Skip to content

Partition Functions

A partition function is a basis-independent trace of an equilibrium statistical weight. In the fixed-particle-number canonical ensemble,

ZN(β)=Tr⁡HNe−βHN.Z_N(\beta) = \operatorname{Tr}_{\mathcal H_N} e^{-\beta H_N}.

In the grand-canonical ensemble,

Ξ(β,μ)=Tr⁡Fe−β(H−μN).\Xi(\beta,\mu) = \operatorname{Tr}_{\mathcal F} e^{-\beta(H-\mu N)}.

These quantities normalize density operators, but normalization is only their first role. Their logarithms generate connected fluctuations, thermodynamic potentials, and responses. Their factorization records independence. Their analytic and asymptotic structure encodes spectra, phases, and low-energy physics.

This page is the canonical home for partition functions as mathematical and physical generating objects. The Canonical Ensemble and Grand-Canonical Ensemble pages own the thermodynamics of their respective control variables.

Let {∣n⟩}\{|n\rangle\} be any orthonormal basis of the relevant Hilbert space. Then

Z=∑n⟨n∣e−βH∣n⟩.Z = \sum_n \langle n| e^{-\beta H} |n\rangle.

If UU is unitary and

∣n′⟩=U∣n⟩,|n'\rangle = U|n\rangle,

then cyclicity of the trace gives

∑n⟨n′∣e−βH∣n′⟩=Tr⁡(U†e−βHU)=Tr⁡e−βH.\begin{aligned} \sum_n \langle n'| e^{-\beta H} |n'\rangle &= \operatorname{Tr} \left( U^\dagger e^{-\beta H}U \right) \\ &= \operatorname{Tr} e^{-\beta H}. \end{aligned}

The partition function is therefore independent of the basis used to evaluate it. The energy basis is convenient, not fundamental.

For

H=∑aEaPa,H = \sum_a E_aP_a,

functional calculus gives

e−βH=∑ae−βEaPa.e^{-\beta H} = \sum_a e^{-\beta E_a}P_a.

Taking the trace,

Z(β)=∑agae−βEa,Z(\beta) = \sum_a g_a e^{-\beta E_a},

where

ga=Tr⁡Pag_a = \operatorname{Tr}P_a

is the degeneracy of energy EaE_a.

Dropping gag_a is equivalent to discarding states. If energy labels are repeated once per independent eigenvector, the same formula can be written

Z(β)=∑ne−βEn.Z(\beta) = \sum_n e^{-\beta E_n}.

The canonical density operator is

ρβ=e−βHZ(β).\rho_\beta = \frac{ e^{-\beta H} }{ Z(\beta) }.

Its unit trace follows immediately:

Tr⁡ρβ=Tr⁡e−βHZ=1.\operatorname{Tr}\rho_\beta = \frac{ \operatorname{Tr}e^{-\beta H} }{ Z } = 1.

The same logic gives

ρβ,μ=e−β(H−μN)Ξ\rho_{\beta,\mu} = \frac{ e^{-\beta(H-\mu N)} }{ \Xi }

in the grand-canonical ensemble.

The exponential must have a dimensionless argument:

βH\beta H

is dimensionless because

[β]=energy−1.[\beta] = \text{energy}^{-1}.

The trace of e−βHe^{-\beta H} is dimensionless, so ZZ and Ξ\Xi are dimensionless.

A classical phase-space partition function requires a reference cell such as hdNh^{dN} to make the integral dimensionless. In quantum mechanics, the Hilbert-space trace supplies the state count directly.

In finite dimension,

0<Z(β)<∞0<Z(\beta)<\infty

for every finite real β\beta.

In infinite dimension, the operator

e−βHe^{-\beta H}

must be trace class. Equivalently,

∑ne−βEn<∞\sum_n e^{-\beta E_n} < \infty

for a discrete spectrum counted with multiplicity.

A Hamiltonian bounded below is not by itself sufficient. The number of states below energy EE must not grow too rapidly relative to the Boltzmann suppression.

For β>0\beta>0, low energies are favored. Confined systems with spectra growing sufficiently rapidly often have finite ZZ.

At β=0\beta=0,

Z(0)=Tr⁡I=dim⁡H.Z(0) = \operatorname{Tr}\mathbb I = \dim\mathcal H.

This is finite only in finite dimension.

For β<0\beta<0, high energies are favored. A finite partition function then generally requires a spectrum bounded above or another explicit cutoff.

Even when a finite-volume partition function exists, the total ZVZ_V usually grows exponentially with volume. The bulk object is often

lim⁡V→∞1Vln⁡ZV,\lim_{V\to\infty} \frac{1}{V} \ln Z_V,

or the associated free-energy density, not a finite infinite-volume trace.

Shift the Hamiltonian by a constant:

H′=H+CI.H' = H+C\mathbb I.

Then

Z′=Tr⁡e−β(H+C)=e−βCZ.Z' = \operatorname{Tr} e^{-\beta(H+C)} = e^{-\beta C}Z.

The normalized density operator is unchanged:

e−βH′Z′=e−βHZ.\frac{ e^{-\beta H'} }{ Z' } = \frac{ e^{-\beta H} }{ Z }.

The Helmholtz free energy shifts by

F′=F+C,F' = F+C,

as expected when every energy is shifted by CC. Probabilities and energy differences are invariant.

Assume HH has no explicit β\beta dependence. Differentiating gives

∂Z∂β=−Tr⁡(He−βH).\frac{\partial Z}{\partial\beta} = -\operatorname{Tr} \left( He^{-\beta H} \right).

Therefore

−∂ln⁡Z∂β=⟨H⟩β.-\frac{\partial\ln Z}{\partial\beta} = \langle H\rangle_\beta.

A second derivative gives

∂2ln⁡Z∂β2=⟨H2⟩β−⟨H⟩β2.\frac{\partial^2\ln Z}{\partial\beta^2} = \langle H^2\rangle_\beta -\langle H\rangle_\beta^2.

Thus

∂2ln⁡Z∂β2=Var⁡β(H)≥0.\frac{\partial^2\ln Z}{\partial\beta^2} = \operatorname{Var}_\beta(H) \geq 0.

The convexity of ln⁡Z\ln Z in β\beta is a variance statement.

Within the canonical state at inverse temperature β\beta, consider

⟨etH⟩β.\left\langle e^{tH} \right\rangle_\beta.

Because HH commutes with every function of itself,

⟨etH⟩β=Tr⁡(etHe−βH)Z(β)=Z(β−t)Z(β),\begin{aligned} \left\langle e^{tH} \right\rangle_\beta &= \frac{ \operatorname{Tr} \left( e^{tH}e^{-\beta H} \right) }{ Z(\beta) } \\ &= \frac{ Z(\beta-t) }{ Z(\beta) }, \end{aligned}

whenever both traces converge.

The cumulant-generating function of energy is

KH(t)=ln⁡Z(β−t)−ln⁡Z(β).K_H(t) = \ln Z(\beta-t) -\ln Z(\beta).

Consequently,

κn(H)=(−1)n∂nln⁡Z∂βn.\kappa_n(H) = (-1)^n \frac{ \partial^n\ln Z }{ \partial\beta^n }.

For example,

κ1(H)=⟨H⟩,\kappa_1(H) = \langle H\rangle,

and

κ2(H)=Var⁡(H).\kappa_2(H) = \operatorname{Var}(H).

Higher derivatives generate higher connected energy fluctuations.

Introduce a source JJ coupled to an observable AA:

H(J)=H−JA.H(J) = H-JA.

Define

Z(β,J)=Tr⁡exp⁡[−β(H−JA)].Z(\beta,J) = \operatorname{Tr} \exp \left[ -\beta(H-JA) \right].

The first derivative is

∂ln⁡Z∂J=β⟨A⟩J.\frac{\partial\ln Z}{\partial J} = \beta \langle A\rangle_J.

This remains true even if

[H,A]≠0.[H,A]\neq0.

The derivative of a noncommuting operator exponential is handled by the Duhamel formula, and cyclicity of the trace reduces the first derivative to the thermal expectation value.

The second derivative requires more care. For noncommuting AA and H(J)H(J), it produces an imaginary-time or Kubo–Mori correlation rather than simply

β2Var⁡(A).\beta^2\operatorname{Var}(A).

The ordinary variance is recovered when the relevant operators commute. Fluctuations and Susceptibilities develops the Kubo–Mori covariance, static response interpretation, and thermodynamic examples.

Parameter Derivatives and Generalized Forces

Section titled “Parameter Derivatives and Generalized Forces”

Let

H=H(λ).H = H(\lambda).

Then

∂ln⁡Z∂λ=−β⟨∂H∂λ⟩.\frac{\partial\ln Z}{\partial\lambda} = -\beta \left\langle \frac{\partial H}{\partial\lambda} \right\rangle.

The Helmholtz free energy

F=−1βln⁡ZF = -\frac{1}{\beta}\ln Z

satisfies

∂F∂λ=⟨∂H∂λ⟩.\frac{\partial F}{\partial\lambda} = \left\langle \frac{\partial H}{\partial\lambda} \right\rangle.

If the generalized force is defined by

Xλ≡−∂H∂λ,X_\lambda \equiv -\frac{\partial H}{\partial\lambda},

then

⟨Xλ⟩=−∂F∂λ.\langle X_\lambda\rangle = -\frac{\partial F}{\partial\lambda}.

Explicit β\beta or temperature dependence in an effective Hamiltonian adds derivative terms and must be stated.

Grand Partition Function as a Number Generator

Section titled “Grand Partition Function as a Number Generator”

Assume

[H,N]=0.[H,N]=0.

Fock space decomposes as

F=⨁N=0∞HN.\mathcal F = \bigoplus_{N=0}^{\infty} \mathcal H_N.

The grand partition function is

Ξ(β,μ)=∑N=0∞eβμNZN(β).\Xi(\beta,\mu) = \sum_{N=0}^{\infty} e^{\beta\mu N} Z_N(\beta).

With fugacity

z≡eβμ,z \equiv e^{\beta\mu},

this becomes

Ξ(β,z)=∑N=0∞zNZN(β).\Xi(\beta,z) = \sum_{N=0}^{\infty} z^NZ_N(\beta).

Thus Ξ\Xi is an ordinary generating function for the fixed-NN canonical partition functions.

The probability of particle number NN is

Pr⁡(N)=zNZNΞ.\Pr(N) = \frac{ z^NZ_N }{ \Xi }.

If Ξ(z)\Xi(z) is analytic around the origin,

ZN=[zN]Ξ(z),Z_N = [z^N]\Xi(z),

where [zN][z^N] denotes the coefficient of zNz^N.

Equivalently,

ZN=12πi∮CΞ(z)zN+1 dz,Z_N = \frac{1}{2\pi i} \oint_{\mathcal C} \frac{ \Xi(z) }{ z^{N+1} } \,dz,

where C\mathcal C encloses the origin inside the domain of analyticity.

This contour formula is exact. Saddle-point evaluation at large NN connects fixed-number and fixed-chemical-potential descriptions.

It is convenient to define

α≡βμ=ln⁡z.\alpha \equiv \beta\mu = \ln z.

Then

∂ln⁡Ξ∂α=⟨N⟩,\frac{\partial\ln\Xi}{\partial\alpha} = \langle N\rangle,

and

∂2ln⁡Ξ∂α2=Var⁡(N).\frac{\partial^2\ln\Xi}{\partial\alpha^2} = \operatorname{Var}(N).

More generally,

κn(N)=∂nln⁡Ξ∂αn.\kappa_n(N) = \frac{ \partial^n\ln\Xi }{ \partial\alpha^n }.

Keeping α\alpha, zz, or μ\mu fixed while differentiating with respect to β\beta are different operations. The variable held fixed must be stated.

Suppose

H=HA⊗HB,\mathcal H = \mathcal H_A \otimes \mathcal H_B,

and

H=HA⊗IB+IA⊗HB.H = H_A\otimes\mathbb I_B +\mathbb I_A\otimes H_B.

The two terms commute, so

e−βH=e−βHA⊗e−βHB.e^{-\beta H} = e^{-\beta H_A} \otimes e^{-\beta H_B}.

Using

Tr⁡A⊗B(XA⊗YB)=Tr⁡AXATr⁡BYB,\operatorname{Tr}_{A\otimes B} (X_A\otimes Y_B) = \operatorname{Tr}_A X_A \operatorname{Tr}_B Y_B,

one obtains

ZAB=ZAZB.Z_{AB} = Z_AZ_B.

Therefore

ln⁡ZAB=ln⁡ZA+ln⁡ZB.\ln Z_{AB} = \ln Z_A+\ln Z_B.

The logarithm is additive because it generates connected, extensive thermodynamic quantities.

Naive factorization fails when:

  • the Hamiltonian contains interactions between AA and BB;
  • the Hilbert space does not factor into independent tensor factors;
  • a global constraint couples otherwise independent subsystems;
  • identical-particle symmetrization links particle labels;
  • gauge or superselection constraints restrict the physical subspace;
  • the trace is projected onto one total-charge sector.

Weak interactions can sometimes be treated perturbatively, but

ZAB≠ZAZBZ_{AB} \neq Z_AZ_B

is the generic interacting case.

Single-Particle Versus Many-Body Partition Functions

Section titled “Single-Particle Versus Many-Body Partition Functions”

Let the one-particle Hamiltonian hh act on a one-particle space h\mathfrak h, with energies ϵi\epsilon_i. Define

q(β)≡Tr⁡he−βh=∑ie−βϵi.q(\beta) \equiv \operatorname{Tr}_{\mathfrak h} e^{-\beta h} = \sum_i e^{-\beta\epsilon_i}.

The symbol qq is a one-particle partition function. It is not automatically the partition function of NN identical quantum particles.

Routes from a one-particle partition function to distinguishable, identical-particle, and grand-canonical many-body partition functions

The same one-particle spectrum leads to different many-body partition functions depending on the Hilbert space and number constraint. Exchange symmetry and the sum over particle-number sectors are structural choices, not optional correction factors.

For NN labeled, noninteracting particles,

Hdist=h⊗N,\mathcal H_{\mathrm{dist}} = \mathfrak h^{\otimes N},

and

HN=∑a=1Nha.H_N = \sum_{a=1}^{N} h_a.

The partition function factorizes:

ZNdist=q(β)N.Z_N^{\mathrm{dist}} = q(\beta)^N.

This formula counts permutations of one-particle occupations as distinct labeled states.

For identical particles, the trace is restricted to the symmetric or antisymmetric subspace:

ZN(±)=Tr⁡HN(±)e−βHN.Z_N^{(\pm)} = \operatorname{Tr}_{\mathcal H_N^{(\pm)}} e^{-\beta H_N}.

The plus sign denotes bosons and the minus sign fermions. In general,

ZN(±)≠qN.Z_N^{(\pm)} \neq q^N.

For noninteracting identical particles, an exact recurrence is

ZN(η)=1N∑ℓ=1Nηℓ−1q(ℓβ)ZN−ℓ(η),Z_N^{(\eta)} = \frac{1}{N} \sum_{\ell=1}^{N} \eta^{\ell-1} q(\ell\beta) Z_{N-\ell}^{(\eta)},

with

Z0(η)=1,η={+1,bosons,−1,fermions.Z_0^{(\eta)}=1, \qquad \eta= \begin{cases} +1,&\text{bosons},\\ -1,&\text{fermions}. \end{cases}

The terms with ℓ>1\ell>1 are exchange corrections.

In the dilute, nondegenerate regime, exchange corrections become small and the identical-particle canonical partition function approaches

ZNMB≃q(β)NN!.Z_N^{\mathrm{MB}} \simeq \frac{ q(\beta)^N }{ N! }.

The factor 1/N!1/N! prevents classical overcounting of particle labels. This is an approximation to the quantum bosonic or fermionic result, not an exact formula at arbitrary density.

Classical Limit of Quantum Statistics derives the controlling parameter nλTd/gn\lambda_T^d/g, the low-fugacity cycle expansion, and the leading Bose and Fermi virial corrections.

Suppose

H−μN=∑i(ϵi−μ)ni.H-\mu N = \sum_i (\epsilon_i-\mu)n_i.

The grand partition function factorizes over modes.

For bosons,

ΞB=∏i∑ni=0∞(ze−βϵi)ni=∏i11−ze−βϵi,\begin{aligned} \Xi_{\mathrm B} &= \prod_i \sum_{n_i=0}^{\infty} \left( ze^{-\beta\epsilon_i} \right)^{n_i} \\ &= \prod_i \frac{1}{ 1-ze^{-\beta\epsilon_i} }, \end{aligned}

provided

∣ze−βϵi∣<1\left| ze^{-\beta\epsilon_i} \right| < 1

for every mode.

For fermions,

ΞF=∏i∑ni=01(ze−βϵi)ni=∏i(1+ze−βϵi).\begin{aligned} \Xi_{\mathrm F} &= \prod_i \sum_{n_i=0}^{1} \left( ze^{-\beta\epsilon_i} \right)^{n_i} \\ &= \prod_i \left( 1+ze^{-\beta\epsilon_i} \right). \end{aligned}

Taking logarithms turns products into sums:

ln⁡ΞB=−∑iln⁡(1−ze−βϵi),\ln\Xi_{\mathrm B} = -\sum_i \ln \left( 1-ze^{-\beta\epsilon_i} \right),

and

ln⁡ΞF=∑iln⁡(1+ze−βϵi).\ln\Xi_{\mathrm F} = \sum_i \ln \left( 1+ze^{-\beta\epsilon_i} \right).

Let

ϵ0=0,ϵ1=Δ,\epsilon_0=0, \qquad \epsilon_1=\Delta,

and define

x≡e−βΔ.x \equiv e^{-\beta\Delta}.

The one-particle partition function is

q=1+x.q = 1+x.

For two labeled particles,

Z2dist=q2=1+2x+x2.Z_2^{\mathrm{dist}} = q^2 = 1+2x+x^2.

The coefficient 2x2x counts which labeled particle occupies the excited level.

The allowed occupation states are

(n0,n1)=(2,0),(1,1),(0,2).(n_0,n_1) = (2,0),(1,1),(0,2).

Therefore

Z2B=1+x+x2.Z_2^{\mathrm B} = 1+x+x^2.

Pauli exclusion allows only

(n0,n1)=(1,1).(n_0,n_1) = (1,1).

Hence

Z2F=x.Z_2^{\mathrm F} = x.

The three answers differ despite sharing the same one-particle spectrum.

Define

ω(E)=Tr⁡δ(E−H).\omega(E) = \operatorname{Tr} \delta(E-H).

Then

Z(β)=∫dE ω(E)e−βE.Z(\beta) = \int dE\, \omega(E)e^{-\beta E}.

The partition function is the Laplace transform of the density of states.

For a discrete spectrum,

ω(E)=∑agaδ(E−Ea),\omega(E) = \sum_a g_a\delta(E-E_a),

and the integral reduces to the spectral sum.

Formally, the inverse transform is

ω(E)=12πi∫c−i∞c+i∞dβ eβEZ(β),\omega(E) = \frac{1}{2\pi i} \int_{c-i\infty}^{c+i\infty} d\beta\, e^{\beta E}Z(\beta),

where the vertical contour lies in a domain of convergence. Inverse Laplace transforms are often numerically ill-conditioned; formal invertibility does not guarantee stable reconstruction from noisy data.

If

ω(E)∼exp⁡[S(E)kB],\omega(E) \sim \exp \left[ \frac{S(E)}{k_{\mathrm B}} \right],

then

Z(β)∼∫dE exp⁡[S(E)kB−βE].Z(\beta) \sim \int dE\, \exp \left[ \frac{S(E)}{k_{\mathrm B}} -\beta E \right].

In a regular thermodynamic limit, the dominant energy satisfies

1kB∂S∂E=β.\frac{1}{k_{\mathrm B}} \frac{\partial S}{\partial E} = \beta.

Thus the partition function packages microcanonical state counting into a fixed-temperature generating object. The Microcanonical Ensemble page owns shell definitions and entropy conventions.

In finite dimension,

e−βH=I−βH+β22H2+O(β3).e^{-\beta H} = \mathbb I -\beta H +\frac{\beta^2}{2}H^2 +O(\beta^3).

Therefore

Z(β)=d−βTr⁡H+β22Tr⁡H2+O(β3),Z(\beta) = d -\beta\operatorname{Tr}H +\frac{\beta^2}{2} \operatorname{Tr}H^2 +O(\beta^3),

where

d=dim⁡H.d = \dim\mathcal H.

As β→0\beta\to0,

ρβ⟶Id.\rho_\beta \longrightarrow \frac{\mathbb I}{d}.

In infinite dimension, this expansion need not define a trace because Z(0)Z(0) may diverge.

Let the ground energy be E0E_0 with degeneracy g0g_0, and let the next distinct energy be

E1=E0+Δ.E_1 = E_0+\Delta.

Then

Z(β)=e−βE0[g0+g1e−βΔ+⋯ ].Z(\beta) = e^{-\beta E_0} \left[ g_0 +g_1e^{-\beta\Delta} +\cdots \right].

At large positive β\beta,

ln⁡Z=−βE0+ln⁡g0+O(e−βΔ).\ln Z = -\beta E_0 +\ln g_0 +O \left( e^{-\beta\Delta} \right).

The leading exponential reveals the ground energy, while the constant term reveals its degeneracy.

For a finite spectrum,

Z(β)=∑agae−βEaZ(\beta) = \sum_a g_ae^{-\beta E_a}

is analytic for finite real β\beta and strictly positive there. Consequently,

ln⁡Z(β)\ln Z(\beta)

has no real-axis singularity at finite system size.

Sharp thermodynamic phase transitions require a limit in which system size diverges. In the complex plane, finite-system partition functions can have zeros. Families of such zeros can approach the physical real axis in a thermodynamic limit, producing nonanalytic bulk free energies.

The trace

Z=Tr⁡e−βHZ = \operatorname{Tr} e^{-\beta H}

is an evolution trace over imaginary time of length

βℏ.\beta\hbar.

For

H=p22m+V(q),H = \frac{p^2}{2m} +V(q),

the Trotter product divides the interval into MM slices. In the limit,

Z=∫q(βℏ)=q(0)Dq e−SE[q]/ℏ,Z = \int_{q(\beta\hbar)=q(0)} \mathcal Dq\, e^{-S_{\mathrm E}[q]/\hbar},

with Euclidean action

SE[q]=∫0βℏdτ[m2(dqdτ)2+V(q)].S_{\mathrm E}[q] = \int_0^{\beta\hbar} d\tau \left[ \frac{m}{2} \left( \frac{dq}{d\tau} \right)^2 +V(q) \right].

The trace imposes periodic endpoint identification on coordinate paths.

This formula is a preview. Finite-Temperature QM Overview places the thermal trace in the wider KMS, Matsubara, spectral, and path-integral toolkit. Path Integrals for Statistical Mechanics develops the cyclic measure, normalization, exchange sectors, ring-polymer representation, and thermal oscillator determinant. Euclidean and Imaginary-Time Path Integrals retains the open-kernel and Wick-rotation construction.

For NN identical particles, the trace over symmetric or antisymmetric states can be represented schematically as

ZN(±)=1N!∑P∈SN(±1)P∫dq ⟨q∣e−βH∣Pq⟩.Z_N^{(\pm)} = \frac{1}{N!} \sum_{P\in S_N} (\pm1)^P \int dq\, \langle q| e^{-\beta H} |Pq\rangle.

Bosons sum permutations with positive sign. Fermions include permutation parity.

In coherent-state thermal path integrals, bosonic fields are periodic and fermionic Grassmann fields are antiperiodic in imaginary time. Coherent-State Path Integrals Preview derives these conditions from the trace and checks the corresponding Gaussian determinants. The distinction from first-quantized permutation closure is made explicit in Path Integrals for Statistical Mechanics.

Directly summing

Z=∑agae−βEaZ = \sum_a g_ae^{-\beta E_a}

can underflow at low temperature. Let Emin⁡E_{\min} be the smallest retained energy. Then

ln⁡Z=−βEmin⁡+ln⁡∑agae−β(Ea−Emin⁡).\ln Z = -\beta E_{\min} +\ln \sum_a g_a e^{-\beta(E_a-E_{\min})}.

Every exponential in the remaining sum is at most one for β>0\beta>0.

For independent modes or large lattices, compute ln⁡Z\ln Z or ln⁡Ξ\ln\Xi directly rather than forming exponentially large products.

For example,

ln⁡ΞF=∑iln⁡(1+ze−βϵi)\ln\Xi_{\mathrm F} = \sum_i \ln \left( 1+ze^{-\beta\epsilon_i} \right)

is more stable than multiplying all factors.

If a spectral sum is truncated, vary the energy cutoff until ln⁡Z\ln Z and target observables are stable. A cutoff adequate at low temperature may fail as temperature increases.

Compute traces in the intended Hilbert space. A full-space trace can differ exponentially from a fixed-charge or fixed-symmetry trace.

Check:

β→0,β→+∞,\beta\to0, \qquad \beta\to+\infty,

known noninteracting limits, and exact small-system results.

This page owns:

  • basis-independent trace and spectral-sum definitions;
  • convergence and trace-class conditions;
  • moment, cumulant, source, and number-generating interpretations;
  • exact factorization criteria;
  • the distinction among one-particle, distinguishable-particle, bosonic, fermionic, canonical, and grand partition functions;
  • density-of-states transforms;
  • high- and low-temperature asymptotics;
  • imaginary-time trace and path-integral preview;
  • numerical stabilization.

Other pages own:

  • fixed-T,N,VT,N,V free energies, energy fluctuations, and reservoir interpretation: Canonical Ensemble;
  • fixed-T,μ,VT,\mu,V thermodynamics, convergence in particle-number sectors, and compressibility: Grand-Canonical Ensemble;
  • equilibrium static response, ensemble dependence, and Kubo–Mori covariance: Fluctuations and Susceptibilities;
  • the regime and accuracy of the Maxwell–Boltzmann approximation: Classical Limit of Quantum Statistics;
  • energy shells and entropy derivatives: Microcanonical Ensemble;
  • natural variables and Legendre transforms among potentials: Thermodynamic Potentials;
  • Bose–Einstein and Fermi–Dirac occupations: Quantum Statistics and Ideal Gases;
  • full path-integral derivations: Quantum Dynamics;
  • compact lookup identities: Ensemble Formula Sheet.

Tr⁡HN\operatorname{Tr}_{\mathcal H_N} and Tr⁡F\operatorname{Tr}_{\mathcal F} define different ensembles.

An energy value with degeneracy gag_a contributes

gae−βEa.g_ae^{-\beta E_a}.

Calling the one-particle sum the many-body partition function

Section titled “Calling the one-particle sum the many-body partition function”

q(β)q(\beta) becomes qNq^N only for labeled noninteracting particles. Exchange symmetry changes the state count.

Dividing by the factorial outside its regime

Section titled “Dividing by the factorial outside its regime”

qN/N!q^N/N! is the Maxwell–Boltzmann approximation for dilute identical particles, not the exact Bose or Fermi answer at arbitrary density.

Independent terms and a tensor-product trace are required for ZAB=ZAZBZ_{AB}=Z_AZ_B.

A formal trace that diverges does not normalize a density operator.

Fixed fugacity, fixed α=βμ\alpha=\beta\mu, and fixed chemical potential are not interchangeable under β\beta derivatives.

Replacing quantum susceptibility by an ordinary variance

Section titled “Replacing quantum susceptibility by an ordinary variance”

For noncommuting observables, second source derivatives generate imaginary-time correlations.

Use logarithms and shifted energies to avoid overflow and underflow.

Let UU be unitary. Show that

Tr⁡e−βUHU†=Tr⁡e−βH.\operatorname{Tr} e^{-\beta UHU^\dagger} = \operatorname{Tr} e^{-\beta H}.
Solution

Functional calculus gives

e−βUHU†=Ue−βHU†.e^{-\beta UHU^\dagger} = Ue^{-\beta H}U^\dagger.

Using cyclicity,

Tr⁡e−βUHU†=Tr⁡(Ue−βHU†)=Tr⁡(U†Ue−βH)=Tr⁡e−βH.\begin{aligned} \operatorname{Tr} e^{-\beta UHU^\dagger} &= \operatorname{Tr} \left( Ue^{-\beta H}U^\dagger \right) \\ &= \operatorname{Tr} \left( U^\dagger Ue^{-\beta H} \right) \\ &= \operatorname{Tr} e^{-\beta H}. \end{aligned}

The partition function depends on the spectrum with multiplicity, not on the chosen basis.

For H′=H+CIH'=H+C\mathbb I, derive Z′Z', ρ′\rho', and F′F'.

Solution

Because CIC\mathbb I commutes with HH,

e−βH′=e−βCe−βH.e^{-\beta H'} = e^{-\beta C} e^{-\beta H}.

Therefore

Z′=e−βCZ.Z' = e^{-\beta C}Z.

The normalized state is

ρ′=e−βCe−βHe−βCZ=ρ.\rho' = \frac{ e^{-\beta C}e^{-\beta H} }{ e^{-\beta C}Z } = \rho.

Finally,

F′=−kBTln⁡Z′=−kBT(−βC+ln⁡Z)=C+F.\begin{aligned} F' &= -k_{\mathrm B}T\ln Z' \\ &= -k_{\mathrm B}T \left( -\beta C+\ln Z \right) \\ &= C+F. \end{aligned}

Starting from

KH(t)=ln⁡Z(β−t)−ln⁡Z(β),K_H(t) = \ln Z(\beta-t) -\ln Z(\beta),

show that the first two cumulants are the mean and variance of HH.

Solution

The first derivative at t=0t=0 is

∂KH∂t∣t=0=−∂ln⁡Z∂β=⟨H⟩.\left. \frac{\partial K_H}{\partial t} \right|_{t=0} = -\frac{\partial\ln Z}{\partial\beta} = \langle H\rangle.

The second derivative is

∂2KH∂t2∣t=0=∂2ln⁡Z∂β2.\left. \frac{\partial^2K_H}{\partial t^2} \right|_{t=0} = \frac{\partial^2\ln Z}{\partial\beta^2}.

Direct differentiation gives

∂2ln⁡Z∂β2=⟨H2⟩−⟨H⟩2.\frac{\partial^2\ln Z}{\partial\beta^2} = \langle H^2\rangle -\langle H\rangle^2.

Hence

κ1(H)=⟨H⟩,κ2(H)=Var⁡(H).\kappa_1(H) = \langle H\rangle, \qquad \kappa_2(H) = \operatorname{Var}(H).

For one-particle energies 00 and Δ\Delta, compute the two-particle partition function for distinguishable particles, identical bosons, and identical spinless fermions.

Solution

Set

x=e−βΔ.x=e^{-\beta\Delta}.

For labeled particles,

Z2dist=(1+x)2=1+2x+x2.Z_2^{\mathrm{dist}} = (1+x)^2 = 1+2x+x^2.

For bosons, the occupations (2,0)(2,0), (1,1)(1,1), and (0,2)(0,2) have energies 00, Δ\Delta, and 2Δ2\Delta, so

Z2B=1+x+x2.Z_2^{\mathrm B} = 1+x+x^2.

For two spinless fermions, both available orbitals must be occupied, giving total energy Δ\Delta:

Z2F=x.Z_2^{\mathrm F} = x.

For two fermionic modes with one-particle energies ϵ1\epsilon_1 and ϵ2\epsilon_2, expand

Ξ=(1+ze−βϵ1)(1+ze−βϵ2)\Xi = \left( 1+ze^{-\beta\epsilon_1} \right) \left( 1+ze^{-\beta\epsilon_2} \right)

and identify Z0Z_0, Z1Z_1, and Z2Z_2.

Solution

Expanding,

Ξ=1+z(e−βϵ1+e−βϵ2)+z2e−β(ϵ1+ϵ2).\begin{aligned} \Xi ={}& 1 \\ &+ z \left( e^{-\beta\epsilon_1} +e^{-\beta\epsilon_2} \right) \\ &+ z^2 e^{-\beta(\epsilon_1+\epsilon_2)}. \end{aligned}

Comparing with

Ξ=∑N=02zNZN\Xi = \sum_{N=0}^{2} z^NZ_N

gives

Z0=1,Z_0=1, Z1=e−βϵ1+e−βϵ2,Z_1 = e^{-\beta\epsilon_1} +e^{-\beta\epsilon_2},

and

Z2=e−β(ϵ1+ϵ2).Z_2 = e^{-\beta(\epsilon_1+\epsilon_2)}.

The N=2N=2 sector has one state because Pauli exclusion forces both modes to be occupied.

Let

H=HA⊗IB+IA⊗HB+gVAB.H = H_A\otimes\mathbb I_B +\mathbb I_A\otimes H_B +gV_{AB}.

For which value of gg does exact factorization follow immediately? Explain why commutation of VABV_{AB} with the uncoupled Hamiltonian is not by itself enough to give Z=ZAZBZ=Z_AZ_B.

Solution

At

g=0,g=0,

the Hamiltonian is a sum of operators on independent tensor factors, so

ZAB=ZAZB.Z_{AB} = Z_AZ_B.

If g≠0g\neq0 and VABV_{AB} commutes with the uncoupled Hamiltonian, then

e−βH=e−βHA⊗e−βHBe−βgVAB,e^{-\beta H} = e^{-\beta H_A} \otimes e^{-\beta H_B} e^{-\beta gV_{AB}},

with the tensor factors understood on the full space. The extra operator e−βgVABe^{-\beta gV_{AB}} generally couples the trace over AA and BB, so its trace does not split into ZAZBZ_AZ_B.

Commutation simplifies the exponential but does not remove the interaction or restore a product state count.

  • R. K. Pathria and P. D. Beale, Statistical Mechanics, 4th ed., Elsevier (2021), chapters 3–8.
  • K. Huang, Statistical Mechanics, 2nd ed., Wiley (1987), chapters 7–12.
  • M. Kardar, Statistical Physics of Particles, Cambridge University Press (2007), chapters 2–6.
  • L. D. Landau and E. M. Lifshitz, Statistical Physics, Part 1, 3rd ed., Butterworth–Heinemann (1980), sections 31–56.
  • A. L. Fetter and J. D. Walecka, Quantum Theory of Many-Particle Systems, Dover (2003), chapters 1 and 2.
  • R. P. Feynman and A. R. Hibbs, Quantum Mechanics and Path Integrals, emended ed., Dover (2010), chapters 2 and 10.
  • H. Kleinert, Path Integrals in Quantum Mechanics, Statistics, Polymer Physics, and Financial Markets, 5th ed., World Scientific (2009), chapters 2 and 7.
  • C. N. Yang and T. D. Lee, “Statistical Theory of Equations of State and Phase Transitions. I. Theory of Condensation”, Physical Review 87, 404–409 (1952).