Skip to content

Hard-Thermal-Loop Effective Theory

Hard-thermal-loop (HTL) theory is the gauge-invariant response of hard plasma particles to soft gauge fields. When external momenta satisfy KgTK\sim gT and internal thermal momenta satisfy PTP\sim T, an infinite family of one-loop amplitudes contributes at leading order. Their resummation produces collective poles, Landau damping, screened interactions, and the vertices required by non-Abelian Ward identities.

Required background. Scale hierarchies supplies the separation TgTT\gg gT, and thermal self-energies distinguishes pole, screening, and asymptotic masses. Helpful background. Soft and collinear singularities supplies the real-time infrared logic used in later kinetic applications.

Let vμ=(1,v)v^\mu=(1,\mathbf v) with v=1|\mathbf v|=1. A color perturbation Wa(x,v)W^a(x,\mathbf v) of an isotropic hard distribution obeys the collisionless gauge-covariant Vlasov equation

(v ⁣ ⁣D)abWb(x,v)=v ⁣ ⁣Ea(x).(v\!\cdot\!D)^{ab}W^b(x,\mathbf v)=\mathbf v\!\cdot\!\mathbf E^a(x).

The induced current is

jindμa(x)=mD2dΩv4πvμWa(x,v).j_{\rm ind}^{\mu a}(x) =m_D^2\int\frac{d\Omega_{\mathbf v}}{4\pi}\,v^\mu W^a(x,\mathbf v).

Together with DνFνμ=jindμ+jextμD_\nu F^{\nu\mu}=j_{\rm ind}^\mu+j_{\rm ext}^\mu, these local equations are equivalent to the nonlocal HTL effective action. Covariant current conservation follows directly:

Dμjindμ=mD2dΩv4π(v ⁣ ⁣D)W=mD2dΩv4πv ⁣ ⁣E=0.D_\mu j_{\rm ind}^\mu =m_D^2\int\frac{d\Omega_{\mathbf v}}{4\pi}\, (v\!\cdot\!D)W =m_D^2\int\frac{d\Omega_{\mathbf v}}{4\pi}\, \mathbf v\!\cdot\!\mathbf E=0.

The final equality uses angular isotropy. It is the kinetic origin of the Ward identities. Keeping an HTL-resummed propagator while dropping the corresponding HTL vertices generally violates this structure.

Linearization gives the retarded polarization tensor

ΠRμν(K)=mD2[δμ0δν0+ωdΩv4πvμvνωv ⁣ ⁣k+i0+],\Pi_R^{\mu\nu}(K) =m_D^2\left[ -\delta^{\mu0}\delta^{\nu0} +\omega\int\frac{d\Omega_{\mathbf v}}{4\pi} \frac{v^\mu v^\nu}{\omega-\mathbf v\!\cdot\!\mathbf k+i0^+} \right],

up to the metric/index convention implicit in the displayed components. Direct contraction verifies KμΠRμν=0K_\mu\Pi_R^{\mu\nu}=0. The i0+i0^+ is essential: it fixes retarded response and generates an imaginary part when ω<k|\omega|<k.

Rotational invariance decomposes the spatial tensor into projectors longitudinal and transverse to k\mathbf k. With x=ω/kx=\omega/k,

ΠL(ω,k)=mD2[1x2ln ⁣(x+1+i0+x1+i0+)],\Pi_L(\omega,k) =m_D^2\left[1-\frac{x}{2} \ln\!\left(\frac{x+1+i0^+}{x-1+i0^+}\right)\right],

and

ΠT(ω,k)=mD22[x2+x(1x2)2ln ⁣(x+1+i0+x1+i0+)].\Pi_T(\omega,k) =\frac{m_D^2}{2}\left[ x^2+\frac{x(1-x^2)}{2} \ln\!\left(\frac{x+1+i0^+}{x-1+i0^+}\right) \right].

With the scalar convention displayed here, the physical resummed sectors are

DLR(ω,k)=1k2+ΠL(ω,k),DTR(ω,k)=1ω2k2ΠT(ω,k).D_L^R(\omega,k)=\frac{1}{k^2+\Pi_L(\omega,k)},\qquad D_T^R(\omega,k)=\frac{-1}{\omega^2-k^2-\Pi_T(\omega,k)}.

Gauge fixing supplies an additional longitudinal four-dimensional component; it is not a new plasma mode. Poles and cuts of the two denominators above give the collective response.

Conventions for the scalar functions vary; physical denominators must be checked rather than formulas compared by name. In the static limit ΠL(0,k)=mD2\Pi_L(0,k)=m_D^2 while ΠT(0,k)=0\Pi_T(0,k)=0 at leading perturbative order. For ω<k|\omega|<k, the logarithm crosses a branch cut and describes collisionless Landau damping. For timelike kinematics, zeros of the resummed inverse propagator yield longitudinal and transverse plasmons, developed on the collective-modes page.

Gauge invariance ties the nn-point HTL vertices together. Formally, the gluonic action can be written

LHTL=12mD2dΩv4πTr ⁣[Fμαvαvβ(v ⁣ ⁣D)2Fμβ].\mathcal L_{\rm HTL} =-\frac12m_D^2 \int\frac{d\Omega_{\mathbf v}}{4\pi}\, \operatorname{Tr}\!\left[ F_{\mu\alpha}\frac{v^\alpha v^\beta}{(v\!\cdot\!D)^2} F^{\mu}{}_{\beta}\right].

Expanding (v ⁣ ⁣D)1(v\!\cdot\!D)^{-1} generates vertices of all orders. For soft fields AgTA\sim gT, the covariant derivative term gAgA competes with gT\partial\sim gT, so the entire series can matter. Braaten and Pisarski showed that HTL-resummed perturbation theory restores a systematic leading-order expansion for soft amplitudes Braaten and Pisarski 1990, §§ 2–4; Frenkel and Taylor established the gauge-invariant generating functional Frenkel and Taylor 1990.

For fermions, an analogous angular integral produces the thermal self-energy with asymptotic mass parameter m2g2CRT2m_\infty^2\sim g^2C_RT^2. In kinetic applications, asymptotic masses regulate collinear energy denominators, whereas mDm_D screens soft longitudinal exchange. Treating them as a single “thermal mass” loses essential physics.

Equilibrium HTL assumes an isotropic, homogeneous hard distribution and soft external momentum. It is collisionless at the scale ω,kgT\omega,k\sim gT. Widths that are parametrically smaller, ultrasoft color relaxation, and LPM interference require collision kernels or further matching. An anisotropic distribution admits a hard-loop generalization but may be unstable; it cannot be obtained by substituting a direction-dependent mDm_D into the isotropic formulas.

A trustworthy HTL calculation should pass:

  • transversality of every self-energy and the associated vertex Ward identities;
  • the static limits ΠL(0,k)=mD2\Pi_L(0,k)=m_D^2 and ΠT(0,k)=0\Pi_T(0,k)=0 at leading order;
  • retarded analyticity in the upper half ω\omega plane and the correct sign of spectral weight;
  • cancellation of the separation scale when hard and soft momentum regions are combined;
  • independence of the final gauge-invariant observable from gauge-fixing parameters.

These tests and the scope of the approximation are recorded in the hot-gauge plasma validity table.

The branching structure shows both the reach of HTL matching and the observables it does not contain.

Hard thermal particles induce a gauge-covariant HTL response for soft real-time fields; static EQCD and MQCD, ultrasoft Bödeker dynamics, and hard-quasiparticle kinetic theory are separate matched descriptions rather than limits of one HTL propagator.

HTL amplitudes are fixed by hard-particle response and satisfy non-Abelian Ward identities, making them the appropriate branch for soft collisionless real-time correlators. The static, magnetostatic, ultrasoft, and quasiparticle branches require additional reductions or collision kernels and cannot be substituted without matching. The diagram is schematic and not to scale.

The text equivalent is to identify the external frequency and momentum before resumming: use HTL for soft real-time response, avoid double counting hard contributions already encoded in its vertices, and change effective theory when static, magnetic, collisional, or ultrasoft physics dominates.

1. Ward identity. Contract the linearized polarization tensor with KμK_\mu, using K ⁣ ⁣v=ωk ⁣ ⁣vK\!\cdot\!v=\omega-\mathbf k\!\cdot\!\mathbf v.

Solution

The integral term becomes mD2ωdΩvν/(4π)m_D^2\omega\int d\Omega\,v^\nu/(4\pi) after cancellation of the denominator. Its only nonzero angular average is the ν=0\nu=0 component, equal to mD2ωδν0m_D^2\omega\delta^{\nu0}, which cancels the contracted local term. Hence KμΠRμν=0K_\mu\Pi_R^{\mu\nu}=0.

2. Landau cut. For real x<1|x|<1, determine the imaginary part of the logarithm in ΠL\Pi_L with the retarded prescription.

Solution

The denominator x1+i0+x-1+i0^+ lies just above the negative real axis, while x+1>0x+1>0. Therefore ln[(x+1+i0+)/(x1+i0+)]=ln(1+x)/(1x)iπ\ln[(x+1+i0^+)/(x-1+i0^+)]=\ln|(1+x)/(1-x)|-i\pi. Because ΠL\Pi_L contains minus one half of this logarithm, the displayed convention gives ImΠL=+πmD2x/2\operatorname{Im}\Pi_L=+\pi m_D^2x/2. Damping is determined from the full retarded propagator and its spectral convention, not from this scalar sign in isolation.

Continue to collective modes and dynamical screening or to effective kinetic theory.

  • Braaten, Eric, and Robert D. Pisarski. “Soft Amplitudes in Hot Gauge Theories: A General Analysis.” Nuclear Physics B 337, no. 3 (1990): 569–634. DOI.
  • Frenkel, Jean, and John C. Taylor. “High Temperature Limit of Thermal QCD.” Nuclear Physics B 334, no. 1 (1990): 199–216. DOI.
  • Le Bellac, Michel. Thermal Field Theory. Cambridge Monographs on Mathematical Physics. Cambridge: Cambridge University Press, 1996. DOI.
  • Taylor, John C., and S. M. H. Wong. “The Effective Action of Hard Thermal Loops in QCD.” Nuclear Physics B 346, no. 1 (1990): 115–128. DOI.