The Approved CMS Phi Signal From 900 GeV Data
    By Tommaso Dorigo | January 8th 2010 07:56 AM | 19 comments | Print | E-mail | Track Comments
    About Tommaso

    I am an experimental particle physicist working with the CMS experiment at CERN. In my spare time I play chess, abuse the piano, and aim my dobson...

    View Tommaso's Profile
    Shoot. Today I am on strike.

    This morning I decided to post here an article describing the details of a new result just approved by the CMS collaboration, the observation of a nice signal of phi meson decays. It is a result of which I am quite proud, and although not really a big deal, it is a nice way to start the new year, while we wait for more data from the LHC.

    I had just finished writing the 200-lines piece describing the likelihood fit to the mass distribution, when I decided to save the draft with the "publish" box unmarked, to give it a last reading before submitting it. And the crazy web interface logged me off the site instead!

    I optimistically logged in again, thinking that the "autosaving" feature of the web interface must have recorded basically all of the piece. But I was in for a bad surprise: despite Hank's reassuring statements, the thing only remembers the first five lines of text!

    So I am on strike today. You are still getting the plot, but I will not comment it again -I hate to do things twice! Well, I might comment on it if you insist in the thread below...


    and there is this other bug that your latest posts dont show on the start page (fortunately, google reader can see them) ... this is what you get for going corporate ;-)
    In any case, would love to read your comments

    Very nice! Just one quick (hopefully simple) question on this: how were K candidates chosen for this (i.e. what PID or subdetector signal requirements placed on the tracks, if any)? Of course, I don't expect you to rewrite the poor lost post, but just some idea. From the shape of the background it looks like you've done something to cut out fakes, but nothing extremely strict. Of course, my eye is biased from only working on B-Factory data so far, so looks may be deceiving moving to hadron collider data!

    Hi Brian,

    we selected tracks with a good fit by cutting on the track chisquared, we required them to be sufficiently energetic and pointing to the interaction vertex, and we cut on the observed deposit of energy in the silicon layers if the tracks had momentum lower than 1 GeV, by selecting a band in the dE/dx vs p distribution, centered on the fitted kaon signal band in an independent sample.

    There is more information in a post I wrote recently, "First LHC results", although again not very much detail.

    ha ha ... 'corporate'.   A guy who created this on his couch for science outreach can't really be called corporate.   It is what every scientist says they want - non corporate, independent, for the right reasons, etc.

    But I do hate bugs.   Autosave should refresh once per minute so I am unsure (yet) why you only got 5 lines on a reload.  Even someone as prolific as Tommaso cannot write and lose an entire article in under a minute so I am looking into it now.
    Besides, Jester, it's not like wordpress is the paradise, either!

    No, Hank, I do type fast, but not that fast :) It must have been some sort of a glitch. Very annoying one at that, but then I am to blame, because I had understood long ago that the right thing to do was to edit outside and paste when done in the interface... I sometimes forget my own prescriptions. (I will switch back to typing directly in this interface when you give it xemacs-emulation capability).

    We know what happened now but we don't know why.   Patrick watched the movie Sherlock Holmes recently so I told him to find 'the little Sherlock inside all of us' and see what can be done.

    Basically, you did not log out but your user status still somehow changed.   If it sounds impossible, that's because it is - at least in pristine test environments.   But it is right there in the log so impossible or not, it happened.

    I looked at Wordpress (multi-user) in 2006 when figuring out how to do this.   It was fine for a stand-alone person but not up to a million people a month and thousands of users.    Plus, when something goes wrong on Wordpress there is no one to gripe at - and you all have me!   :)
    Yes, that is what happened. I was clicking the "save" button when I was kicked out. I reloaded the page and the text was gone. Logged in again, and the autosave only showed a few lines of text.

    Probably the autosave got confused by the switch. I do not think it is reliable enough for me to trust it: if I lose 100, 200, or 400 lines of text as once happened, I am not going to write them again -so I prefer an outside editor, since I love each and every post I draft :)

    T - As to the why, I'm guessing your cookie expired or perhaps you cleared your browser cache/history for some reason. In my blogging days I used Pivot as a CMS and for the very now and then I had something similar to yours happen (or maybe just a browser crash) I could sometimes find the post in the cache directory or the user profile temp directory. Worth a check if you lose a real long one. I also got in the habit of copy/pasting anything long into another document as I wrote it...just in case :)

    and why does this appear on but not on ??? Why do your posts always appear first on either one or the other URLs (seemingly randomly) and then take _several days_ to appear on the other one???

    It's on both.  I never look for his stuff on the front page and I don't even visit the categories (I use the updates on my friends list here to know which category articles are in) so it always appears on his column for me.

    There was an issue a few weeks back where one of his articles was made 'sticky' and forced to appear at the top but that was resolved.
    Now I've got the same issue, this time with Microsoft's browser rather than Mozilla's; that is, this post doesn't show on the top, instead it's the previous one.

    Really? I don't see it on (at least yet). Do you??

    It finally arrived today. Why the delay?

    Local ISPs use caching too.    It was here in all places within 3 seconds after it published but we can't control how often ISPs update their cache.  They think it improves speed but it really just increases annoyance.   If you register for an account, that issue goes away.
    This is not an ISP problem. It occurs in different cities (in fact in different countries), at universities and national laboratories as well as public ISPs. It is a problem with It needs to be fixed. is the one that is caching old data, not some grand conspiracy of ISPs across the world.

    could it also be your browser settings? For instance, firefox can be set to check once per session, every time you load a page, or every time it thinks the page has been changed.

    You need to get rid of the grey background and add units to the vertical axis. ;)

    More seriously - why did you take the PDG value for the φ width but not the mass? Our magnetic field calibration is known to be good to better than 0.1%.


    Hi Michael,

    you know, I used to have gRoot->SetStyle("plain") everywhere in my macros, but as of late I have started liking a non-white background in my plots. I know, it's not PRL-ready, so what :)

    We took the width from the PDG because the observed width is not a parameter that may be directly affected by our measurement (in a sensitive fit which uses a convolution of Breit-Wigner and Gaussian as the one above). Instead the mass and the mass resolution are: with the comparison of observed and PDG mass we get access to a check of the momentum scale calibration, and with a measurement of the mass resolution we access information on the track momentum resolution in these early data.

    (In another fit which was given as a check, we also left the width free to vary, and results agree, but of course the large correlation of width and mass resolution yields a larger uncertainty on both).

    I know, we know that the scale is correct and the resolution is ok from 400 million cosmics... But had we taken this stand, we would have left these resonances alone anyway!

    In principle, one might have fixed both gamma, mass, and expected mass resolution (the one we get in the MC). The final fit result, of course, would have been almost indistinguishable. But we would have learned fewer things about how well we are doing... Or rather, we would have had fewer things to show in propaganda presentations!

    Yup, this all makes sense, especially in the context of a public relations plot.

    Fitting the width can be tricky for the phi, since it sits upon a rapidly-rising background. That the mass value comes out right is nice, for sure, and tells us things that the cosmic-ray studies didn't (such as the correctness of the modeling energy loss of kaons in the silicon). If the width also comes out right, then it shows that your techniques are good enough to handle a non-trivial background shape - this would be quite important for any physics study involving φ mesons. Anyway, I see nothing wrong with your choice and presentation.

    On a completely different note, do you have any comment on the problems in Rosarno, which are very much in the news here in the US? See, for example, We all know you are not shy when it comes to such things...