# Even More Perturbative QCD ...

Lorenzo Magnea Universit` di Torino a I.N.F.N. Torino magnea@to.infn.it XI Seminario Nazionale di Fisica Teorica

Abstract These lectures describe the treatment of mass divergences in perturbative QCD, concentrating on hadron production in electron-positron annihilation. I perform an explicit and detailed one-loop calculation, and use it to infer some results and techniques valid to all orders in perturbation theory. I introduce some of the tools necessary to prove factorization theorems, and show how they can be used also to resum certain classes of logarithmic contributions to all orders in perturbation theory.

Ubi maior ... : G. Sterman: An introduction to quantum field theory. P. Nason: http://castore.mib.infn.it/~nason/misc/QCD... M.L. Mangano: http://home.cern.ch/~mlm/talks/cern98... G. Sterman: hep-ph/9606312.

– Parma, 05/09/2002 –

Outline

• On mass divergences in perturbative QCD – Mass divergences, low energies e long distances. – Cancellation for physical observables, KLN theorem. – Factorizable and IR/C safe observables in PQCD. • An explicit example: Re+ e− – Deﬁnitions, cut diagrams, tree level. – O(αs ): the calculation. – Approximations and observations. • Other infrared ﬁnite quantities – Sterman–Weinberg jets. – Event shapes. • Methods for all-order calculations – Singularities of Feynman diagrams and trapped surfaces. – Landau equations and Coleman-Norton picture. – Infrared power–counting and ﬁniteness of Re+e− . • Factorization and resummation – Multi-scale problems and large logarithms. – From factorization to resummation. – Examples: the quark form factor; the thrust.

– Parma, 05/09/2002 –

1

**Mass divergences: qualitative discussion
**

• Fact: in quantum ﬁeld theory, two kinds of divergences are associated with the presence of massless particles. – Infrared (IR). Emission of particles with vanishing four momentum (λDB → ∞); in gauge theories only; they are present also when matter particles are massive. – Collinear (C). Emission of particles moving parallel to the emitter; they are present if all particles in the interaction vertex are massless. • Example: a massless fermion emits a gauge boson in the ﬁnal state.

k

M

p+k

p

→

− igu(p)/(k)ta

i(p + k / /) M, (p + k)2 + iε

**Singularities: 2p · k = 2p0k0 (1 − cos θpk ) = 0 , → k0 = 0 (IR); cos θpk = 0 (C). Note: p0 = 0 not a problem (singularities are always integrable).
**

– Parma, 05/09/2002 – 2

• Origin of mass singularities – In covariant perturbation theory ( pµ conserved in every vertex; intermediate particles generally oﬀ–shell): the emitting fermion is on–shell, so that it can propagate indeﬁnitely. – In time-ordered perturbation theory (all particles are on– shell, energy however is not generally conserved in the interaction vertices): the IR/C emission vertex conserves energy, so it can be placed at arbitrary distance from the primary process. – Mass divergences originate from physical processes happening at large distances. • Available terapies . – The sickness is serious. Because of mass divergences, the S matrix cannot be constructed in the Fock space of quarks and gluons – Observe. Mass divergences are associated with the existence of experimentally indistinguishable, energy degenerate states. All physical detectors have ﬁnite resolution in energy and angle. – KLN Theorem. Physically measurable quantities (such as transition probabilities, cross sections, after summing coherently over all physically indistinguishable states) are ﬁnite, mass divergences cancel.

– Parma, 05/09/2002 – 3

In an asimptotically free ﬁeld theory the limit m → 0 and the high-energy limit formally coincide. • The situation in perturbative QCD – Long distance (d > 1fm).
– Parma. Consider a theory deﬁned by its hamiltonian H . and let D (E0) be the set of eigenstates of H characterized by energies E0 − ≤ E ≤ E0 + . 05/09/2002 – 4
. Then the quantity
P (E0. ∗ long-distance eﬀects can be isolated in universal factors. with = 0. Masses acquire a scale dependences such that m2(µ2) → 0 as µ2 → ∞. depending on the initial state but not on the hard process being studied (factorizable cross sections). ) ≡
i. or low energy (E < 1GeV) ∼ ∼ physics is not perturbatively calculable. – Working at the perturbative. such that ∗ long-distance eﬀects are suppressed thanks to cancellations (IR–safe cross sections).j∈D (E0 )
X
P (i → j)
is ﬁnite in the massless limit to all orders in perturbation theory • Note. Let P (i → j) be the transition probability per unit volume and per unit time from eigenstate i to eigenstate j .• KLN theorem. partonic level one identiﬁes suﬃciently inclusive cross sections. – The KLN theorem is not directly applicable when a sum over initial states is necessary (we have no control on the structure of hadronic initial states).

with an infrared regulator (e. valid ` ´ modulo O (ΛQCD /Q)p corrections. 2 µ
2
(
m (µ ) . such that
σpart = f m2 µ2 F ! ∗ σpart b Q2 µ2 . g. cross sections σpart are derived with a perturbative calculation. admitting a perturbative expansion in powers of αs (Q2) 1. αs (µ2 ). αs (µ2). the goal is constructing factorizable quantities. One computes
σpart = σpart Q .
σpart = σpart ! Q2 .
• Factorization. 0} + O 2 µ ( m2 µ2 !p . proved at parton level. 05/09/2002 – 5
. • With hadrons in the initial state. is transcribed at hadron level. are interpreted as estimates of the corresponding hadronic quantities. )! . µ2
2
2
)!
.
• IR–safe quantities are selected. having a ﬁnite limit when the IR regulator removed ( → 0. b
– Parma. m2 (µ2) → 0). requiring the presence of at least one hard scale Q2. inclusive quantities. {0. Distribution functions f are measured.: = 2 − d/2 < 0). F µ2 µ2 ! +O m2 µ2 F !p ! .The strategy of perturbative QCD
• All calculations are performed at partonic level.
• These partonic.

2
2
2
” e2 µ2 1 − “ µν 2 σtot (q ) = −g Hµν (q ) . k2 )H (q ) . H (q ) = e µ qf
X
This leads to
−g
” “ µ ν 2 µν H(q 2 ) . to leading order in α. 2q 2
Current conservation implies qµHµν = qν Hµν = 0. 2 2q 4
X spin
normalized dividing out the total muon-pair production cross section “ ”
Re+ e− ≡ σtot e+ e− → hadrons σtot (e+ e− → µ+ µ−)
In d = 4 − 2 . 05/09/2002 – 6
.
σtot (q ) =
µν 2
1 µν 2 Lµν (k1 . k2 ) = k1 k2 + k1 k2 − k1 · k2 g µν . q4 X µν 2 2 2 2 d d 0|Jµ(0)|X X|Jν (0)|0 (2π) δ (q − pX ) . (q ) = q q − q g
2
µν
Hµν (q ) = (3 − 2 ) q H(q ) .An explicit example: Re+e−
The prototype of IR–safe cross sections is the total annihilation cross section for e+ e− → hadrons. 2q 4 3 − 2
2
– Parma.
Z 1X 1 X dΓX σtot (q 2 ) = |M(k1 + k2 → X)|2 . so that
H
µν
´ e2 µ2 ` µ ν ν µ L (k1 .

.
• the rules apply for ﬁxed ﬁnal state momenta. . • for particles with spin = 0 the cut carries the sum over polarizations. γµi γ5 . if needed. the loop momentum integral for cut loops. γµn ω2 = ˆ ˜ ω2 γµn . becomes the phase space integral. (ta)ij
Note that
= (ta )ji. . 05/09/2002 – 7
. . . ˆ ˜∗
as well as hermiticity of group generators. γµ2 γµ1 ω1 . σµν . . . σµν . • cut fermion loops carry the expected minus sign. γµi γ5 . Pictorially
ui uf ui uf
∗
vi ui
vf
vi ui
vf
∗
vi
p vi
p
2πδ+ (p2 ) p /
To the right of the cut all explicit i’s in the Feynman rules and all momentum components change sign. .A technical interlude: cut diagrams
A useful representation for |M|2 can be constructed in terms of cut diagrams. . .
– Parma. Consistency with spinor and color algebra is easily veriﬁed using
` ˆ ˜ ´∗ ω 1 γµ1 γµ2 . . .

the trace evaluates to 4(1 − )q2 . e e f
– Parma. use the two δ+ distributions to perform k0 and |k| integrals. Γ(d)
In d = 4 − 2 one ﬁnds then
Γ(2 − ) 2 µ −Hµ = 2 α q Γ(2 − 2 ) 4πµ2 q2
!
whence the famous result.
f
X 2 (0) R + − = Nc qf . →
X 2 Nc qf . /γ / / d−2 p
q 4
2
!1−
Ω2−2 . for
X 2 4πα2 σtot = Nc qf 3q 2
f
→ 0.Exercise: Re+e− at tree–level
k 2 2 1− σtot (q 2 ) = e µ4 3−2 (−g µν ) 2q q µ ν
µ 2 −Hµ = e2 µ2 qf
Z
dd k (2π)
In he center of mass frame (k − q)2 = q2 − 2 q2k0 . (2π)2−2
The d-dimensional solid angle is given by the classic formula
2dπd/2 Γ(d/2) Ωd = . 05/09/2002 –
8
. Summing over quark colors and ﬂavors one ﬁnds
X 2 µ −Hµ = 2(1 − )e2 µ2 Nc qf
f
` ´ δ+(k2 )δ+ ((k−q)2 )Tr k µ (k − q )γ µ .

– Parma.
Real emission
It is convenient to compute separately the transition probability and three-body space.R) −Hµ = Z δ+ (p )δ+(k )δ+ ((p + k − q) ) −Hµ µ (2π)2d−3 ddpdd k
2 2 2
h
i
. One can integrate over the energies of p ˆ and k using the respective δ+.
Summing over positions of the cuts one obtains real and virtual diagrams in turn. 05/09/2002 – 9
. ˆ δ+ ((p + k − q) ) = ϑ p0 δ s − 2 s(p + k) ˆ ˆ
2
“
ˆ where p = |p|. k = |k|.
The transition probability depends on a single polar angle. Following the rules for cut diagrams one ﬁnds
h i µ (1. All angular integrations are trivial except the one on u. Let u ≡ cos θpk in the center of mass frame: one ﬁnds
” “ ” √ ˆ + 2pk(1 − u) . We examine them separately.t.Radiative corrections
Enumerating graphs contributing to the one-loop correction is easy in terms of cut diagrams
h
µ −Hµ
i(1)
=
+ c.

+ 2 (2p · k)
and can be simpliﬁed using Tr(tata) = NcCF . p·k = sxyz/2.
i 0 h Tr γµ (p + k)γσ pγ µ (−p − k)γ σ p / / / / / / X 2 2 2 µ a 2 2 qf g µ Tr(ta t ) @ −Hµ = −2e µ (2p · k)(2p · k) f h i1 σ (p + k)γ µ p Tr γµ (p + k)γσ pγ / / / / / / A . / /p / /q
γµ p k /γ µ / /q
and with the identiﬁcations p·q = sx/2. 1 − yz 1 − yz 0
The transition probability can be computed from the Feynman rules. 05/09/2002 – 10
. x= √ . / 4p · k − 2 p k . // −2q k / + 2 p k / .R) −Hµ = 8 (2π)5−2
„ «1−2 Z 1 Z 1 s 1−2 dxx dzz 1−2 2 0 0 „ « Z 1 i 1 1−z h − µ × dy [y(1 − y)] δ x− −Hµ . k·q = sz/2.Introduce dimensionless variables (quark and gluon energy fractions)
ˆ 2k 2p ˆ z= √ . s s and deﬁne y = (1 − u)/2. The result is h i 1 Ω2−2 Ω1−2 µ (1. Cliﬀord algebra identities such as
γµ p γ / γµ p k / /γ
µ µ
= = =
−2(1 − )p .
– Parma.

05/09/2002 –
11
.
Note: mass singularities are regulated choosing < 0. • Collinear: y−1− e (1 − y)−1− . singular when y → 0 (gluon collinear to the quark).R)
X 2 Ω2−2 Ω1−2 2 = 2Nc CF qf ααs (1 − ) q (2π)3−4
f
2µ q2
2
!2
(1 − z)
(1 − yz)
−2−2
z
1−2
[y(1 − y)]
−
(1 − yz)2 − yz 2 (1 − y)
+ y 1+ !# .
h −Hµ
µ
i(1. The y and z integrations yield Euler B functions (typical of oneloop calculations in one-scale problems). Expanding around = 0 one gets the ﬁnal result for real emission. using the remaining δ .
1− × Γ(2 − 2 )
»
2
2
+
3
19 −π + + O( ) 2
2
– Parma. displaying the double infrared–collinear pole.R)
X 2 αs 2 = Nc C F α qf q π
f
4πµ q2
2
!2 – . and y → 1 (gluon collinear to the antiquark). gives a pole in when the gluon energy z tends to 0.One can integrate out the quark energy fraction x. The expression simpliﬁes considerably and one gets
h
µ −Hµ
Z 1
0
» dzdy (1 − )(1 − z)−2 (1 − yz)−2−2 z 1−2 (1 − y)1−
1−2
i(1.
1
One recognizes the announced singularities
• Infrared: z −1−2 .

) ≡ p1 .Virtual contribution
Purely virtual contributions to the production amplitude are given to all orders by the quark form factor.
Γµ (p1 . 05/09/2002 – 12
. with an overall factor given by 2 ReΓ.
„ µ ∂ ∂ + β( . • The form factor is renormalization group invariant (it has a vanishing anomalous dimension). p2 . as a consequence of the conservation of the electromagnetic current.
p1 “ ” 2 Γν p 1 .
– Parma. =
ν p2
The calculation greatly simpliﬁes by taking into account the general properties of the form factor. αs (µ2 ).
As a consequence.
QCD does not violate the QED Ward identity. multiplying the Dirac structure of the threee amplitude. 2 µ !
= −ieqf u(p1 )γµv(p2 ) Γ
. Z1 = Zψ . the transition probability is proportional to the tree–level result. αs .
• For massles quarks. p 2 . the form factor is expressed in terms of a single scalar function. µ . αs ) ∂µ ∂αs « Γ Q . µ2 . p2 |Jµ (0)|0 q2 . µ2
2
!
=0.

. it is necessary to include these graphs on one of the two lines only. including the respective counterterms.. furthermore it depends on a cancellation of IR and UV eﬀects . p 2 . k2 (p1 − k)2 (p2 + k)2 (2π)d
– Parma. i – Note! This is false in general in axial gauges (∃ n · pi). the vertex correction
(1) Γν
“ ” 2 p1 . µ . −1/2 every external line must be multiplied times Rψ . • In Feynman gauge and in dimensional regularisation all 1PR graphs with the loop on the external fermion line vanish because they are expressed by scale-less integrals (p2 = 0). At one loop these observations are summarized by the identity
0
One is left with only one graph to be computed. according to the reduction formulas. 05/09/2002 –
13
. Since. =
which can be written as
(1) 2 2 Γν = −eqf g µ CF
Z
dd k u(p1 )γσ (p 1 − k ν (p 2 + k σ v(p2 ) / /)γ / /)γ . reconstruct the residue Rψ of the quark propagator.• Reducible (1PR) graphs on each fermion line.

. then isolate integrals with diﬀerent / / powers of k.
• It may be useful to decompose tensor integrals ` la Passarino– a Veltman. 05/09/2002 –
14
..
h
Z Z Z
1 dd k .
α β
α β
α β
β α
– Parma. Summarizing
• Sistematically use the mass–shell conditions (Dirac equation).The steps to compute this diagram are standard.
(1) 2 2 2 α Γν = −eqf g µ CF u(p1 ) 2q γν I0 + 2(γν γα p 1 − p 2 γα γν )I / /
−γ γα γν γβ γσ I
σ
αβ
• The tensor integrals
I0 =
i
v(p2 ) . 1 2 g
αβ
I3 + p1 p1 I4 + p2 p2 I5 + (p1 p2 + p1 p2 ) I6 . (2π)d k2 (p1 − k)2 (p2 + k)2
Iα
=
Iαβ
=
can be computed with the usual Feynman parametrization. Iα may have collinear poles . and only Iαβ can have UV divergences. u(p1 )p 1 = p 2 v(p2 ) = 0. (2π)d k2 (p1 − k)2 (p2 + k)2 dd k kα . in order to get directly the scalar form factor Γ. (2π)d k2 (p1 − k)2 (p2 + k)2 kα kβ dd k . The result is
Iα I
αβ
= =
pα I1 + pα I2 . – Note: only I0 can have IR divergences.

the poles cancel and one can take the limit → 0.
– Parma.• In terms of the scalar integrals I1. . the (classical) result is
X 2 (0) R + − = Nc qf e e f „ αs + O(α2 ) 1+ s π « . 05/09/2002 –
15
. . ! » –
• The ﬁnal result for the form factor is
αs (1) Γ = − CF 4π 4πµ −q 2
2
h
2
2
i
Γ2 (1 − )Γ(1 + ) Γ(1 − 2 )
2
2
+
3
+ 8 + O( ) . Because of the double pole. the virtual contribution has the same IR-C poles as real emission.
Result
We observe that. as expected. Summing the two. the factor exp(−iπ ) must be expanded to second order in . with the result
X 2 4πα2 σtot = Nc qf 3q 2
f
« „ αs 3 2 CF + O(αs ) . where CF = 4/3. 1+ π 4
For SU (3). It generates numerically important contributions. .
When taking the real part one must use
(−q 2 + iε)− = (q 2 )− e−iπ . .
Note! The sign of q2 is dictated by the Cutkosky rules. I6 one ﬁnds
Γ
(1) 2 2
= g µ CF 4(1 − ) I3 − 2q (I0 + I2 − I1 + (1 − )I6 ) .

p. i k. a µ p .j µ
# / /)ε ε (k)(p + k µ / / /)Γ Γµ (p + k /(k) aµ a Aij = gtij u(p) − v(p ) .j
Aij =
aµ
k. 05/09/2002 – 16
. / /
The result is
˛ aµ ˛ Aij ˛ = gta ij soft " # p·ε p ·ε µ − A0 . just as was the case for the virtual diagram. p·k p ·k
where Aµ = u(p)Γµv(p ) is the Born amplitude (whatever the 0 explicit form of the vertex Γµ). 2p · k 2p · k
"
When the gluon is soft one can
• neglect k in the numerator. in the IR and C limits the amplitude for the emission of a real gluon becomes proportional to the Born amplitude. a p . and in the deﬁnition of p .
– Parma. • commute p and p in order to impose the mass-shell condition. i p. / / p v(p ) = u(p)p = 0. This result can be made systematic introducing the soft approximation.Soft approximation
The cancellation that was just exhibited is possible only because.

p·k p ·k
To get the cross section one integrates over phase space
soft 2 σq q g = g C F σq q ¯ ¯
Z
d3 k 2p · p . 2|k|(2π)3 p · k p · k
In the center of mass frame (q = 0) and in the soft approximation the quark and the antiquark are still back to back. which is over colors and polarizations (using ν allowed in this case). The transition probability can be computed summing P εµ ε∗ = −gµν . One then recovers the structure of IR and C singularities. they only detect the global color charge and the direction of motion These considerations generalize to multiple emission. |k| (1 − cos θpk )(1 + cos θpk ) −1 0
– Parma.
Soft cross section It’s easy to recover the singular part of the real emission cross section.
|Asoft | = g CF |A0 |
2 2 2
2p · p .Remarks
• The soft amplitude is gauge–invariant (it vanishes if ε ∝ k). • Identical considerations apply to gluon emission from gluons. 05/09/2002 –
17
. Long-wavelength gluons cannot analyze the short-distance properties of the emitter (spin.
αs soft σq q g = σq q C F ¯ ¯ π Z 1 Z ∞ d|k| 2 d cos θpk . internal structure). • Soft gluon emission has universal characters.

or the approximation may break down. v . v± = √ (v 0 ± v 3 )/ 2. v = v . – Note: the approximation is not uniformly valid.
• When kµ pi · k in denominators. containing the virtual double pole.
– Parma.
• Consider for example the integral I0. • Using light-cone coordinates one can set
” “ − (p ) = 0. (p ) . with some care. v⊥ . one can neglect k2 with respect to
” “ + p = p .
→
COLL . 0⊥ .
µ µ
p q 2 . ∀µ. |k | ∼ λ q 2 . as well as k in numerators (eikonal approximation). in some cases it becomes necessary to deform integration contours. 05/09/2002 –
18
. ” “ µ + − – Note: For a generic four-vector. q q 2 2 . v 2 = 2v + v − − |v⊥|2 . 0. In the eikonal approximation and in d = 4
1 (eik) = I0 32π4 q 2 Z dk+ dk−d2 k⊥ . ∼λ q ⊥ k ∼λ
µ
→
IR . 0⊥ . (−k− + i )(k+ + i )(2k+ k− − |k⊥ |2 + i )
There are three integration regions giving rise to divergences. One can parametrize them introducing a scaling variable λ. ∀µ .Virtual diagrams The soft approximation can be applied to virtual diagrams as well. k
q 2 . according to q
k
±
q ∼ q2 .

– Parma. angular ordering) • It links perturbative and non–perturbative regimes (resummations. In that frame the quark and the antiquark are typically emitted forming a small angle (θpp π). and one has
soft dσq qg ¯
=
1 − cos θpp αs d|k| dφk dσq q CF d cos θk ¯ π |k | 2π (1 − cos θpk )(1 − cos θp k ) dσq q CF ¯ ´ αs d|k| dφ 1 ` d cos θk k Wq + Wq ¯ π |k | 2π 2 1 1 − . ¯ The full angular distribution is positive deﬁnite.
• It displays the universal properties of soft color radiation (color transparency.Angular ordering
The soft approximation is of great practical relevance in perturbative QCD. hadronization Monte Carlo)
The simplest example of angular ordering is the diﬀerenzial cross section dσq qg in a frame in which the decaying photon has a large ¯ momentum q. The partial distributions Wi are not positive deﬁnite. but singular for emissions parallel to either the quark or the antiquark. (1 − cos θpk ) (1 − cos θp k )
=
where
Wq =
(1 − cos θpk )(1 − cos θp k )
1 − cos θpp
+
while Wq is found exchanging p ↔ p . 05/09/2002 – 19
. but they enjoy special properties.

Properties of Wq e Wq ¯ • Wq is singular only when cos θpk → 1. 05/09/2002 – 20
.
Z 2π ” “ 1 2 dφ Wq (φ) = Θ θpp − θpk . so that color singlet parton clusters are likely to be formed inside collimated particle beams (jets). Comments • This property can be generalized to higher orders. Gluons which are radiated later are forced on average to the inside of the cones deﬁned by previously emitted gluons and quarks. ¯ • The azimuthal average of Wq (with respect to the axis deﬁned by p) vanishes if θpk > θpp . Thus: the distribution of soft radiation on average is given by a sum of uncorrelated contributions from the quark and from the antiquark. 2π 0 1 − cos θpk
as can be proven using
cos θp k = cos θpk cos θpp + sin θpk sin θpp cos φ . each vanishing outside the cones built rotating the direction of one of the fermions around that of the other one. ¯ • Azimuthal averages are positive deﬁnite and can be interpreted as probability distributions for the emission of soft gluons independently of the quark and of the antiquark.
An identical equation applies to Wq . • Perturbative evolution in the soft limit is local in phase space. while the opposite is true for Wq .
– Parma.

Thus it is not necessary to integrate real emission over the entire phase space in order to obtain a ﬁnite result. such that gluon emission be integrated over IR and C conﬁgurations. All other events are three-jet events. as expected. • Virtual contributions are two-jet events. At parton level:
• All events are two-jet events at leading order. such that all the energy. except at most a fraction . or C (with any energy). Therefore the partonic two-jet cross section is ﬁnite. 05/09/2002 – 21
.Sterman–Weinberg jets
Momentum conﬁgurations responsible for singularities and needed for their cancellation are infrared and collinear. • At O(αs) two-jet events are those in which the gluon is IR (and emitted in any direction). it is suﬃcient to consider suﬃciently inclusive observables. ﬂows into the cones. Prototype: two-jet cross section
Eout < √ δ s √ Ein > (1 − ) s
Deﬁnition: an event is a two-jet event iﬀ ∃ two opposite cones with opening angle δ .
– Parma.

π 3 4
Observe:
• The total cross section is dominated by two-jet events at large q 2 (asymptotic freedom for jets . jets are more collimated. δ) = σtot − σ3j ( . δ → 0. f 3q 2
h i X 2 αs 2 Z 1 µ (1. δ) . • For increasing q2 the perturbative result remains reliable for narrower cones.
• It is easy to compute the dominant contributions as . Combining with the leptonic tensor one ﬁnds
" # αs π2 7 (1) (0) σ3j ( .
(1) • σ3j is easily computed from the real emission matrix element with appropriate cuts on phase space. as expected for spin 1/2 quarks.). • It is possible to compute (and verify experimentally) the angular distribution of two-jet events dσ2j /d cos θ ∝ 1 + cos2 θ . δ) = σtot CF 4 log(δ) log(2 ) + 3 log(δ) + − . In d = 4
P
2 2 qf 4πα . 05/09/2002 – 22
.
– Parma.R) −Hµ = 2Nc CF qf α q dz 3j π 2
f
×
Z 1−δ 2 (1−z 2 /2)
δ2
z(1 − y) 1−z dy + y(1 − yz)2 yz(1 − y)
»
–
...More precisely:
(0) (0) • At LO one ﬁnds simply σ2j ( . δ) = σtot = Nc (1) (1) (1)
• At NLO one ﬁnds only two.or three-jet events. so that
σ2j ( .

pj + pk . µ p →0 j lim E (p . .
lim Em+1 (p1 . . . . pm) be the observable. . . . . . = 2 de 2q m
and its moments (and in particular the average value) are
e
n
=
Z emax
emin
de e
n dσ
de
. . µ m+1 1 µ p →αp j k
– Parma. .
˛ Z Z ˛ (R) (1V ) σ(e)˛ “ dσm+1 + dσm + .) . . . The corresponding distribution is deﬁned by
Z dσ 1 X dLIPSm |Mm|2 δ (e − Em (p1 . . pj−1 . . .) = Em (p1 . Given a ﬁnal state with m partons. .
Note: These are “weighted” cross sections. . . . . . 05/09/2002 –
23
. pm)) . . . . . . At order αm−1 one must sum contributions with m + 1 partons s in the ﬁnal state. . pj+1 . ” = ˛ O αm+1
s
IR-C cancellation is preserved if the observable takes the same value for those conﬁgurations diﬀering only by the IR/C radiation. pj . .Event–shape variables
The mechanism underlying the cancellation of IR and C divergences suggests a further generalization: study distributions of observables constructed with ﬁnal state momenta so as to assign equal weights to events diﬀering by IR or C emissions. . and so on. . . . . let Em(p1. . . pj . with one virtual m real partons. .) = Em (p1 . . pk . . . . .) . .

• Great phenomenological relevance (for example: determination of αs . The deﬁnition can be expressed in terms of the eigenvalues of the space part of the energymomentum tensor of the ﬁnal state. 05/09/2002 – 24
. two-jet eventi have C = 0. • Thrust Pm
Clearly 0 < Tm ≤ 1.
1 = 2@ q
0
pi ∈H
X
pi A
12
Observe • Perturbative distributions are singular in the two-jet limit (logarithms of the form αn log2n−1 C ). but expectation values s are ﬁnite. C = 3(λ1λ2 + λ1 λ3 + λ2 λ3 ).Examples of event–shape variables
There is a variety of available IR/C safe event shapes.
i=1 Tm = maxn Pm ˆ
ˆ |pi · n| . hadronization corrections). |pi | i=1
• C parameter
Also in this case 0 ≤ Cm ≤ 1. • Jet algorithms and the related deﬁnitions of multi-jet events can be seen as particular event-shapes. Cm = 3 − 2 (pi · q) pj · q i. and Tm = 1 corresponds to two precisely collimated back to back particle beams.j=1
• Jet masses ρm
(H)
H is one of the two hemispheres identiﬁed by the thrust axis.
– Parma.
` ´2 m pi · pj 3 X ` ´ .

the ¯ asimptotic states of QCD are not quarks and gluons. which is ﬁnite. – Interpretation: it is not possible to produce only qq . q 2 ) # . σBorn is a good approximation. ¯ – Therefore σBorn is not a good approximation for σ = 0 (conﬁnement). as well as collinear divergences in the massless limit. – Interpretation: it is not possibile to produce only µ+µ−. ∆). consider for example the process e+ e− → µ+µ−. This is not a problem.
– Parma. On the other hand it is a good approximation for σtot(e+ e− → hadrons).
• In QED. the situation is almost ¯ analogous.
so that lim∆→0 σtot(∆) = 0. The true observable is σtot(∆) = P + − → µ+ µ− + nγ. – Therefore σBorn is not a good approximation for σ. – σ(e+e− → µ+µ− diverges starting at O(α3). • In QCD. – IR divergences in QED can be explicitly resummed
α σtot (∆) = σﬁn exp log π " ∆2 q2 ! f (m2 . There are similarities and diﬀerences. 05/09/2002 – 25
. considering e+e− → qq .A comparison with QED
QED also has IR divergences. asimptotic states of QED are not isolated fermions. and for σtot (∆). which n σ(e e is ﬁnite. – σ(e+e− → qq diverges starting at O(α2αs ).

• The vanishing of D0 on one integration contour is not suﬃcient to determine a singularity of the integral. dyi ˆ y1 k2 + y2 (p − k)2 + y3 (p + k)2 + i (2π)d 0 i=1 dd k Z 1Y 3
Let D0 be the denominator. 05/09/2002 – 26
P
.
Example: the scalar form factor
Introduce Feynman parameters y1. The only possible singularities of I0 must lie on surfaces where D0 = 0.
P1 O P2
– Parma. y3. Furthermore
• The integrand is a function of the complex variables kµ. I0 . • Only in two cases the singularity cannot be avoided by a deformation: either the contour is trapped between two poles. The contour can be deformed. yi. a pole migrates to the edge of an integration contour. y2. Let us begin with a simple example. or one has end-point singularities. Then
I0 = 2 Z δ(1 − y1 − y2 − y3 ) ˜.On singularities of Feynman diagrams
To study IR and C divergences to all orders one must characterize the generic singularity structure of Feynman diagrams. with an analytic structure determined by the i prescription.

D0 can be independent of yi on the surface D0 = 0. so that yi becomes useless for the deformation.
kµ = 0 . y2 = 0 . D0 however is quadratic in kµ.
µ where li is the momentum ﬂowing through the line with parameter yi. This is expressed by the Landau equations
y1 kµ − y2 (p − k)µ + y3 (p + k)µ = 0 yi = 0 or and
2 li = 0 . p.
“ ” ∂ µ D0 yi . k .• Singularities of I0 – dkµ integrals cannot have end-point singularities (I0 is UV convergent). y2 /y1 = y3 /y1 = 0
µ
IR C C
k = αp . 05/09/2002 – 27
. p = 0 . y3 = 0 . • Landau equations for I0 A necessary condition for a singularity of I0 is that all integration variables be trapped. since D0 is linear in yi . αy1 = (1 − α)y2
µ
µ
µ
k = −βp . so that two poles can trap the contour if they coalesce. ∂kµ
– dyi integrals can only have end-point singularities (in yi = 0). Alternatively. βy1 = (1 − β)y3
We recognize the expected IR and C singularities. • Solutions of Landau equations It’s easy to ﬁnd the expected solutions. Are there other solutions of Landau equations?
– Parma.

– Parma. vi =
µ 0 µ µ
li 1. for each on-shell line li . µ – Let ∆xµ ≡ yili . Then i
∆xi = ∆xi vi . 0 li
!
. – On–shell lines describe physically admissible processes for the classical propagation of massless particles.
• Interpretation: the solutions of Landau equations are given by reduced diagrams.• Coleman–Norton physical picture The search for solutions of Landau equations is simpliﬁed by the fact that they admit a simple physical representation. where – Oﬀ–shell lines are contracted to points. Observe that – if line i in a loop is oﬀ–shell it must be that yi = 0.
Interpretation: ∆xµ describes classical propagation of a i massless particle with momentum li . 05/09/2002 –
28
. – The Landau equations for I0 can now be written as
X
i
σ(i) ∆xi = 0 ∆xi = 0
µ
µ
on shell oﬀ shell .

. lˆ ` D yi . kl . kl . . D yi .
µ µ µ µ µ
– Parma. where all remaining loops can be interpreted as classically permitted processes. ∆xij ≡ xi − xj . ai = QN i “P ”PN a Γ(ai ) 0 i=1 D i=1 i N y D i=1 i=1 i i=1 i i N Y
an arbitrary Feynman diagram G(pr ) can be written as
G(pi ) = linee 0 Y Z 1 dyi δ @1 − 0 X i yi A 1 loops Y Z dd k
where the denominator D is a sum of propagators
“ ” X ` ´ 2 (p. + ∆xM 1 = 0 . • Landau equations
X i µ ηij ∆x = 0 i µ ∆x = 0 i on shell . ∀j/i ∈ j . M such that
∆x12 + . it must be possible to associate to all its vertices coordinates µ xk . kl . pr = yi li i linee
` ´ N yi . pr N
and the momenta li are linear functions of pr and kl . 05/09/2002 –
29
. . oﬀ shell .General case
• Feynman parametrization Thanks to the identity
“ “P ” ” P N a Z 1 N „ « δ 1 − N yi Γ Y 1 a −1 i=1 i i=1 dyi y i . pr ´˜ . . For any loop.
• Coleman–Norton picture Solutions are again reduced diagrams (with oﬀ–shell lines contracted to points). . k) − m2 + i . . k = 1.

” “p 2 µ In fact. Proof:
• Normal thresholds are solutions of Landau equations. If one produced particle has non-vanishing momentum the other ones must compensate moving in the opposite direction. m2 ) = q q
Theorem: the only singularities of the diagram (and thus of G(q 2 . An example with n = 4 is
q q
• No other reduced diagram satisﬁes Coleman–Norton. n = 1. which do not move. 2 . 05/09/2002 – 30
.
– Parma. and interact until they are absorbed. m2 )) are normal thresholds q 2 = n2 m2 . .Example: the two-point function
Consider an arbitrary 1PI diagram for the two-point function G(q 2 . The Coleman–Norton process is the creation of n > 1 particles at rest. .
G(q 2 . when q > 0 one can choose q = q 2 .. Once separated they cannot meet again in free motion. in a theory with only one species of particle with mass m2 . for an arbitrarily long time. m2 ). 0 .

• Construct the homogeneous integral for S . similar to those employed in the UV. k p. ˆ p p Example: for I0 .
• Given a diagram. k p. identify among the {ki } intrinsic coordinates (movement in S ) and normal coordinates (distance from S ). • Introduce a scaling variable determine the relative weight (integration volume)/singularity. the homogeneous integral is the eikonal one. Example: for I0. 6 One must develop power counting techniques. k− ∼ λ2 q2 . If phase space has high enough dimension singularities are suppressed (e.
– Parma. Set ni = λai ni and consider ˆ λ → 0. 05/09/2002 – 31
. yi }. µ • For every S . ni ﬁnite.
n ˆ ˆ2 i ˆ1 i
S
Example: for I0 . use the CN representation to identify a µ trapped surface S in the space {ki . to determine the strength of the singularities.: IR divergences in φ3). {k−. k+ is intrinsic.IR/C power counting
The Landau equations are only necessary conditions for the manifestation of IR/C divergences.g. k⊥} are normal. |k⊥| ∼ λ q2 . taking the dominant power of λ in every factor of the graph.

05/09/2002 –
32
.
ˆ ˜ d4xeiqx 0|T Jµ (x)Jν (0) |0
Unitarity gives
X ˜ ˆ 2 4 4 2 Im ρµν (q) = e 0|Jµ(0)|n n|Jν (0)|0 (2π) δ (q − pn ) . k) − n m2 → λAi f (ˆ ). then i
nS = X
i
ai −
X
i
Ai + nnum . If.
n
– Parma.
Application: ﬁniteness of Re+e−
The all-order ﬁniteness of σtot(e+e− → hadrons) follows from its relationship with the correlator of two elettromagnetic currents.
nS ≤ 0 signals a divergence. logarithmic when nS = 0.• The degree of divergence is given by the power of λ associated 2 with the homogeneous integral. for every line. Deﬁne then
ρµν (q) ≡ ie
2
” “ 2 2 = qµ qν − q gµν π(q ) . li (p.
P
C
2 Im pi Z C kj pi kj
generalizing T T † = −i(T − T † ). which in turn comes from unitarity.

– Parma. 0
”
S
H
These however are ﬁnite by power counting. 05/09/2002 – 33
. in d dimensions.
• In a frame in which q = one sees that there are no allowed classical processes with non-vanishing momenta such that the photon decays and then reconstitutes. Thus there are no trapped surfaces with non-vanishing momenta.
• To see it note that fermion lines at zero momentum are less singular than gluon lines. In that case. The worst case is then if S contains only gluons. q
It is then suﬃcient to prove the ﬁniteness of π(q2 ).σtot(e e
+ −
h i e2 2 → hadrons) = 2 Im π(q ) . with LS loops and gS gluons in S . This follows from the Coleman-Norton representation. and reduced diagram
µ
“p
q2. • The only trapped surfaces are those with all particles having vanishing momentum. as may be expected (all lines in H are oﬀ–shell).
nS = d LS − 2gS = 2(1 − )gS
which is positive in d > 2.

n · A = 0. The most general reduced diagram is
“ ” 2 Γν p 1 .Application: the quark form factor
Considering Γ(q2) one ﬁnds collinear divergences associated with observed quarks.
– Parma. −n n·k (n · k)2
so that the degree of IR/C singularity is reduced. plus a new soft propagator dominated by the new soft loop).
Contracting with kµ cancels the denominator
k Gµν (k) =
µ ax
kν nν 2 . • These considerations suggest a factorization Γ = J1 J2SH . 05/09/2002 – 34
. =
J H
S
J
Further simpliﬁcations are possible
• Gluons connecting S directly to H are suppressed (one more oﬀ–shell propagator. p 2 . ¯ • In an axial gauge. µ . • Fermion lines connecting diﬀerent subgraphs (except the necessary q e q) are suppressed. In fact the axial gauge gluon propagator is
ax Gµν (k) =
1 k2 + iε
„
nµ kν + nν kµ 2 kµ kν −gµν + −n n·k (n · k)2
«
. only the (anti)quark line connects the jet J to H .

in general non local). k⊥ . 05/09/2002 – 35
. recognizing only charge and direction. k2 (p1 − k)2 k · u2
The gluon-antiquark coupling simpliﬁes. con k+
µ
σ + − u(p1 )γσ (p1 − k)γµ (p2 + k)γ v(p2 ) → u(p1 )γ (p1 − k)γµ (p2 + k)γ v(p2 ) / / / / / / / /
• Ward identity: k = (p 2 + k − p 2 . Microexample: again the one-loop form factor. • Grammer-Yennie approximation in the numerator
→ → 1 u(p1 )γ + (p1 − k)γµ (p2 + k)k+ γ − v(p2 ) → / / / / + k 1 + u(p1 )γ (p1 − k)γµ (p2 + k) k v(p2 ) . k⊥ }.Diagrammatics of factorization
Identifying the leading integration regions in momentum space is only the ﬁrst step towards factorization. One must then • exploit simpliﬁcations in Feynman diagrams in the leading regions (soft/collinear approximations. it becomes eikonal. / / / / / k · u2 ˆ
µ p1 =
“
p+ . Feynman gauge.
– Parma. k = k . 0. One ﬁnds / / /) /
Γ(coll) ∝ µ Z dd k
u(p1 ) u 2 (p 1 − k γµ v(p2 ) / / /) . k . Ward identities).
• Kinematics: {k− . 0⊥ 1
”
” “ + − . collinear region k p1 . • organize all-order subtractions a tutti gli ordini so as to avoid double counting (diﬀerent factor functions have operator deﬁnitions.

• Typical result In an axial gauge the form factor factorizes
Γ q2 µ2 ! = J1 (p1 · n)2 µ 2 n2 ! J2 (p2 · n)2 µ 2 n2 ! S (ui · n) H q2 µ2 ! .
– Parma. One uses then eikonal identities like
1 1 1 1 1 1 = + .• Eikonal lines Graphically. one ﬁnds sistematic cancellations. a = iguα ta ij j i j u i
ij = u·k+iε iδ
in terms of these rules the previous calculation reads
coll
• More than one gluon Summing over all possible insertions of two or more collinear gluons. 05/09/2002 –
36
. one may introduce eikonal Feynman rules
α. due to Ward identities. k1 · u k2 · u (k1 + k2 ) · u k1 · u (k1 + k2 ) · u k2 · u
All collinear gluons couple to the same eikonal line.

• The transverse momentum distribution in Drell–Yan. and be left with“IR/C poles.
Most problems however have several hard scales. giving double log(q⊥/q2 ).
• The “Sudakov” form factor. . Double log(1 − q2/s) are resummed a ` la Sudakov. The two hard scales are Q2 = −q2 and W 2 = (p + q)2 = Q2 (1 − x)/x. and double log(1 − x). The two 2 2 scales are q2 and q⊥. the two hard scales ¯ 2 2 are s = (p1 + p2) and q . There are single log(1/x).
– Parma. • Drell–Yan. αs (µ ) µ2 ! “ ” 2 = σ 1. 05/09/2002 – 37
In a massless theory one may choose µ2 = q2 . resummed by the BFKL equation. αs (q ) . With masses one ﬁnds duble logarithms ” 2 2 2 log q /m .
.. They all arise from IR/C dynamics. If for example 2 2 q1 q2 ΛQCD . resummed ` la a Sudakov..Multi-scale problems and large logarithms
For observables depending on a single hard scale all logarithms are resummed using the renormalization group
σ q2 2 . with p ≤ n (single s logarithms) or p ≤ 2n (double logarithms). the reliability of perturbation theory is put 2 2 in jeopardy by terms like αn logp (q1 /q2 ).
Γ q2 µ2 !
αs =1− CF log2 4π
q2 µ2
!
+ . The process is qq → µ+µ−(q2).
• DIS. Some examples.

αs (Q2 ) m2 µ2 F ! . can already be seen from the renormalization group. αs (Q2 ) m ! e =C N. d log µF
The quark form factor
Consider the factorization
Γ q µ2
2
!
= J1
(p1 · n1 ) µ 2 n2 1
2
!
J2
(p2 · n2 ) µ 2 n2 2
2
!
S (ui · ni ) H
q . 2 . 05/09/2002 –
38
. g(µ)) . . ni µ2
2
!
. In terms of Mellin moments
e F2 Q2 N. g0 ) = Zi (Λ/µ. The Altarelli–Parisi equation similarly leds to the resummation of (single) logarithms of µF .
e dF2 =0 → dµF
“ ” e d log f 2 = γN αs (Q ) . 2 . αs (Q2 ) µF Q2 ! e f N. Λ. i=1
dG0
(n)
dµ
=0 →
d log GR d log µ
(n)
=−
Solving this equation resums in exponential form the logarithmic dependence on µ. Or: use eﬀective ﬁled theory (SCET).
n X i=1
γi (g(µ)) .Factorization and resummation
The deep connection between factorization and resummation. Ordinary renormalization group is not suﬃcient. µ.
– Parma.
n Y 1/2 (n) (n) G0 (pi .
Double logarithms are more diﬃcult. g(µ)) GR (pi . Gauge invariance plays a key role.

λ2 .
. . ) which implies Γ(q2 = 0. αs (µ2 ). independent of q2 . ∂p1 · n1 ∂ log(p1 · n1 ) ∂ log(p1 · n1 ) ∂ log(u1 · n1 )
The two functions on the r. Then
“ ” ∂ log J 2 + GJ = KJ αs (µ ). ∂ log q q2 . µ2 ! .
The K function contains all singularities in (H is ﬁnite as → 0). αs (µ2) becomes αs (µ2. ! Z 2 1 µ dλ2 + γ 2 ξ2 λ2 K α
G
−1. so that
∂ log Γ ∂ log J1 ∂ log H ∂ log S =0 → =− − . < 0) = 0. αs (µ + :2 0 ξ2 ! . Furthermore the form factor is renormalization group invariant. s
– Parma.s. 2 µ ξ2 µ2 ! 8 < 1 Z −Q2 dξ2 h “ ” 2) = exp K . 05/09/2002 – 39
!!39 = 5 . The function G contains all q2 dependence. α
. so that
dG dK =− = γK (αs (µ)) . the exponent has only single poles in of the form αn / n+1 . 2 s µ
The exponentiation is non trivial. α (µ2 ). then
Γ Q2 . Since Γ is divergent. The equation for Γ can be solved. The whole form factor obeys an equation of identical form. one needs to keep consistently the dependence on < 0 to all orders.h. αs (µ2 ). αs (µ2 ). have diﬀerent arguments. d log µ d log µ
with a ﬁnite anomalous dimension.The form factor is gauge invariant.

leading logarithms are in the jet functions J . which satisfy
ˆ ∂ log J ˆ = KJ ∂ log q q 2 . σ0 dT π 1−T π
Note: dσ/dT → 0 as T → 1 (“Sudakov suppression”). . αs (µ ) 2 Nµ
2
!
ˆ + GJ
q 2 . so they have a non-vanishing invariant mass m2 ∝ (1 − T )q2 as T → 1. These logs can be resummed with s similar methods. N µ2 n2 µ 2 ! ˆ ˆ ˆ J2 S H .
ˆ ˆ One can then solve for J(N ) in terms of J(1). The jets J now enter the ﬁnal state. as
ˆ J q2 N µ2 ! ˆ =J 2 3 ! „ Z 2 “ ” “ ”« q2 1 q µ dλ2 5. 05/09/2002 – 40
. As T → 1 the distribution can be factorized in a manner similar to Γ. exp4− log Γ ˆ αs (λ2 ) − Γ ˆ αs (λ2 ) 2 2 J 2 q 2 /N λ λ J µ
Leading logarithms are determined by ΓJ = γK + .The thrust distribution
The thrust distribution is singular as T → 1. behaving as αn log2n−1 (1 − T )/(1 − T ).
– Parma. J
Z 1 1 N dσ ˆ σ(N ) ≡ dT T = J1 σ0 0 dT q 2 (p1 · n)2 . Inverting the Mellin transform.
– » 1 dσ αs log(1 − T ) αs 2 = −2CF exp −CF log (1 − T ) .. . αs (µ ) 2 µ
2
!
. Neglecting ˆ ˆ running coupling eﬀects one easily ﬁnds J ∼ exp(αs log2 N ).
In an axial gauge.

one observes that • The perturbative expansion in powers of αs(q2 ) diverges.
• The integral is ambiguous. 1 + β0 αs (q 2 ) log k2 /q 2
αs (q 2 )
∞ ∞ X n n+1 2 Z ∞ X −az n n fa (q ) = (q ) β0 αs (q ) dze z = cn αs n! .
2 2 a n=0 0 n=1
` ´ . but the ambiguity is suppressed by powers of q2 .The borders of the perturbative regime
Resummations test the limits of the perturbative theory. In fact
fa (q ) = (q ) αs (q )
2 2 a 2
Z ∞
0
e−az dz . 1 − β0 αs (q 2 )z
The Landau pole on the integrazion contour induces an ambiguity of the same parametric size as the residue
» ˛ ˛ ˛ 2 ˛ ˛δfa (q )˛ ∝ exp −
– Parma. In fact
αs (k ) =
2
Letting z ≡ log q2 /k2. 05/09/2002 –
a β0 αs (q 2 )
–
=
Λ2 QCD q2
!a
. One gets in fact integrals of the form
fa (q ) =
2
Z q2
0
dk2 2 a (k ) αs (k2 ) . 2 k
One sees esplicitly the non–convergence of perturbation theory at high orders. consequence of the Landau pole.
41
.