MPX Dcors as LHC Luminosiy Monior arxiv:5.4v [physics.ins-d] 5 Dc 5 André Sopczak, Babar Ali, Ndaa Asbah, Bndik Brgmann, Khald Bkhouch 3, David Caforio, Michal Campbll 4, Erik Hijn, Claud Lroy, Anna Lipniacka 5, Marzio Nssi 4, Sanislav Pospíšil, Frank Sifr, Jaroslav Šolc, Paul Souid, Michal Suk, Danil Turčk, Zdněk Vykydal Insiu of Exprimnal and Applid Physics, Czch Tchnical Univrsiy in Pragu, Czch Rpublic Group of Paricl Physics, Univrsiy of Monral, Canada 3 Faculé ds Scincs d la Tchnologi, Univrsié Mohammd Khidhr d Biskra, Algria 4 CERN, Swizrland 5 Dparmn for Physics and Tchnology, Brgn Univrsiy, Norway Absrac A nwork of 6 Mdipix- (MPX silicon pixl dvics was insalld in h ATLAS dcor cavrn a CERN. I was dsignd o masur h composiion and spcral characrisics of h radiaion fild in h ATLAS xprimn and is surroundings. This sudy dmonsras ha h MPX nwork can also b usd as a slf-sufficin luminosiy monioring sysm. Th MPX dcors collc daa indpndnly of h ATLAS daa-rcording chain, and hus hy provid indpndn masurmns of h bunch-ingrad ATLAS/LHC luminosiy. In paricular, h MPX dcors locad clos nough o h primary inracion poin ar usd o prform van dr Mr calibraion scans wih high prcision. Rsuls from h luminosiy monioring ar prsnd for daa akn a proon-proon collisions. Th characrisics of h LHC luminosiy rducion ra ar sudid and h ffcs of bam-bam (burn-off and bam-gas (singl bunch inracions ar valuad. Th sysmaic variaions obsrvd in h MPX luminosiy masurmns ar blow.3% for on minu inrvals. Prsnd a h IEEE 5 Nuclar Scinc Symposium, San Digo, USA
MPX Dcors as LHC Luminosiy Monior André Sopczak, Snior Mmbr, IEEE, Babar Ali, Ndaa Asbah, Bndik Brgmann, Khald Bkhouch, David Caforio, Michal Campbll, Erik Hijn, Fllow IEEE, Claud Lroy, Mmbr IEEE, Anna Lipniacka, Marzio Nssi, Sanislav Pospíšil, Snior Mmbr, IEEE, Frank Sifr, Jaroslav Šolc, Paul Souid, Michal Suk, Danil Turčk, Zdněk Vykydal Absrac A nwork of 6 Mdipix- (MPX silicon pixl dvics was insalld in h ATLAS dcor cavrn a CERN. I was dsignd o masur h composiion and spcral characrisics of h radiaion fild in h ATLAS xprimn and is surroundings. This sudy dmonsras ha h MPX nwork can also b usd as a slf-sufficin luminosiy monioring sysm. Th MPX dcors collc daa indpndnly of h ATLAS daa-rcording chain, and hus hy provid indpndn masurmns of h bunch-ingrad ATLAS/LHC luminosiy. In paricular, h MPX dcors locad clos nough o h primary inracion poin ar usd o prform van dr Mr calibraion scans wih high prcision. Rsuls from h luminosiy monioring ar prsnd for daa akn a proon-proon collisions. Th characrisics of h LHC luminosiy rducion ra ar sudid and h ffcs of bam-bam (burn-off and bam-gas (singl bunch inracions ar valuad. Th sysmaic variaions obsrvd in h MPX luminosiy masurmns ar blow.3% for on minu inrvals. Indx Trms Pixl dcors, Luminosiy, LHC, Mdipix I. INTRODUCTION A comprhnsiv analysis of daa akn by a nwork of Mdipix- (MPX dvics as a slf-sufficin luminosiy monioring sysm is prsnd. Th MPX nwork was insalld a diffrn locaions in h ATLAS dcor [] a CERN and in is cavrn. Th MPX dvics ar basd on h Mdipix- hybrid silicon pixl dcor which was dvlopd by h Mdipix- Collaboraion []. I consiss of a cm silicon snsor marix of 56 56 clls, bump-bondd o a radou chip. Each marix lmn (55 55 m pixl, 3 m hick is conncd o is rspciv radou chain ingrad on h radou chip. Puls high discriminaors drmin h inpu nrgy window and provid nois supprssion. A counr in ach pixl rcords inracing quana of radiaion, phoons, nurons, lcrons, minimum ionizing paricls, and ions wih nrgy dposis falling wihin h prs nrgy window [3], [4]. A. Sopczak, B. Ali, B. Brgmann, D. Caforio, E. Hijn, S. Pospíšil, F. Sifr, J. Šolc, M. Suk, D. Turčk, and Z. Vykydal ar wih h Insiu of Exprimnal and Applid Physics, Czch Tchnical Univrsiy in Pragu, Horska 3a, CZ, Czch Rpublic (-mail: andr.sopczak@crn.ch. N. Asbah, C. Lroy, and P. Souid ar wih h Group of Paricl Physics, Univrsiy of Monral, 9 boul. Édouard-Monpi, Monréal QC H3T J4T, Canada. K. Bkhouch is wih h Faculé ds Scincs d la Tchnologi, Univrsié Mohammd Khidhr d Biskra, BP 45 RP, 7 Biskra, Algria. M. Campbll and M. Nssi ar wih CERN, CH Gnva 3, Swizrland. A. Lipniacka is wih h Dparmn for Physics and Tchnology, Univrsiy of Brgn, Allégan 55, N-57 Brgn, Norway. Th ATLAS and CMS collaboraions hav labora sysms of luminosiy masurmns, dscribd in [5] and [6], rspcivly. A comparaiv sudy of hir rsuls and h MPX luminosiy monioring is byond h scop of his aricl. Th mhods and chniqus dscribd in his aricl wr pionrd for high-prcision luminosiy drminaion for h daa-aking priod a h LHC. Ths chniqus and analysis mhods ar basd on prcision couning of paricls passing h snsors. Similar rquirmns apply o analysis chniqus usd in ohr filds of rsarch (.g. mdical applicaions and spac scinc whr high prcision and long-rm im-sabiliy of masurmns ar ndd. MPX dvics hav alrady bn succssfully applid in hs aras of rsarch and h fundamnal sudis prsnd in his aricl can lad o furhr advancs in hir applicaion. Th capabiliy of h MPX dvics for luminosiy monioring has bn invsigad bfor [7]. I is shown in his aricl ha h MPX nwork is slf-sufficin for luminosiy monioring. In paricular, van dr Mr (vdm scans [] can b usd for absolu luminosiy calibraion. Daild analysis of h MPX daa allows h quanificaion of h long-rm sabiliy ovr on yar of daa-aking. I also provids shor-rm (minu by minu prcision. This informaion is crucial o valua h prformanc of h MPX nwork as a luminosiy monioring sysm. Th dcion of chargd paricls in h MPX dvics is basd on h ionizaion nrgy dposid by paricls passing hrough h silicon snsor. Th signals ar amplifid and cound during an adjusabl im window (fram for ach pixl. Nural paricls, howvr, nd o b convrd o chargd paricls bfor hy can b dcd. Thrfor, a par of ach silicon snsor is covrd by a 6 LiF convrr. Th MPX pixl dcor can b oprad in racking or couning mod [3], [4]. Evry pixl rcords h numbr of his wihin an adjusabl im inrval (acquisiion im. On of h imporan faurs of h MPX dvics is h abiliy o rcord and idnify clusrs. Clusrs ar dfind as parns of adjacn pixls wih nrgy dposis dfind in [4, Sc..]. Diffrn paricls ha ravrs h dvic caus diffrn clusr shaps. Ths shaps allow paricl idnificaion and h disincion bwn kv-mv lcrons, phoons, nrgic hadrons, alpha paricls and ion-fragmns. Th nrgy dposid during h acquisiion im can b simad as wll. Th daa is sord fram-by-fram. Afr h daa acquisiion is closd, i aks abou 6 s o ransmi h saus of h full 65536 pixl marix. Th dvic is no
snsiiv during h radou procss (dad-im. Th primary goal of h MPX nwork was o provid informaion on h radiaion composiion wihin h ATLAS cavrn (including h hrmal nuron componn. Th MPX dcors also allow h masurmn of bam-inducd radioaciviy during and afr collisions. This ral-im masurmn of h LHC-gnrad background radiaion prmis h validaion of background radiaion simulaion sudis. Rsuls from - MPX daa-aking hav bn rlasd [4, Sc. 4]. Th us of h MPX nwork for rlaiv luminosiy masurmns in proon-proon collisions proposd in [3] is sudid hr in dail. Thirn ou of h sixn insalld dvics wr usd for h analysis. Two dcors wr noisy du o radiaion damag. On dcor was locad oo far away from h inracion poin. Tabl I liss h locaions of h dcors and numbr of rgisrd vns (clusrs pr uni snsor ara and pr uni ingrad luminosiy. Figur shows an xampl of h luminosiy from hi couning masurd wih MPX for LHC fill 649. Th papr is srucurd as follows. Scion II dscribs h luminosiy masurmns from hi couning, and Scion III dscribs h luminosiy masurmns from havy blob (hrmal nuron couning. Th rlaion bwn his and clusrs usd o valua h saisical prcision is discussd in Scion IV. Dails of h analysis of MPX daa akn during LHC vdm scans for an absolu luminosiy calibraion ar givn in Scion V. Th shor-rm MPX prcision valuad from a daild sudy of LHC luminosiy curvs is prsnd in Scion VI. Conclusions ar givn in Scion VII. II. MPX LUMINOSITY FROM HIT COUNTING Th six MPX dvics wih highs clusr ra, spcifid in Tabl I ar usd (MPX-5,3. Each dvic masurs h TABLE I MPX DEVICE LOCATIONS WITH RESPECT TO THE INTERACTION POINT. Z IS THE LONGITUDINAL DISTANCE FROM THE INTERACTION POINT AND R IS THE DISTANCE FROM THE BEAM AXIS. ONLY DEVICES WITH LOW CLUSTER RATES ARE USED FOR THE HEAVY BLOB (THERMAL NEUTRON COUNTING ANALYSIS AS INDICATED. ORDERING IN THE TABLE IS GIVEN WITH DECREASING PARTICLE FLUX. REJECTED DURING ANALYSIS (SEC. III. Dvic Z R Masurd MPX clusrs pr Usd for (m (m uni snsor ara and pr uni h. nuron luminosiy (cm /nb analysis MPX 3.4.77 55 No MPX3 3.4.44 3 No MPX 3.4.5 3 No MPX4 7..3 No MPX5 7..36 47 No MPX3.94 3.57 3 No MPX6 7. 3.36 Ys MPX9 5.39.56 5. Ys MPX 7.3 6.5 3.9 Ys MPX 4. 4.4. Ys MPX. 5.9. Ys MPX7.35 4.59.45 Ys MPX 4.6 6.69.3 Ys s L MPX (b 55 LHC fill 649 5 Normalizaion vdm scan LHC fill 336 MPX 45 4 35 4 5 6 7 9 Hours on May Fig.. Tim hisory (CEST of h MPX luminosiy. Th small dips, visibl as variaions from h dscnding curv, corrspond o ims whn h LHC opraors prformd small-ampliud bam-sparaion scans o opimiz h luminosiy, Th normalizaion bwn MPX hi ra and luminosiy is basd on van dr Mr scans using LHC fill 336, daild in Sc. V. luminosiy indpndnly and is cross-chckd wih h ohr dvics. This is an inrinsic advanag of h MPX nwork. Each MPX dvic has is own acquisiion im window indpndn of h ohr dvics. In ordr o compar luminosiy masurmns from diffrn dvics, a common im window is inroducd, calld Luminosiy Block (LB which is ypically on minu long. A small numbr of noisy pixls could hav a significan ffc on h luminosiy masurmn. Thr indpndn mhods diffring in procdur and criria for noisy pixl rmoval hav bn sd. Th rsuls of hir rspciv luminosiy masurmns ar compard as ponial sourcs of sysmaic uncrainy. Mhod : A pixl is dfind as noisy if is coun ra is mor han fiv sandard dviaions from h avrag. If a pixl is found o b noisy in a 4 hour rfrnc priod (chosn as h day in ha had h largs numbr of noisy pixls i is rmovd from h nir daa-aking priod. A linar inrpolaion is mad bwn h ras (numbr of his pr scond in diffrn frams. Th hi ra a h middl of an LB is obaind from his inrpolaion. Mhod : Noisy pixl rmoval is don fram by fram, i.. a diffrn s of noisy pixls is rmovd in ach fram. Noisy pixls ar hos wih a couning ra ha diffrs from h man by mor han a luminosiydpndn hrshold. Th MPX luminosiy from frams falling wihin an LB is usd wihou an inrpolaion. A corrcion is mad for h rlaiv duraion of h MPX frams and of h LB. Mhod 3: Noisy pixl rmoval is don fram by fram. Th couns of 5 frams (h fram undr invsigaion and 7 frams bfor and afr ar summd and a pixl is rmovd if h sum of hs couns is abov a hrshold. An inrpolaion of h fram hi ra a h im of ach LB is don as in mhod.
3 All hr mhods show a significan incras in h numbr of noisy pixls wih im, whn applid o MPX daa. This migh indica possibl radiaion damag in h radou chip. In mhod (mhod, h numbr of noisy pixls in MPX incrass from lss han (3 in April o abou 3 ( a h nd of Novmbr. Th ohr dvics hav a smallr numbr of noisy pixls sinc hy ar xposd o a much lowr paricl flux (Tabl I. Th luminosiis masurd wih h hr mhods wr compard in shor (fram-by-fram and long (7 monhs im priods. Dpnding on h MPX dvic considrd, h framby-fram agrmn varis from a fw prcn o lss han.% (for MPX. Th largs variaion is bwn mhod and h ohr wo mhods. W hav drmind ha mos of his variaion is aribud o h convrsion bwn frams and LBs in mhod. In h following, mhod is usd and hus h sam noisy pixls ar rmovd for h whol daa-aking priod. During h analysis of h MPX daa, im-shifs bwn h hr radou PCs wr noicd. Thrfor, hy wr synchronizd off-lin by analyzing h rising and falling luminosiy curvs whn an LHC fill sars and nds. A im accuracy br han h LB lngh was hrfor achivd. In h procss of prforming h luminosiy drminaion wih MPX dvics, h acivaion of h ATLAS dcor marial was invsigad and found o hav a ngligibl ffc. Firs, h hi ras pr fram N his / acquisiion ar convrd ino hi ras pr LB for ach MPX dvic sparaly. Th procdur is dscribd blow. Frams wihin h im window of h LB ar slcd. Th hi ras of hs frams ar avragd. Thus, on hi ra is sord pr LB. Givn ha h acquisiion ims vary bwn 5 and s h numbr of frams usd pr LB varis for h six MPX dvics. If hr is no hi ra for a givn LB, h prvious LB hi ra is usd. In h MPX luminosiy analysis only hos LBs ar usd for which all six MPX dvics (MPX-5 and MPX3 wr opraional. Th hi ra for h MPX dvic is normalizd o unis of luminosiy by muliplying wih h facor n f =.56 3 b /hi, drivd in Sc. V. Thn, h ohr dvics ar normalizd o MPX basd on h avrag hi ra for h Jun o Novmbr running priod. Tabl II summarizs h normalizaion facors. Th avrag luminosiy (MPXav for all ohr dvics (xcluding h on undr considraion is calculad LB-by- LB. Using his normalizaion facor, ach MPX dvic obains an qual wigh, alhough h MPX dvics hav largly varying paricl fluxs (hi ras. Th raio MPX/MPXav is calculad LB-by-LB for h six MPX dvics. Figur shows h luminosiy raio pr LB for h daa-aking priod April-Novmbr for all dvics, xcp for MPX using h daa from Jun o Novmbr. A singl Gaussian fi is applid using h saisical uncrainy nris in ach bin. Tabl II summarizs h Gaussian fi valus. Th widh of hs fis vary bwn.6% and.% dpnding on h MPX dvic. In addiion, h long-rm im sabiliy of h six MPX dvics is sudid. For his sudy h LBs for which all h MPX dvics wr opraional ar groupd ino 4 im priods, such ha ach im priod conains h sam numbr of LBs. Th luminosiy raio of an individual MPX dvic o h avrag of all ohr MPX dvics (MPXav is calculad for ach im priod and givn in Fig. 3. A normalizaion is applid such ha h raio is uniy in im priod for ach MPX dvic. A linar fi is applid o h MPX/MPXav luminosiy raio vrsus im for h Jun o Novmbr daa-aking priod. Th slop of h linar fi is akn as a masur of im sabiliy. Th obaind slop valus and hir uncrainis ar summarizd in Tabl III. Th varianc of hs slop masurmns is.69 [% pr days]. Th rsuling sandard dviaion of.3 [% pr days] is usd as an simaion of h sysmaic uncrainy. In summary, for h six high saisics MPX dvics h widh of h flucuaions LB-by-LB is bwn (.6.%, and h im-sabiliy from Jun o Novmbr is br han %. This givs us an indicaion of h im sabiliy of MPX luminosiy monioring. III. MPX LUMINOSITY FROM HEAVY BLOB (THERMAL NEUTRON COUNTING Thrmal nurons ar dcd by MPX dvics via 6 Li(n,α 3 H racions in a 6 LiF convrr layr wih hicknss of -3 mg/cm on avrag [4, Sc..3]. In MPX racking mod, rions and alpha paricls ar rgisrd by Si-snsors as so-calld havy blobs, HB (larg round-shapd pixl clusrs. Th ypical dcion fficincy for hrmal nurons is %, drmind from individual calibraions of h MPX dvics in a hrmal nuron fild [4, Sc..3]. Hnc, h HB coun ra is usd as a masur of insananous luminosiy sinc nurons ar gnrad in h LHC collisions. Th MPX6 o MPX dvics ar usd for h HB (hrmal nuron couning sinc h pixl marix occupancy for hs dvics is sufficinly small for parn rcogniion. A ddicad sudy was prformd o drmin h misidnificaion of havy blobs which ar los du o h ovrlap wih ohr clusrs [4, Sc..]. Th rsuling corrcion facors, spcific o ach MPX dvic, dpnd on h numbr of clusrs pr fram (i.., on h LHC collision ra, on h dvic locaion and on h acquisiion im. Th prcision of hs corrcion facors was simad o b blow % wih h xcpion of MPX6 wih h largs pixl occupancy. Thrfor, MPX6 was no usd for h prcision sudy and combinaion wih h ohr dvics. An analyic modl for h clusr ovrlap probabiliy has bn dvlopd and agrs wih h xprimnal rsuls for simpl clusr shaps. Th disribuion of havy blobs pr fram rcordd wihin h MPX rgion covrd by a 6 LiF convrr is wll dscribd by a Poisson disribuion, dmonsrad in [7]. For luminosiy drminaion, ovrlap corrcd numbr of havy blobs (chb is usd for ach MPX dvic. Th numbr of chb pr fram is convrd ino LBs, similar o h hi analysis. Frams which li wihin h im window of h LB ar slcd. Th numbrs of chb of hs frams ar avragd
4 35 3 5 Daa MPX Widh =.7 % 35 3 5 Daa MPX Widh =.69 % Enris 5 Enris 5 5 5.95.96.97.9.99...3.4.5 L MPX /L MPX-5,3.95.96.97.9.99...3.4.5 L MPX /L MPX,3-5,3 Enris 6 4 6 4 Daa MPX3 Widh =.6 %.95.96.97.9.99...3.4.5 L MPX3 /L MPX-,4-5,3 Enris 3 5 5 5 Daa MPX4 Widh =.6 %.95.96.97.9.99...3.4.5 L MPX4 /L MPX-3,5,3 Enris 4 6 4 6 4 Daa MPX5 Widh =. %.95.96.97.9.99...3.4.5 L MPX5 /L MPX-4,3 Enris 4 35 3 5 5 5 Daa MPX3.95.96.97.9.99...3.4.5 L MPX3 /L MPX-5 Widh =.64 % Fig.. Luminosiy raio MPX/MPXav for h six MPX dvics wih highs paricl flux (MPX-5 and MPX3. Th disribuions ar approximad by Gaussian fis. Th widh of h Gaussian is an sima of h LB-by-LB uncrainis. LHC fills from Jun o Novmbr. TABLE II NORMALIZATION FACTOR/n f, WIDTH OF GAUSSIAN FITS OF LUMINOSITY RATIO MPX/MPXAV FOR THE SIX MPX DEVICES (MPX-5 AND MPX3, SHOWN IN FIG.. THEχ /ndf VALUES INDICATE THAT IN ADDITION TO THE STATISTICAL UNCERTAINTIES, SOME SYSTEMATIC UNCERTAINTIES ARE PRESENT. FOR A REALISTIC WIDTH UNCERTAINTY DETERMINATION OF AN INDIVIDUAL DEVICE, THE ERRORS ARE SCALED TO OBTAINχ /ndf =. MPX /n f (hi/b R widh (% σr widh (% χ /ndf σr widh 639..7.4 596/97.5 4.393.69. 3/97.34 3.579.599.3 39/97.76 4.36.63. 46/97.6 5.5534.39.35 66/97.9 3 5.9.6394. 36/97.3
5 L MPX /L MPX-5,3.5. Daa MPX.5.995.99 L MPX /L MPX,3-5,3.5. Daa MPX.5.995.99.95 9/6 9/7 / 7/9 7/ 6/ Das in.95 9/6 9/7 / 7/9 7/ 6/ Das in L MPX3 /L MPX-,4-5,3.5. Daa MPX3.5.995.99 L MPX4 /L MPX-3,5,3.5. Daa MPX4.5.995.99.95 9/6 9/7 / 7/9 7/ 6/ Das in.95 9/6 9/7 / 7/9 7/ 6/ Das in L MPX5 /L MPX-4,3.5. Daa MPX5.5.995.99 L MPX3 /L MPX-5.5. Daa MPX3.5.995.99.95 9/6 9/7 / 7/9 7/ 6/ Das in.95 9/6 9/7 / 7/9 7/ 6/ Das in Fig. 3. Tim hisory of h luminosiy raio MPX/MPXav for h six MPX dvics wih highs paricl flux (MPX-5 and MPX3 usd for hi couning. A linar fi is applid o drmin h slops. Th daa is dividd ino 4 im priods wih an qual numbr of LBs. Th daa is scald such ha h valu of h firs bin is uniy. Th siz of h saisical rror bar is blow h siz of h daa poin. I is idnical for vry daa poin of h sam dvic. Th siz of h rror bar which would giv χ /ndf = is givn in Tabl III for ach MPX dvic. LHC fills from Jun o Novmbr. TABLE III SLOPE OF TIME HISTORY OF THE LUMINOSITY RATIO MPX/MPXAV FOR THE SIX MPX DEVICES WITH HIGHEST PARTICLE FLUX (MPX-5 AND MPX3 USED FOR HIT COUNTING. THE SLOPE VALUES AND THE UNCERTAINTIES ARE GIVEN PER SECOND, AND IN PERCENT FOR DAYS. FOR A REALISTIC SLOPE UNCERTAINTY DETERMINATION OF AN INDIVIDUAL DEVICE, THE ERRORS ARE SCALED TO OBTAINχ /ndf =. σr IS THE SIZE OF THE ERROR BARS WHICH ARE THE SAME FOR EACH DATA POINT (FIG. 3. MPX Slop σslop Slop σslop σr ( s ( s (%/d (%/d 3.9.94.534.336.349...364.7.94 3 3.4.9.559.5.4 4 5.95.36.9.35.45 5 5.6.3.74.5.34 3.66..459.9.9
6 for all opraional MPX dvics. Only hos LBs for which all MPX dvics (MPX7 wr opraional ar usd. Ths LBs ar groupd ino 4 im priods, such ha ach im priod has an qual numbr of LBs, in h sam way as i was don for h hi analysis. For ach im priod h numbr of chb ar summd for ach MPX dvic. Th summd chb ar convrd ino luminosiy by using a normalizaion facor such ha h luminosiy raio MPX/MPXav of im priod is uniy. In ordr o calcula h luminosiy raio MPX/MPXav for ach im priod, h wighd luminosiy avrag of all h dvics (xcluding h dvic undr considraion usd in MPXav is drmind. For ach dvic and ach im priod h saisical uncrainy is / N HB, whr N HB is h summd numbr of havy blobs. A linar fi is applid o drmin h prcision of h im hisory of h raio bwn individual MPX dvics wih rspc o h wighd avrag of all ohr dvics, shown in Fig. 4. Tabl IV summarizs h slop valus and h uncrainis of h linar fis. Th varianc of hs slop masurmns is.3 [% pr days]. Th rsuling sandard dviaion of.6 [% pr days] is usd as an simaion of h sysmaic uncrainy. Th χ /ndf is clos o uniy, hus saisical uncrainis dscrib h flucuaions. In h analysis of HB couning in h 6 LiF-covrd dcor rgion h saisical uncrainis ar dominan sinc h HB coun ra is rahr small (a fw HB pr fram. In summary, h long-rm im-sabiliy from Jun o Novmbr 5 of h HB analysis is br han % comparing singl MPX dvics wih h wighd avrag of all ohr MPX dvics. IV. RELATION BETWEEN HITS AND CLUSTERS Th rlaion bwn h numbr of his and clusrs is invsigad in ordr o drmin h saisical uncrainy in luminosiy from hi couning. Th dfiniion of six clusr yps is basd on diffrn shaps obsrvd. Thy ar dos, small blobs, curly racks, havy blobs, havy racks and sraigh racks [4, Sc..]. During physics daa-aking MPX opras in couning mod. Howvr, during h so-calld van dr Mr scans h occupancy of h dvic is sufficinly low for racking paricls. As an xampl h las horizonal van dr Mr scan of Novmbr (LHC fill 336 is analysd o drmin h raio bwn his and clusrs. Th daa covrs a im priod of 6 sconds, in which 3 frams wr akn. Th oal numbr of clusrs is 55. Figur 5 shows h numbr of his pr clusr for h six MPX dvics wih h highs clusr ras wihou disinguishing clusr yps. In summary, h raio N hi /N cl = 435/55 =.65 is smalls for MPX and largs for MPX3 wih N hi /N cl = 57/65 = 4.9. Tabl V liss h numbr of clusrs, hi/clusr raios and RMS valus. Assuming ha on clusr is crad by on paricl, his raio corrsponds o h hi ra pr inracing paricl. Th flucuaions in h numbr of paricls, no h numbr of his, conribu o h saisical uncrainy of h luminosiy masurmn. V. VAN DER MEER SCANS Van dr Mr (vdm scans ar usd for absolu luminosiy calibraion a h LHC []. Th vdm scan chniqu was pionrd a CERN in h 96s o drmin h luminosiy calibraion in a simpl way. I involvs scanning h LHC bams hrough on anohr o drmin h siz of h bams a hir poin of collision. Th scans ar prformd o drmin h horizonal and vrical widhs of h bams. Ths widh masurmns ar hn combind wih informaion on h numbr of circulaing proons, allowing h drminaion of an absolu luminosiy scal. Svral ATLAS and CMS subdcors ar usd for vdm scans [5], [6] sinc h luminosiy calibraion is vry imporan for physics analyss. Th sudy of h MPX daa akn during LHC vdm scans in April, July and Novmbr focuss on h horizonal and vrical widh drminaion of h LHC proon bams. Th daa usd in his vdm scan sudy was akn wih h MPX dvic wih h highs coun ra. This sudy dmonsras ha h opraion of h MPX nwork is fully adapd o h low-luminosiy rgim of vdm scans and h high-luminosiy rgim of rouin physics running. Owing o h low saisics in havy blob (hrmal nuron couning, only h hi couning mod can b usd for h vdm daa analysis. Th bams ar ypically scannd ransvrsly across ach ohr in 5 sps. During ach sp, h bam orbis ar lf unouchd ( quiscn bams and h luminosiy rmains consan for approximaly 9 s. Th bam sparaion is hn incrmnd by svral ns of microns ( non-quiscn bams ovr an inrval of svral sconds, during which h luminosiy varis rapidly and h luminosiy masurmns ar unrliabl. Sinc h MPX xposur (acquisiion im is abou 5 s pr fram, followd by a 6 s radou priod, wo frams ypically occur wihin ach quiscn-bam scan sp. Occasionally, h MPX dvics nd o rload hir configuraion fils, in which cas h dad im can b as long as 3 s. Thrfor, only on fram is rcordd in som scan sps. Th bam sparaion dpndnc of h masurd MPX luminosiy is wll rprsnd by h sum of a singl Gaussian and a consan (Fig. 6. Th saisical uncrainy for ach MPX fram, calculad from h numbr of his, is scald up by a facor.65 o accoun for h raio bwn his and clusrs, as xplaind bfor in Sc. IV. In his approach i is assumd ha on paricl inracing wih h MPX dvic cras on clusr and i is h numbr of paricls which lad o h saisical uncrainy. Th prcision of h MPX dvic can b drmind wih rspc o h xpcd saisical prcision. For his sudy, h pull (daa-fi/σ daa is calculad (Fig. 7 for h las horizonal vdm scan in Novmbr, whr σ daa = R σsa hi and R =.65. Th sigma of h pull disribuion is.7, which indicas ha addiional uncrainis ar prsn byond h drmind saisical uncrainis. Th daa shows ha h snsiiviy of h MPX dvic is sufficin o masur h luminosiy down o abou.5 b s, four ordrs of magniud smallr han h luminosiy ypical of rouin physics daa-aking.
7.4 Daa.3 MPX7.4 Daa.3 MPX L MPX7 /L MPX...99 L MPX /L MPX7,9...99.9.9.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in L MPX9 /L MPX7-,.4 Daa.3 MPX9...99.9 L MPX /L MPX7-9,.4 Daa.3 MPX...99.9.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in.4 Daa.3 MPX.4 Daa.3 MPX L MPX /L MPX7,...99 L MPX /L MPX7...99.9.9.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in.97 /5 9/6 9/7 / 7/9 7/ 6/ Das in Fig. 4. Tim hisory of h luminosiy raio MPX/MPXav for h six MPX dvics wih lows paricl flux (MPX7 usd for havy blob (hrmal nuron couning. A linar fi is applid o drmin h slops. Th daa is dividd ino 4 im priods wih an qual numbr of LBs. Th daa is scald such ha h valu of h firs bin is uniy. Th saisical rror bars ar shown. LHC fills from May o Novmbr. TABLE IV SLOPE OF TIME HISTORY OF THE LUMINOSITY RATIO MPX/MPXAV FOR THE SIX MPX DEVICES WITH LOWEST PARTICLE FLUX (MPX7 USED FOR HEAVY BLOB (THERMAL NEUTRON COUNTING. THEσSLOPE VALUE RESULTS FROM THE STATISTICAL UNCERTAINTY/ N HB PER TIME PERIOD. THE χ /ndf VALUES OF THE LINEAR FITS FROM FIG. 4 ARE ALSO SHOWN.. MPX Slop σslop Slop σslop χ /ndf ( s ( s (%/d (%/d 7.45 6.7 -.5.7 9.4 / 4.93 4..5.76 7. / 9-3. 3.77 -.536.64.3 / 4.57 4.4.79.757 7.57 /.7 9.5.34.573 9.5 / -.4 3. -.49.553.7 /
Numbr of clusrs 5 4 3 Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX Numbr of clusrs Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX 5 5 5 3 His pr clusr 5 5 5 3 His pr clusr Numbr of clusrs Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX3 Numbr of clusrs Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX4 5 5 5 3 His pr clusr 5 5 5 3 His pr clusr Numbr of clusrs Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX5 Numbr of clusrs Las hor. scan Nov. LHC fill 336 Clusr: all yps MPX3 5 5 5 3 His pr clusr 5 5 5 3 His pr clusr Fig. 5. Numbr of his pr clusr combind for six clusr yps. Daa is akn by MPX-5 and MPX3 during h las horizonal van dr Mr scan of Novmbr (LHC fill 336. TABLE V NUMBER OF CLUSTERS, AVERAGE RATIO HITS PER CLUSTER AND RMS VALUES OF THE HITS PER CLUSTER DISTRIBUTIONS, SHOWN IN FIG.5. DATA IS TAKEN BY MPX-5 AND MPX3 DURING THE LAST HORIZONTAL VAN DER MEER SCAN OF NOVEMBER (LHC FILL 336. FOR COMPARISON THE CORRESPONDING VALUES ARE GIVEN FOR ONE OF THE VERTICAL VAN DER MEER SCANS OF JULY (LHC FILL 55. IN THIS STUDY, NO NOISY PIXELS ARE REMOVED FOR THE HIT COUNTING OR CLUSTER COUNTING, THE SMALLER RATIO IN NOVEMBER FOR MPX CAN BE EXPLAINED BY THE INCREASE IN THE NUMBER OF NOISY PIXELS DURING THE YEAR. MPX 3 4 5 3 N cl 55 7 7 94 5 65 NOV. N hi /N cl.65 4.3 4.4 4. 4. 4.9 RMS 6.7 5.5 4.4 4. 4.64 7.3 N cl 66 6 9 3 47 JULY N hi /N cl 5.4 4.95 4.65 4. 3.5 4.67 RMS.35 6.54 5.9 4.7 3.7 6.
9 MPX hi ra (s 4 6 4 6 4 LHC fill 336 MPX -.6 -.4 -...4.6 Nominal sparaion (mm Fig. 6. Luminosiy from hi couning as a funcion of nominal bam sparaion masurd wih MPX during h las horizonal vdm scan of Novmbr. Each daa poin shows h masurd insananous luminosiy bfor background subracion and avragd ovr on acquisiion im. Bcaus h acquisiion im is significanly shorr han h duraion of a scan sp, hr can b mor han on MPX sampling pr scan sp. Th MPX samplings ha parially or oally ovrlap wih non-quiscn scan sps (varying bam sparaion ar no shown. Th fi funcion is h sum of a singl Gaussian (rprsning h propr luminosiy in his scan and a consan rm ha accouns for insrumnal nois and singl-bam background. Th MPX normalizaion uss his horizonal and a vrical bam widh from LHC vdm fill 336. Enris 6 4 Las hor. scan Nov. LHC fill 336 MPX - -6-4 - 4 6 MPX luminosiy pull (daa-fi/σ daa.5.5.5 hi σ daa =.65 σ sa Pull.7 Fig. 7. Pull disribuion dfind as (daa-fi/σ daa, whr σ daa = R σ hi sa and h raio R = N hi/n cl =.65 for MPX. Th daa shown in Fig. 6 is usd. LHC fill 336. Th luminosiy can b calculad as: MPX luminosiy (b L MPX = N b N p N p f/(πσ x Σ y, ( whr N b is h numbr of bunch crossings producing collisions pr machin rvoluion, N p and N p ar h avrag bunch populaions (numbr of proons in bam and bam, rspcivly, f is h machin rvoluion frquncy (45.5 Hz, and Σ x (Σ y ar h convolud horizonal (vrical bam sizs. Th spcific luminosiy is dfind as: L spcific = L MPX /(N b N p N p = f/(πσ x Σ y. ( Tabl VI summarizs h scan rsuls for all vdm scans rgisrd wih h MPX dvic. s Las hor. scan Nov. TABLE VI MPX VAN DER MEER (VDM SCAN RESULTS FOR DATA. THE FIT RESULTS FOR THE BUNCH-AVERAGED HORIZONTALΣ x AND VERTICAL Σ y CONVOLUTED BEAM SIZES ARE GIVEN. THE UNITS OF THE SPECIFIC LUMINOSITY,L spcific, ARE 7 cm s ( proons. APRIL 5 JULY 55 NOV. Fill Scan Σ x Σ y L spcific numbr (m (m 33 5.94 33.7 6 6.4 3.65 5 3 6.7 33.64 99 4.5 7.6 6 5 4.7 7.9 6 3...6 7. 6.4. 4 9.3 9.6 6 4 39.9 3. 97 336 5 9.3 33.6 In his papr, h las horizonal and vrical Novmbr scans ar usd for h absolu luminosiy calibraion. Ths wo scans ar wll dscribd by a singl Gaussian. Th horizonal scan (Fig. 6 has χ /ndf = 36/47 and a similar valu is obaind for h vrical scan χ /ndf = 4/44. Ths raios indica ha in addiion o h saisical uncrainy (augmnd by h facor.65 sysmaic uncrainis ar also prsn. Th widhs of h horizonal and vrical nominal bam sparaions and hir uncrainis ar (9.3 ±.6 m and (33.6 ±.9 m, rspcivly. Th LHC paramrs for fill 336 ar [9]: Numbr of bunchs: 9 Avrag numbr of proons (in unis pr bunch in bam and in bam : 5.3/9 =.7 and 5.7/9=.6, rspcivly. Thus, h rsuling luminosiy is L MPX =.55 b s. Th corrsponding numbr of MPX his a h pak is drmind from a Gaussian fi plus a consan background. Th fi provids (69.34±.3 his/s a h pak abov h background. Thus, h normalizaion facor n f bwn h MPX hi ra and h insananous LHC luminosiy is n f =.55 b s 69.34his =.56 3 b /hi. (3 Th normalizaion facor for h absolu luminosiy is only approxima sinc h MPX acquisiion im is much longr han h bunch spacing. Thrfor, h bunch-ingrad luminosiy avrags ovr h diffrn bunch profils. Th uncrainy on n f du o h bunch-ingraion by h MPX daa-aking has bn simad by simulaing 9 ovrlapping Gaussian disribuions, corrsponding o 9 colliding bunchs, wih varying individual widhs. Th simulad bunchs vary in widh in qual disancs up o ±5%. Th nvlop of h summd Gaussians is fid and h rsuling widh and high ar compard wih h nominal valu wihou
variaions. Th widh uncrainy is.36% and high uncrainy is.49%. Th fi is rpad ims wih diffrn ss of random numbrs o s h rproducibiliy. For h drminaion of luminosiy uncrainy i is assumd ha h horizonal and vrical widh uncrainis ar corrlad, hus h luminosiy uncrainy is.7%. Furhrmor, i is assumd ha h widhs and high (fid hi ra uncrainis ar corrlad, hus h uncrainy on h normalizaion facor is.7%+.5% =.%. Alhough furhr uncrainis could aris from non- Gaussian shaps, his sudy shows ha h Gaussian approximaion of h sum of Gaussians is qui robus and h luminosiy approximaion by bunch ingraion is a snsibl approach. No amp is mad for a prcis drminaion of h oal uncrainy which would rquir a ddicad sudy [5]. In summary, Fig. 6 shows h hi ra and h absolu luminosiy drmind from h scan widhs. Th rsuling normalizaion facor is usd hroughou his papr. VI. LHC LUMINOSITY CURVE AND MPX SHORT-TERM PRECISION Th MPX nwork prcisly masurs h LHC luminosiy as a funcion of im. As a proof of principl i is dmonsrad ha h MPX nwork has h capabiliy o sudy h undrlying mchanisms of h ra of rducion of LHC luminosiy. Th LHC luminosiy rducion is mainly causd by bambam inracions (burning-off h proon bunchs and bamgas (singl bunch inracions by h proons of h circulaing bams wih rmaining gas in h vacuum pip. Th paricl loss ra du o proon burn off in collision is proporional o h numbr of proons in h scond powr sinc proons ar los in boh colliding bunchs. In h cas ha h proons in a bam collid wih rmaining gas in h vacuum pip, h paricl loss ra is proporional o h numbr of proons in h bam. Th loss ra of proons N in h colliding bam is hus govrnd by: dn/d = N /N + N, (4 whr N is h iniial numbr of proons, and and ar consans rlad o bam-bam and bam-gas inracions, rspcivly. This quaion has a known soluion: N λg N( = + ( λg, (5 wih wo wll-known bordr cass: N( = N λg for and, (6 N( = N + for. (7 In h following w will b inrsd in h im dpndnc of h luminosiy and of h avrag numbr of inracions pr bunch crossing. By dfiniion is proporional o h luminosiy L. Sinc boh of hs quaniis ar proporional o N w xpc h im dpndnc of o b dscribd by: λg ( = [+ ( λg ]. ( A fi is applid o h daa using q. (. Th uncrainy on h fi rsul is valuad by svral indpndn masurmns of h MPX nwork. Th LHC fill 336, akn on -9 Ocobr, has bn invsigad in his sudy. Th MPX luminosiy is convrd o an avrag inracion pr bunch crossing by: = L σ inl /(k f, (9 whr k = 36 colliding bunchs, f = 45.5 Hz and h inlasic cross-scionσ inl = 73 mb. Th fill was chosn sinc i has a larg rang from abou = 35 a h bginning o abou = a h nd. Whn sudying h MPX masurmns of h LHC luminosiy, srucurs rsuling from LHC paramr uning can b nod. Ths bam uning adjusmns chang h shap of h luminosiy dcras, and ar no dscribd by q. (. I is nod ha in h firs half of a fill hs unings wr frqun (abou on vry hour whil in h scond half of a fill, adjusmns of h bam wr rarly mad. Thrfor, only h scond half of a fill is usd for his sudy and h rang = 5 o = is usd for fiing h daa disribuion. Th fis ar shown in Fig. individually for MPX-5 and MPX3. Th fis giv h iniial numbr of inracions pr bunch crossing, and, summarizd in Tabl VII. Th fis wih boh paramrs and dscribs h daa significanly br compard o using only on of h wo. Figur 9 shows h diffrnc bwn h fi and h daa for MPX. Th obsrvd srucurs could b aribud o small LHC luminosiy flucuaions no dscribd by h fi funcion. Howvr, h origin of hs flucuaions canno b drmind from h MPX daa alon, and hus hy ar consrvaivly aribud o MPX sysmaics. Th siz of hs flucuaions simad as dparur of h daa from h fid curv, amouns o an RMS of approximaly.3%, corrsponding o fc sys =.3 for =. For all high-saisics dvics usd (MPX-5 and MPX3, h uncrainis of h fis ar dominad by sysmaic ffcs. Th fi of h MPX daa, for xampl, has a χ /ndf = 4.4 7 /499 much largr han on. Thrfor, h fis ar rpad wih a consan sysmaic uncrainy for ach MPX daa poin such ha χ /ndf =. Ths fi paramrs ar also summarizd in Tabl VII. Th λ man valus of h bam-bam and bam-gas inracions ar and = (6.±.3 6 s ( = (3.±. 6 s, ( whr h givn sandard dviaions ar calculad as h squar roo of h varianc from h numbrs in Tabl VII for consan sysmaic uncrainis. Th fi rsuls indica ha h LHC luminosiy rducion is prdominanly rducd by h bam-bam inracions sinc a largr valu of λ corrsponds o a shorr lifim. In addiion o h inracions bwn h proon bams and h rmaining gas in h vacuum pip, hr could b ohr procsss which, assuming ha hy also dpnd linarly on h numbr of proons, ar incorporad in h valu of.
MPX 7 6 5 4 3 9 LHC fill 336 MPX (= fi [+ - 4..4457 λ.46 bb - (- ] - MPX 7 6 5 4 3 9 LHC fill 336 MPX (= fi [+ - 4..6 λ.55 bb - (- ] - 7 3 4 5 6 7 9 7 3 4 5 6 7 9 Hours on 9 Oc. Hours on 9 Oc. MPX3 7 6 5 4 3 9 LHC fill 336 MPX3 (= fi [+ - 4.4. λ.6676 bb - (- ] - MPX4 7 6 5 4 3 9 LHC fill 336 MPX4 (= fi [+ - 4.79.393 λ.46 bb - (- ] - 7 3 4 5 6 7 9 7 3 4 5 6 7 9 Hours on 9 Oc. Hours on 9 Oc. MPX5 7 6 5 4 3 9 LHC fill 336 MPX5 (= fi [+ - 4..39 λ.534 bb - (- ] - MPX3 7 6 5 4 3 9 LHC fill 336 MPX3 (= fi [+ - 4.3.44 λ.449 bb - (- ] - 7 3 4 5 6 7 9 7 3 4 5 6 7 9 Hours on 9 Oc. Hours on 9 Oc. Fig.. Avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX dvics wih highs paricl flux (MPX-5 and MPX3. Th disribuion is approximaly dscribd by a funcion as givn in h figur. Th paramrs ar dfind in h x. Th saisical uncrainis pr daa poin ar indicad. Thy dpnd on h hi saisics scald by a facor N hi /N cl givn in Tabl V. In addiion o h fi rsul dscribd by h solid lin, h dashd lin shows h rsul of bam-gas inracions alon xp(, wih and obaind from h hr paramr fi. LHC fill 336. TABLE VII FIT VALUES OF, AND FOR THE MPX DEVICES WITH HIGHEST PARTICLE FLUX FOR STATISTICAL UNCERTAINTIES ONLY, AND FOR SYSTEMATIC UNCERTAINTIES ONLY, CONSTANT IN TIME, SUCH THATχ /ndf =. LHC FILL 336. MPX 3 4 5 3 Man Sandard dviaion Saisical uncrainis only 4. 4. 4.4 4.79 4. 4.3 4.. ( 6 s 4.7 6.7 7.7 5.33 6.3 4.6 5.9. ( 6 s 5.6 3.3.34 4.5 3.7 5.9 4.. Sysmaic uncrainis only 4. 4. 4.5 4.79 4. 4.3 4.. ( 6 s 4.9 7.. 5.4 6.37 4.94 6.5.3 ( 6 s 4.9 3.6.9 4.44 3.59 4.7 3..
(daa-fi/fi (% MPX. LHC fill 336.6.4. -. -.4 -.6 -. MPX 3 4 5 6 7 9 Hours on 9 Oc. Fig. 9. Rlaiv diffrnc bwn daa and fid avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX. Th rlaiv dviaions bwn daa and fi hav an RMS of.3%. Th saisical uncrainis / pr daa poin ar indicad and vary from.94.65% o.5.65% whr h facor.65 is h avragd raio of his pr inracing paricl. Th apparn srucur is discussd in h x. LHC fill 336. In h following, h xpcd man lifim of inlasic bam-bam inracions is calculad and compard wih h xprimnal rsuls. Th man lifim from inlasic bambam inracions is givn by []: inl bb = kn /(N xp L σ inl, ( whr k is h numbr of bunchs, N is h iniial numbr of proons pr bunch (kn =. 4 proons [9]. Th iniial luminosiy is L = 736 b s [9], h numbr of xprimns is N xp = (ATLAS [] and CMS []. W obain inl bb =.5 5 s and hus λ inl bb = / inl bb = 4.7 6 s. (3 W no ha λ inl bb dpnds on h iniial luminosiy and h iniial numbr of proons, hus on h saring valu of = 5 for h fi. Sinc L N w can wri λ inl bb L. Thus for h lowr iniial luminosiy in h fi, w xpc a longr lifim from bam-bam inracions and hrfor a smallr λ 5 bb = 5/35 4.7 6 s = 3. 6 s. (4 W obsrv ha h fid valu of q. ( is largr han λ 5 bb calculad using h inlasic cross-scion only, givn in q. (4. This suggss, as xpcd, ha h proonproon inracion cross-scion lading o h proon loss from h bam is indd largr han h inlasic cross-scion. In addiion o inlasic hadronic scaring, hadronic diffraciv, hadronic lasic and Coulomb scaring conribu o h proon burn-off. Anohr fi is prformd using h combind daa from MPX-5 and MPX3 as a consisncy chck of h prviously dscribd analysis procdur which usd h avrag of h fi paramrs from h individual MPX dvics. Th fi using h combind daa is shown in Fig.. I givs, as xpcd, almos idnical fi paramrs o h ons obaind as h avrag valus of h prvious procdur. Th raio of h daa in h firs half and h xrapolad fi curv is sudid sinc h prformd fis wr basd only on daa from h scond half of h LHC fill 336. I was found ha daa a h bginning of h fill is up o 3% abov h fi curv. This could indica ha h luminosiy dcrass mor quickly han xpcd from bam-bam (burn-off and bam-gas inracions alon a h bginning of a fill, possibly du o non-linar ffcs wih small mianc and shor-lngh bunchs. Furhrmor, fasr rducions in collision ras a h bginning of h fill could aris from dnsr bunchs. An inrinsic uncrainy in h MPX luminosiy masurmns rsuls from h bunch ingraion of h MPX nwork sinc h colliding bunchs in h LHC conribu wih diffrn innsiis. MPX-5,3 7 6 5 4 3 9 7 LHC fill 336 MPX-5,3 (= fi [+ - 3 4 5 6 7 9 Hours on 9 Oc. 4..34 λ.53 bb - (- ] - Fig.. Avrag numbr of inracions pr bunch crossing as a funcion of im using combind daa from MPX-5 and MPX3. Th disribuion is approximaly dscribd by h funcion givn in h figur. Th paramrs ar dfind in h x. In addiion o h fi rsul dscribd by h hin lin, h dashd lin shows h rsul of bam-gas inracions alon xp(, wih and obaind from h hr paramr fi. LHC fill 336..5.56.54.5.5.4 (= fi [+ - LHC fill 336 MPX-5,3 - (- ].9.3.3.3.33.34.35.36.37 Fig.. Conour curv of and, givn a, for h fi rsuls of h avrag numbr of inracions pr bunch crossing as a funcion of im using combind daa from MPX-5 and MPX3, shown in Fig.. Th do in h cnr of h conour indicas h fi valus. LHC fill 336. Th conour plo of and, givn a in Fig., shows a srong ani-corrlaion. Th individual uncrainis ar akn as consan (giving qual wigh o h individual MPX dvics in h combind daa sinc h uncrainy
3 of h MPX daa is sysmaically dominad. Th consan uncrainy is scald o =.44 yilding χ /ndf =. This sudy of h LHC luminosiy rducion allows us o invsiga h conribuion of saisical and sysmaic uncrainis for ach MPX dvic wihou rlying on comparisons wih ohr luminomrs, only using MPX daa du o is larg daa saisics. Th pull, (daa-fi/σ daa, is sudid for ach high-saisics MPX dvic individually wih σ daa = R σsa hi. Th R = N hi/n cl valu is givn in Tabl V for ach MPX dvic. Figur shows h pull disribuion for MPX-5 and MPX3. As xpcd from h obsrvd srucurs in Fig. 9, h pull valu 5.47 (dfind as h widh of a Gaussian fi is larg for MPX. Th pull valus vary bwn and 3 for MPX-5 and MPX3, givn in Tabl VIII. Th uncrainy nds o b saisical in naur sinc h pull disribuions ar wll dscribd by Gaussians. Whil h hi saisics of MPX-5 and MPX3 vary mor han a facor (Tabl V h pull is almos dvic indpndn. Thus, h sysmaic uncrainy in addiion o h hi saisics sms o also b of saisical naur. Thrfor, h oal uncrainy (saisical and sysmaic can b dscribd by scaling h saisical uncrainis such ha h pull is uniy. Comparing h pull valu of MPX wih hos of h ohr MPX dvics, h saisical prcision of MPX is high nough ha in addiion o h inrinsic MPX uncrainis (scaling wih h MPX hi saisics, variaions in h LHC luminosiy which ar no dscribd by h fi funcion bcom visibl, shown in Fig. 9. This is corroborad by h obsrvaion ha h pull disribuions as a funcion of im show a srucur only for MPX. Th flucuaions of luminosiy no dscribd by h fi funcion (Fig. 9 ar inrprd as sysmaic uncrainy. Thrfor, an addiional uncrainy corrsponding o h RMS of h daa dviaions from h fi funcion, fc sys =.3, is addd in quadraur o h saisical uncrainy. Figur 3 shows h rsuling pull disribuions for MPX-5 and MPX3, and Tabl VIII summarizs h fi rsuls. Th widh of h pull disribuion is clos o uniy as MPX was usd o driv h sysmaic uncrainy. For h ohr MPX dvics h pull varis bwn.44 and.. Thus, as for h vdm scan, h flucuaions ar abou a facor wo largr han xpcd from hi saisics alon for physics daa-aking which is abou 3 ims largr in luminosiy. Th ffc of h varying hi/clusr raio on h saisical valuaion is sudid wih a simpl Mon Carlo simulaion. Th goal is o drmin whhr using h avrag hi/clusr raio (.65 for MPX, rahr han h varying raios framby-fram, could incras h pull valu. This sudy canno b don wih rcordd daa sinc h pixl occupancy is oo larg during physics daa-aking whn h pull valu is drmind. Firs, a pull disribuion has bn simulad using a Gaussian disribuion wih uniy widh. Thn, h saisical uncrainis ar rducd by a consan valu.65. Alrnaivly, hy ar rducd by a varying facor N hi /N cl fram-by-fram rflcing h MPX disribuion in Fig. 5. Boh rsuling pull disribuions ar fid wih a Gaussian. Th formr givs a widh of.65 =.6, as xpcd, and h lar.. Thrfor, hr is no incras of h widh whn using h varying hi/clusr raios. Compard o h iniial Gaussian disribuion, h pull disribuion wih varying hi/clusr raios shows a highr pak and highr populad sid-bands. Thus, i can b xcludd ha h varying hi/clusr raio is h sourc of h obsrvd pull valus for MPX-5 and MPX3 ranging bwn.44 and. (Tabl VIII. Th following obsrvaions could xplain h abou wic as larg luminosiy flucuaions compard o h saisical xpcaions from h hi saisics alon: Th avrag hi/clusr raio, in paricular for MPX, could b largr as noisy pixls ar no xcludd whn h raio is drmind. Th assumpion ha on clusr corrsponds o on paricl bars an uncrainy involving h clusr dfiniion. Som paricls ar rflcd and could pass h MPX snsor mor han onc. Th showring of paricls in h ATLAS dcor marial incrass h numbr of paricls passing h MPX dvics, hus h numbr of indpndn paricls o b usd in h saisical drminaion of h xpcd uncrainy would b smallr. Th drminaion of his ffc would rquir a full simulaion of h marial disribuion of h ATLAS dcor in fron of ach of h MPX dvics which is byond h scop of his sudy. Howvr, h sam pull valu of abou wo, obsrvd for all six MPX dvics usd in h hi sudy, indicas ha h ffc is indpndn of h MPX dvic and is posiion. Ths ffcs could lad o a smallr numbr of indpndn objcs impacing h luminosiy masurmn. Thus, h saisical uncrainy is prsumably largr han from hi couning alon. A rducion of h numbr of indpndn objcs by abou a facor four incrass h saisical uncrainy by a facor wo, and would rduc h pull disribuion widh o uniy. In summary, Fig. 4 shows h conour plo of and, givn a, for h hi saisical uncrainis and sysmaic uncrainis from luminosiy flucuaions no dscribd by h fi funcion. Thy ar addd in quadraur, givn in Tabl IX. Figur 5 shows h corrsponding plos for χ /ndf =. Th variaions of h fi valus ar wll dscribd by h conours. Owing o h srong ani-corrlaion bwn and a ransformaion of h fi rsuls is prformd. Th ( vrsus ( + plan is usd in ordr o illusra h rlaiv uncrainy bwn h fi valus. Th comparaiv snsiiviis of h MPX dvics a ar summarisd in Fig. 6. Th analysis prformd wih LHC fill 336 daa, has bn rpad wih LHC fill 349 daa, akn 3 Ocobr Novmbr, for = 3.4 o = 9.. Th MPX rsuls from LHC fill 349 lad o a rlaiv prcision blow RMS.3% (Fig. 7 in agrmn wih h rsuls from LHC fill 336. Tabl X givs h rlaiv saisical uncrainis, and liss h oal uncrainis for RMS.% (obaind in LHC fill 349 and RMS.3% (obaind for LHC fill 336. Th corrsponding pull valus for saisical and oal uncrainis ar givn in Tabl XI.
4 Enris 6 4 6 4 LHC fill 336 MPX -5-4 -3-3 4 5 MPX pull (daa-fi/σ daa hi σ daa =.65 σ sa Widh = 5.47 Enris σ = 4.3 σ hi daa sa 6 4 6 4 LHC fill 336 MPX - -6-4 - 4 6 pull (daa-fi/σ MPX daa Widh = 3.5 Enris σ = 4.4 σ hi daa sa 6 4 6 4 LHC fill 336 MPX3 - -6-4 - 4 6 pull (daa-fi/σ MPX3 daa Widh =.4 Enris 5 hi σ daa = 4. σ sa 5 5 LHC fill 336 MPX4 - -6-4 - 4 6 pull (daa-fi/σ MPX4 daa Widh =. 5 LHC fill 336 MPX5 hi σ daa = 4. σ sa Widh =.96 6 4 LHC fill 336 MPX3 hi σ daa = 4.9 σ sa Widh =.64 Enris 5 Enris 6 5 4 - -6-4 - 4 6 pull (daa-fi/σ MPX5 daa - -6-4 - 4 6 pull (daa-fi/σ MPX3 daa Fig.. Pull disribuions dfind as (daa-fi/σ daa, whr σ daa = R σsa hi. Th raio R = N hi/n cl is drmind for ach MPX dvic sparaly. LHC fill 336. TABLE VIII WIDTHS OF THE GAUSSIAN FITS TO THE PULL DISTRIBUTIONS FOR MPX-5 AND MPX3. THE WIDTHS ARE GIVEN FOR UNCERTAINTIES RESULTING FROM THE NUMBER OF HITS SCALED BY A FACTOR Nhi /N cl FOR EACH MPX DEVICE, GIVEN IN TABLE V. THE WIDTHS ARE ALSO GIVEN FOR STATISTICAL AND SYSTEMATIC UNCERTAINTIES ADDED IN QUADRATURE. THE SYSTEMATIC UNCERTAINTIES RESULT FROM LUMINOSITY FLUCTUATIONS NOT DESCRIBED BY THE FIT FUNCTION. TABLE IX INDICATES THE SIZES OF THE UNCERTAINTY APPLIED. LHC FILL 336. MPX 3 4 5 3 Widh pull sa. uncr. 5.47 3.5.4..96.64 Widh pull oal uncr. (.3 sys...6..64..44
5 4 35 3 LHC fill 336 MPX σ daa =.65σ sa Widh =. hi fc σ sys 3 5 LHC fill 336 MPX σ daa = 4.3σ sa Widh =.6 hi fc σ sys Enris 5 5 5 Enris 5 5 - -6-4 - 4 6 MPX pull (daa-fi/σ daa - -6-4 - 4 6 pull (daa-fi/σ MPX daa Enris hi fc σ daa = σ sys 6 4 6 4 LHC fill 336 MPX3 - -6-4 - 4 6 pull (daa-fi/σ MPX3 daa 4.4σ sa Widh =. Enris 5 5 5 LHC fill 336 MPX4 - -6-4 - 4 6 pull (daa-fi/σ MPX4 daa σ daa = 4.σ sa Widh =.64 hi fc σ sys Enris 3 5 5 LHC fill 336 MPX5 σ daa = 4.σ sa Widh =. hi fc σ sys Enris 3 5 5 LHC fill 336 MPX3 σ daa = 4.9σ sa Widh =.44 hi fc σ sys 5 5 - -6-4 - 4 6 pull (daa-fi/σ MPX5 daa - -6-4 - 4 6 pull (daa-fi/σ MPX3 daa Fig. 3. Pull disribuions dfind as (daa-fi/σ daa, whr σ daa = R σsa hi σfc sys. Th raio R = N hi/n cl is drmind for ach MPX dvic sparaly and aks ino accoun h sysmaic uncrainy from h flucuaions no dscribd by h fi funcion. LHC fill 336. TABLE IX STATISTICAL UNCERTAINTY RANGES FOR THE DATA POINTS IN FIG.. THE UNCERTAINTY RESULTING FROM THE NUMBER OF HITS IS SCALED BY A FACTOR Nhi /N cl, GIVEN IN TABLE V. THE VALUES ARE GIVEN FOR THE BEGINNING AND END OF THE FIT RANGE. ALSO SHOWN ARE THE TOTAL UNCERTAINTIES CALCULATED ADDING IN QUADRATURE THE STATISTICAL AND SYSTEMATIC UNCERTAINTIES, WHERE THE SYSTEMATIC UNCERTAINTIES RESULT FROM LUMINOSITY FLUCTUATIONS NOT DESCRIBED BY THE FIT FUNCTION (σsys fc =.3, SHOWN IN FIG. 9. LHC FILL 336. MPX 3 4 5 3 sa..3-.7.49-.9.7-.54.435-.333.679-.54.9-.75 oal (.3 sys..3-.3.39-.355.763-.64.59-.44.74-.595.37-.347
6.75.7 LHC fill 336 MPX.75.7 LHC fill 336 MPX.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.75.7 LHC fill 336 MPX3.75.7 LHC fill 336 MPX4.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.75.7 LHC fill 336 MPX5.75.7 LHC fill 336 MPX3.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5 Fig. 4. Conour curvs of and, givn a, for h fi rsuls of h avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX-5 and MPX3. Th do in h cnr of h conour indicas h fi valus. Th hi saisical uncrainis and sysmaic uncrainis from luminosiy flucuaions no dscribd by h fi funcion ar addd in quadraur, givn in Tabl IX. Th χ valus ar 5,, 666, 45, 7, and 7, for MPX-5 and MPX3, rspcivly, for 499 dgrs of frdom. LHC fill 336. TABLE X SAME AS TABLE IX, BUT FOR LHC FILL 349. THE SYSTEMATIC UNCERTAINTIES RESULT FROM LUMINOSITY FLUCTUATIONS NOT DESCRIBED BY THE FIT FUNCTION (TAKINGσsys fc =. AND.3, SHOWN IN FIG. 7. LHC FILL 349. MPX 3 4 5 3 sa..-.9.37-.4.66-.57.49-.354.653-.556.-.7 oal (. sys..-..3-.6.67-.64.464-.47.6-.59.97-.74 oal (.3 sys..3-.3.33-.363.73-.644.55-.464.65-.63.37-.353
7.75.7 LHC fill 336 MPX.75.7 LHC fill 336 MPX.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.75.7 LHC fill 336 MPX3.75.7 LHC fill 336 MPX4.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.75.7 LHC fill 336 MPX5.75.7 LHC fill 336 MPX3.65.65.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5.6.55.5 (= -.45 fi λ -.4 [+ bb (- ].35..5..5.3.35.4.45.5 Fig. 5. Conour curvs of and, givn a, for h fi rsuls of h avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX-5 and MPX3. Th do in h cnr of h conour indicas h fi valus. Th hi saisical uncrainis and sysmaic uncrainis from luminosiy flucuaions no dscribd by h fi funcion ar addd in quadraur, givn in Tabl IX, and hn scald o yild χ /ndf =. LHC fill 336. TABLE XI WIDTHS OF THE GAUSSIAN FITS TO THE PULL DISTRIBUTIONS FOR MPX-5 AND MPX3. THE WIDTHS ARE GIVEN FOR UNCERTAINTIES RESULTING FROM THE NUMBER OF HITS SCALED BY A FACTOR Nhi /N cl FOR EACH MPX DEVICE, GIVEN IN TABLE V. THE WIDTHS ARE ALSO GIVEN FOR STATISTICAL AND SYSTEMATIC UNCERTAINTIES ADDED IN QUADRATURE, WHERE THE SYSTEMATIC UNCERTAINTIES RESULT FROM LUMINOSITY FLUCTUATIONS NOT DESCRIBED BY THE FIT FUNCTION (TAKING. AND.3. TABLE X INDICATES THE SIZES OF THE UNCERTAINTY APPLIED. LHC FILL 349. MPX 3 4 5 3 Widh pull sa. uncr..69.76.74..6.9 Widh pull oal uncr. (. sys..4.7.57.7.93.96 Widh pull oal uncr. (.3 sys...54.4.67..53
-.7.6 s = TV LHC fill 336.5 3 MPX.4 MPX.3 MPX3 5. MPX4. 4 MPX5 3 MPX3 -..4.5.6.7..9 + Fig. 6. Conour curvs of ( and ( +, givn a, for h fi rsuls of h avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX-5 and MPX3. Th dos in h cnr of h conours indica h fi valus. Th hi saisical uncrainis and sysmaic uncrainis from luminosiy flucuaions no dscribd by h fi funcion ar addd in quadraur, givn in Tabl IX, and hn scald o yild χ /ndf =. LHC fill 336. (daa-fi/fi (% MPX. LHC fill 349.6.4. -. -.4 -.6 -. MPX 9 3 Hours on Nov. Fig. 7. Dviaions bwn daa and h fi of h avrag numbr of inracions pr bunch crossing as a funcion of im sn by MPX. Th rlaiv dviaions bwn daa and fi hav an RMS of.%. Th saisical uncrainis / pr daa poin ar indicad and vary from abou.99.65% o abou.6.65% whr h facor.65 is h avragd raio of his pr inracing paricl. Th apparn srucur is similar o h on prviously discussd in h x. LHC fill 349. VII. CONCLUSIONS Th nwork of MPX dvics insalld in h ATLAS dcor cavrn has succssfully akn daa from o 3. Th sudy prsnd hr focuss on h proon-proon collisions from May o Novmbr. This sudy has dmonsrad ha h MPX nwork is wll suid for luminosiy monioring. Th slops of h long-rm im-sabiliy of h luminosiy masurmns boh from h hi and havy blob (hrmal nuron couning ar blow %. Th uncrainis for h hi analysis ar dominad by h sysmaic ffcs. For h havy blob (hrmal nuron analysis, h saisical uncrainis dscrib h flucuaions. Th MPX nwork has bn usd as wll o sudy all h van dr Mr scans prformd in in dail. I is dmonsrad ha h MPX luminosiy masurmns can cop wih a collision ra / of h on characrisic for physics daa-aking. Alhough no spcifically dsignd for luminosiy masurmns, h MPX nwork givs rliabl supplmnary informaion for h luminosiy drminaion of LHC proon-proon collisions. I is dmonsrad ha h prcision of h MPX nwork is sufficin o sudy h undrlying mchanisms of h LHC luminosiy rducion. Th valuaion of h variaions around h fid im-dpndnc of h LHC luminosiy is found o giv imporan informaion on h prcision of h individual MPX dvics. This sudy shows ha h rlaiv uncrainy on h luminosiy masurmn is blow.3% for on minu inrvals. A nwork of TPX dvics (upgradd succssors of h MPX dvics has bn insalld as h rplacmn of h MPX nwork in prparaion for h Run- LHC opraion []. ACKNOWLEDGMENT Th auhors would lik o hank warmly h ATLAS Luminosiy Group for usful discussions and inracions. Th projc is suppord by h Minisry of Educaion, Youh and Spors of h Czch Rpublic undr projcs numbr MSM 649, LA 3, LG 39 and LG 33, and h Naural Scincs and Enginring Rsarch Council of Canada (NSERC. Calibraion masurmns wr prformd a h Pragu Van-d-Graaff acclraor fundd by h Minisry of Educaion, Youh and Spors of h Czch Rpublic undr projc numbr LM 3. REFERENCES [] ATLAS Collaboraion, Th ATLAS Exprimn a h CERN Larg Hadron Collidr, JINST, vol. 3, p. S3,. [] Mdipix- Collaboraion, Projc Wbpag [Onlin]. Availabl: hp://mdipix.wb.crn.ch/mdipix, 5. [3] M. Campbll, C. Lroy, S. Pospisil, and M. Suk, Masurmn of Spcral Characrisics and Composiion of Radiaion in ATLAS by MEDIPIX-USB Dvics, Projc Proposal [Onlin]. Availabl: hps://dms.crn.ch/documn/565, 6. [4] M. Campbll al., Analysis of Radiaion Fild in ATLAS Using - Daa from h ATLAS-MPX Nwork, ATL-GEN-PUB-3-, 3. [5] ATLAS Collaboraion, Improvd luminosiy drminaion in pp collisions a s = 7 TV using h ATLAS dcor a h LHC, EPJC, vol. 73, p. 5, 3. [6] CMS Collaboraion, CMS Luminosiy Basd on Pixl Clusr Couning - Summr 3 Upda, 3, CMS Public Analysis Summary CMS- PAS-LUM3-. [7] A. Sopczak, Luminosiy Monioring in ATLAS wih MPX Dcors, on bhalf of h ATLAS and Mdipix- Collaboraions, in Proc. IPRD3, 7 Ocobr 3, Sina, Ialy. JINST, vol. 9, p. C7, 4. [] S. van dr Mr, Calibraion of h ffciv bam high in h ISR, ISR-PO/6-3 CERN Rpor, 96. [9] LHC, Prfromanc and Saisics hp://lhc-saisics.wb.crn.ch/lhc-saisics, 5. [] K. Eggr, K. Honkavaara, and A. Morsch, Luminosiy considraions for h LHC, 994, CERN-AT-94-4, LHC-NOTE-63. [] CMS Collaboraion, Th CMS xprimn a h CERN LHC, JINST, vol. 3, p. S4,. [] C. Lroy, S. Pospisil, M. Suk, and Z. Vykydal, Proposal o Masur Radiaion Fild Characrisics, Luminosiy and Inducd Radioaciviy in ATLAS wih TIMEPIX Dvics, Projc Proposal [Onlin]. Availabl: hp://cds.crn.ch/rcord/64697, 4.