A Short introduction to HBT Qinghui Zhang HBT'04 Q. H. Zhang 1 Heavy Ion Collisions requires understanding particle distributions in coordinate and momentum space HBT'04 Q. H. Zhang 2 GGLP HBT Goldhaber, Goldhaber, Lee and Pais (1960) First application of “intensity interferometry” in particle physics Hanbury-Brown and Twiss (1950’s) Revolutionary development of intensity interferometry in astronomy B-E HBT'04 Bose-Einstein (1920’s) Role of Bose statistics in fluctuations Q. H. Zhang 3 HBT'04 Q. H. Zhang 4 1 X Source y 2 Neglects Momentum dependence of source Quantum mechanics up to x and y Final State Interactions after x and y C2(q) contains shape information HBT'04 Q. H. Zhang 5 That is, squared Fourier Transforms measure (only) relative separations of Sphere vs source coordinates hemisphere Very different sources r(x) can give very similar distributions in relative separation qOUT R HBT'04 Q. H. Zhang 6 C 2 (q ) 1 | r (q ) | 2 , r (q ) r ( x )e iqx d 4 x q p1 p 2 ( E 1 E 2 , p1 p 2 ) (q 0 , q ) q x (q 0 t q r ) Fourier Transform provides one time and three space extensions of source BUT: 2 2 2 2 E1 E 2 p1 p 2 q0 E1 E 2 E1 E 2 E1 E 2 p1 p 2 ( p1 p 2 ) q V PAIR E1 E 2 That is, q0 is not an independent quantity in the F.T. (e.g., note that q 0 | q | ) Fourier Transform has support in only half of the q0-q plane HBT'04 Q. H. Zhang 7 C 2 (q ) 1 | r (q ) | 2 , r (q ) r ( x )e iqx d 4 x q 0 q V PAIR q x (q V PAIR t q r ) q ( r V PAIR t ) iq ( r V PAIRt ) r (q ) r (q ;V PAIR ) r ( x )e d4x So Fourier transform (+ two-particle kinematics) already provides conclusion: C 2 (q ) 1 | r (q ) | C 2 (q ; V PAIR ) ( r V PAIR t ) 2 2 Additional dynamic effects (expansion, thermal source, …) will also lead to systematic dependence of extracted “radii” on VPAIR (or mT or kT or …) HBT'04 Q. H. Zhang 8 So ur c e q Y-Axis P1 K P2 qSIDE Beam Axis ZAx is X-Axis qLONG Source qOUT There are other decompositions... HBT'04 Q. H. Zhang 9 1 1 x 2x 2 y 2y 2 z 2z 2 t 2t 2 r (rr(,rt ,)t~) ~exp[ exp[ ( ( 2 2 2 2 2 2 2 )]2 )] 2 2RR RR RzRz x x y y 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 22 2 r (rq(q; V; V ) ) exp[ exp[ (q(xq xRR q yq yRR q zq zRzRz (q(q V V ) ) )] )] PAIR PAIR x x y y PAIR PAIR 22 Simplest possible (yet still very useful) case: e ur c So q Choose frame so that x = “Out” (that is, parallel to VPAIR) y = “Side” z = “Long”: Y-Axis P1 K P2 qSIDE Beam Axis ZAx is X-Axis z qLONG y x Source qOUT 2 2 2 2 2 2 (q x R x q y R y q z Rz (q V PAIR ) 2 2 ) q SIDE R SIDE q LONG R LONG qOUT ROUT with ROUT R x V PAIR 2 2 Symmetry HBT'04 2 2 2 2 2 2 2 2 " " R y R x ROUT R x V PAIR 2 R SIDE R y 2 Q. H. Zhang 2 2 2 10 2 Most experiments with charged tracking and particle identification have HBT results: BNL AGS: E859, E866 E877 E895 CERN SPS: NA44 NA49 WA98 RHIC PHENIX HBT'04 STAR Q. H. Zhang 11 The Great Puzzle: Why so little variation with ECM ? Why are lifetimes () ~ 0 ?? ( especially at RHIC ) HBT'04 Q. H. Zhang 12 systematic errors in experimental data is required: Definition of C2 Coulomb corrections Role of l, dependence of R on same Contributions from resonances (especially when comparing between different experiments) A better understanding of theoretical assumptions and uncertainties is required: HBT'04 Resonance contributions Lorentz effects Modeling of expansion, velocity profiles Q. H. Zhang 13 Our favorite (charged) bosons never have pure plane-wave states HBT'04 Even our ideal-case Fourier transform becomes a “Coulomb transform” This particular case is easy to treat analytically via “Gamow” or “Coulomb” corrections: Q. H. Zhang 14 Some indication from both AGS and SPS that RSide(--) > RSide(++) E877 (AGS): RSide(--) ~ (1.5 +/0.4)*RSide(++) R (--) > R (--) TSide TOut NA44 (SPS): RTSide(++) < RTOut(++) Ze Note external Coulomb HBT'04 •Systematic in RSIDE •Cancels in ROUT - - Q. H. Zhang Ze 15 r (q ) r ( x )e iqx d 4 x r (q 0) 1 C 2 ( q 0) 2 versus observed value of 1.1-1.5 This is parameterized away via C2(q) = 1+ l | r(q)|2 , l ~0.1-0.5 “Understood” via assumed resonance contributions: HBT'04 Q. H. Zhang 16 How do resonances affect l ? By creating a “halo” around the source “core”: Some culprits: r (G = 150 MeV S ~ 1.3 fm ) K* K (G = 50 MeV S ~ 4 fm ) w (G = 8.4 MeV S ~ 20 fm ) , h h (G = 0.2 MeV S ~ 200 fm ) 2(1-fCORE )fCORE (1-fCORE )(1-fCORE ) fCORE fCORE (and many more…) Core 2 Effect: C2 develops structure near q=0 to describe the various sources: 1 Momentum resolution pHBT'04 p Halo Combination C2(q) 0 Q. H. Zhang 0 (~ 1%) (~ 1%) p [Ge V/c] 50 q (MeV/c) 100 17 Using Wigner function: See for example: Chao, Gao and Zhang Phys. Rev. C 49,3224 (1994) Use ~this in cascade codes HBT'04 Q. H. Zhang 18 Expect C2 to depend on q and K B.R. Schlei and N. Xu: HBT'04 Q. H. Zhang 19 RQMD vs. Data for HBT Final state phase space points xi are weighted by 1 HBT'04 + cos[(pa-pb)(xa-xb)]Q. H. Zhang 20 This approach is Plausible WRONG Final state phase space points xi are weighted by 1 + cos[(pa-pb)(xa-xb)] with Advantage: Uses precisely what codes produce Disadvantage Can lead to non-physical oscillations in C2 First noted empirically by Weiner et al and Later matin et al. Theoretical analysis by Q.H. Zhang et al. Phys. Lett. B 407, 33 (1997) Quantitatively, is it significant for heavy ion HBT'04 collisions? Q. H. Zhang 21 Answer 1: No (Empirical) Study these effects with a source that explicitly parameterizes x-p correlations: Examples for R=5fm, P=200 MeV/c 500 Naive 400 300 Correct Usual (used in MC's) 2 K 0 p (MeV/c) 100 S=0 C2(q) 200 1 -100 -200 x -300 0 -400 0 100 q (MeV/c) -500 -10 -8 -6 -4 -2 0 R (fm) 2 4 6 8 200 10 500 Naive 400 Correct Usual (used in MC's) 300 2 0 p (MeV/c) 100 S=0.33 C2(q) 200 1 -100 -200 0 -300 0 100 q (MeV/c) -400 200 -500 -6 -4 -2 0 R (fm) 2 4 6 8 10 Naive 500 400 Correct Usual (used in MC's) 2 300 S=0.9 200 100 0 HBT'04 -100 -200 -300 -400 C2(q) -8 1 p (MeV/c) -10 Q. H. Zhang 22 0 0 100 q (MeV/c) 200 Q. Can one remove these pathologies from ~classical predictions? A. Yes (Q.H. Zhang et al.): Smear the phase space points (x,p) with minimum uncertainty wave-packets: with s ~ 1 fm Pictorially: 500 5 400 4 300 3 200 2 100 1 0 p (MeV/c) = -100 0 - -200 - -300 - -400 - -500 -8 -6 -4 -2 0 R (fm) 2 4 6 8 - 10 -10 -8 -6 -4 -2 0 R (fm) 2 4 This can be done analytically with parameterization on previous slide: x 500 400 300 200 100 0 p(MeV/c) -10 -100 -200 -300 -400 -500 -10 -8 -6 -4 -2 R HBT'04 0 (fm 2 4 6 8 10 ) Q. H. Zhang 23 6 8 10 What are the Lorentz properties of C2(q)? d 6n 3 3 E1 E 2 d 6n 3 3 dp dp dp1 dp 2 C 2 ( p1 , p 2 ) 3 1 23 Lorentz Invariant d n d n d 3n d 3n E1 E2 3 3 3 3 dp1 dp1 dp1 dp1 How to write our favorite practice Gaussian in an explicitly Lorentz invariant way? Answered in F.B. Yano and S.E. Koonin, Phys. Lett. B78, 556 (1978). 1 r2 t2 r ( r , t ) ~ exp[ ( 2 2 )] 2 R 1 2 2 r (q ; V PAIR ) exp{ [ | q | R (q V PAIR ) 2 2 ) ]} 2 so 2 2 2 C 2 ( p1 , p 2 ) 1 | r (q ; V PAIR ) | 1 exp{ | q | R (q V PAIR ) 2 2 } i.e. , not explicitly Lorentz covariant Fix by introducin g source four - velocity u s (1, v s ) 2 q q q0 | q |2 , q u s (q 0 q v s ) q 0 (in source rest frame) Then CHBT'04 qH. ) 2 Zhang R (q u) 2 [ R 2 2 ]} 2 ( p1 , p 2 ) 1 exp{( q Q. 2 i.e. , explicitly Lorentz covariant 24 Can Lorentz effects “distort” information? Apply Yano-Koonin-Podgoretzsky p1 result to another toy model: VPAIR uS Beam axis p2 Apply this to ~1-d motion in “Out” direction: Study argument of exponentia l ARG (q q ) 2 R (q u) 2 [ R 2 2 ] for various cases : Source at rest v S 0 : 2 ARG | q | 2 [ R 2 V PAIR 2 ] 2 ROUT R 2 V PAIR 2 Pair at rest V PAIR 0 : *2 *2 ARG | q | 2 s [ R 2 v S 2 ] 2 2 ROUT s [ R 2 v S 2 ] Numerical example : | V PAIR | 1 , | v S | 0.7 2 *2 *2 ROUT (0.4) 2 [ R 2 2 ] 2 Nota Bene: This last case allows ROUT<RSIDE HBT'04 25 even for ~ RSIDE Q. H. Zhang The chief lesson from last 10-15 years of (theoretical) HBT work: The “radii” are a (potentially) complicated mixture of Space-time distribution Spatial flow gradients Temperature gradients This average ( r V PAIR t ) 2 is over space, time, flow profile, temperatur e, ..... More correct terminology: “radii” “lengths of homogeneity” In the presence of source dynamics, “radii” depend on mean pair (energy, momentum, transverse mass, …) HBT'04 Q. H. Zhang 26 But : ROUT/RSIDE trend completely eliminates one “hydro” calculation HBT'04 Soff, Bass, Dumitru, Phys. Rev. Lett. 86, 3981 (2001) Q. H. Zhang 27 Plotted versus KT, there is an amazing consistency between data from AGS, SPS, and RHIC, spanning a factor of 100 in CM energy(!): RHIC AGS HBT'04 SPS Q. H. Zhang 28 There is now an extensive experimental and theoretical literature built on 20+ years of “HBT” studies We’re now ready to start doing things right: Experimentally Theoretically HBT'04 Measure resonance contributions Measure like and unlike particle effects Understand systematic errors Cascade codes for RHIC Right parametrization Coulomb systematics Q. H. Zhang 29 Multipion correlations Each events has large number of pions, therefore, multi-pion correlation will affect Two-pion interferometry (1): If the pion correlation formula is right? (2): If it is not right, which kind of formula should we use! HBT'04 Q. H. Zhang 30 Seems right! Zhang, Phys. Rev. C58, 22 (1998) Proved that for a class model, the present pion correlation function formula is right. However generally, the formula is not right! Zhang, Phys. Rev. C59,1646 (1999). HBT'04 Q. H. Zhang 31 The formula is : C(p1,p2)= A(p1,p2)[1+R(p1,p2)] A(p1,p2) is a function of p1,p2. R(p1,p2) can be written as the same as above. HBT'04 Q. H. Zhang 32 Three-pions correlations HBT'04 Q. H. Zhang 33 HBT'04 Q. H. Zhang 34 HBT'04 Q. H. Zhang 35 HBT'04 Q. H. Zhang 36 HBT'04 Q. H. Zhang 37 HBT'04 Q. H. Zhang 38 Some of slides taken from W. A. Zajc, R. Wilson. Thank you! HBT'04 Q. H. Zhang 39