezione di Bari tel: (+39) 080 544 2351 V. Orabona, 4 fax: (+39) 080 544 2470 70126 Bari (Italy) E-mail: nicola.colonna@baNOSPAMPLEASE.infn.it

Dear all,

I hope you had a nice time in Vienna. Please find attached a short report on the status of the gold (n,gamma) c6d6 data analysis. In summary I have found a mistake in the way I was applying before the threshold correction factor (the correction was actually not applied). Now the new results are 5.5% higher, and hence quite close to ENDF. Still substantial discrepancy with TAC, at least regarding the last TAC data I have from 28th February. Perhaps Cristian has more recent results to compare with? If you have time, probably we could make (video)-conference next week? Best regards,

Cesar.

GSI Planckstraße 1 64291 Darmstadt GERMANY Telf. +49 6159 71 2434

• normalization.pdf: This is the document sent in by Cesar on the data normalization.