Forbidden link at p.246 in the manual

ROOT
1

Hi,

I get a

You don’t have permission to access /thomasj/searchlimits/ecl.html on this server

while following the link CLs in Chapter 13 p. 246 of the July 2006 version 5.12.

regards,

Gernot

2

Hi Gernot,

You are right - this link is not available now (it was in July). I will check with the author and will try to fix it as soon as possible. Thank you for your report.

Cheers, Ilka

3

Hi Gernot,

The dead link will be replaced by root.cern.ch/root/doc/TomJunk.pdf (just in case you need that information asap).

Cheers, Ilka

4

Hi,

My AFS account expired after 6 months of non-use – I was surprised that the web server could not see my web area! Anyhow, the account is now unblocked and the old link now works. Nonetheless, I cannot guarantee it will remain active forever, so it might be nice to have a more permanent storage of anything there. My NIM article is fine, but I also recommend reading and citing Alex Read’s fine article:

Presentation of search results: The CL(s) technique.
Alexander L. Read (Oslo U.) . 2002.
Prepared for Conference on Advanced Statistical Techniques in Particle Physics, Durham, England, 18-22 Mar 2002.
Published in J.Phys.G28:2693-2704,2002 Also in Durham 2002, Advanced statistical techniques in particle physics 11-21

The main thing on this web site is a relatively simple Fortran routine which I used in the LEP Higgs MSSM combination work, and which computes all of the things talked about in these papers. My web site at the University of Illinois which has the same program is

hep.uiuc.edu/home/trj/mclimit/index.html

This too may not survive forever. I do still use this program and a derivative of it, written in C++ using Root libraries, which does MINUIT fits on each pseudoexperiment, and is thus rather slow (but is important in cases where the signal is smaller than the a priori systematic error on the background – typically a background fit is needed, and two in fact – one assuming a signal and one not) – there’s documentation on that and a program, linked from that page.

I still maintain the Fortran version – and even fixed a bug in it just this month. Many thanks to Laurent Duflot who found that the reweighted versions of the results were biased – I found the problem with the weights and updated the program on both of these pages.

Tom Junk

5

Hi Tom,

Thank you very much for your explanation and provided additional links related to the CL(s) technique. I can confirm that the Internet CL link included in the User’s Guide v5.12 (page 246) is available.

Cheers, Ilka