research38 min studyLesson 800 in Mathematics

Modular theory and KMS states

Construct modular flow from a faithful normal state and verify its analytic equilibrium boundary identity with an explicit sign convention.

On this page

What you’ll understand

  • State the cyclic-separating and faithful-normal hypotheses needed for modular operators.
  • Track the anti-linear Tomita operator, its polar decomposition, and modular automorphisms.
  • Verify a KMS boundary identity on analytic matrix units with an explicit complex-time convention.

A faithful normal state can determine its own canonical flow of observables, even when no external Hamiltonian has been specified. Tomita–Takesaki theory constructs that flow from an algebra and a vector. We assume a complex von Neumann algebra admitting a faithful normal state and work in its GNS representation, with a unit cyclic separating vector Omega. These are substantive hypotheses: an arbitrary vector in an arbitrary concrete representation need not be separating, and not every von Neumann algebra admits a faithful normal state.

A state fixes a densely defined anti-linear operator

Cyclic means M Omega is dense; separating means a Omega equal to zero forces a equal to zero. On that dense set define S-zero by sending a Omega to a-star Omega. Separation makes the prescription well-defined, while the adjoint conjugates scalars, so the operator is anti-linear. It is closable; its closure S has polar decomposition J times the positive square root of Delta. Delta is generally an unbounded positive self-adjoint injective operator, and J is an antiunitary involution.

Tomita–Takesaki's theorem states that J conjugates M onto its commutant and the unitary powers of Delta preserve M by conjugation. Thus sigma-t of a, defined using Delta to the it, is an automorphism of M for every real t. The resulting group is appropriately operator-topology continuous, not generally norm-continuous. This theorem is stated here rather than proved: closability, unbounded domains, and invariance of the algebra are the central analysis, not formal consequences of writing a polar decomposition.

S0(aΩ)=a∗Ω,S=JΔ1/2,JMJ=M′,σt(a)=ΔitaΔ−it∈MS_0(a\Omega)=a^*\Omega,\qquad S=J\Delta^{1/2},\qquad JMJ=M',\qquad \sigma_t(a)=\Delta^{it}a\Delta^{-it}\in M
Ω is a unit cyclic separating vector. S₀ is anti-linear on MΩ and is closed before its polar decomposition is used.↔ Scroll wide equations horizontally.

KMS is an analytic boundary relation with an ordering convention

We use the following KMS convention, matching the cited modular-theory source. For a σ-weakly continuous one-parameter group alpha-t of *-automorphisms of M and inverse parameter beta positive, require for every a,b in M a bounded scalar function analytic in the strip with imaginary part between zero and beta, continuous on its closure, whose lower boundary is phi(alpha-t(a)b) and upper boundary is phi(b alpha-t(a)). With this ordering the modular group of a faithful normal state satisfies the condition at beta one. Reversing the placement of the evolving observable changes the associated imaginary-time sign.

For entire analytic elements, the boundary condition can be written phi(alpha at i-beta of a times b)=phi(ba). That shorthand is not valid for an arbitrary bounded a unless its analytic continuation exists in the required domain. The general strip statement avoids pretending every operator has a value at complex time. A suitable invariant analytic *-subalgebra is ultraweakly dense for modular flow. Physical time and its temperature normalization must also be specified before identifying this canonical parameter with a laboratory evolution.

Fa,b(t)=φ(αt(a)b),Fa,b(t+iβ)=φ(bαt(a)),φ(αiβ(a)b)=φ(ba) for analytic aF_{a,b}(t)=\varphi(\alpha_t(a)b),\qquad F_{a,b}(t+i\beta)=\varphi(b\alpha_t(a)),\qquad \varphi(\alpha_{i\beta}(a)b)=\varphi(ba)\ \text{for analytic }a
The displayed ordering fixes the sign convention. For the modular automorphism group σ, β=1 under this convention; the full definition uses bounded strip-analytic scalar functions.↔ Scroll wide equations horizontally.

Faithful density matrices give a computable model

For M_n(C) with an invertible density matrix rho, represent the algebra by left multiplication on Hilbert–Schmidt matrices and take Omega=rho to the one half. This vector is cyclic and separating. The modular operator sends X to rho X rho inverse, and modular automorphisms send a to rho to the it times a times rho to the minus it. All matrix elements are entire analytic, so complex-time calculations here have no hidden domain problem.

If rho is a scalar multiple of the identity, the state is tracial and modular flow is trivial even though M_n can be noncommutative. Conversely, the modular group of a faithful normal state is trivial exactly when the state is tracial. A rank-deficient density matrix cannot be inserted into the inverse-rho formula on the full algebra. One must change the support or representation appropriately; calling its ordinary GNS vector separating would contradict the null positive elements detected in earlier lessons.

Check the KMS sign on one matrix unit

Let ρ=diag(3/4,1/4), φ(a)=Tr(ρa), a=e₁₂ and b=e₂₁. Calculate modular flow and verify the β=1 analytic boundary identity.

  1. Conjugation by ρ^(it) multiplies e₁₂ by ((3/4)/(1/4))^(it)=3^(it). Hence σ_t(e₁₂)=exp(it log3)e₁₂.

  2. At complex parameter i, the factor is 3^(−1)=1/3, so σ_i(a)b=(1/3)e₁₁ and φ(σ_i(a)b)=(1/3)(3/4)=1/4.

  3. The reverse product ba=e₂₂ has φ(ba)=1/4, agreeing with the boundary identity. The state is faithful but not tracial, and the real modular flow is nontrivial.

The result

The analytic shift has the reciprocal factor 1/3, confirming the declared KMS ordering and sign. These finite matrices have entire analytic continuation.

Explore the model

Interactive lab

Interactive lab · quantum

A controlled analogy for a rotating coherence

Pause, reset, then set θ=90 degrees and φ=60 degrees. Predict basis probabilities 1/2 and Bloch coordinates (1/2,√3/2,0). Playback adds relative phase at one radian per lab-time unit. The worked modular matrix unit rotates at log3 per modular-time unit, so comparing rates requires an explicit rescaling. The instrument is a pure two-level state, not a faithful mixed density matrix or a Tomita operator.

A controlled analogy for a rotating coherenceA normalized two-level pure state, not an interacting field simulation. Playback adds one radian per second to the relative phase; basis probabilities stay fixed. Current readings: Probability |0⟩: 71.1%; Probability |1⟩: 28.9%; Bloch (x, y, z): 0.78, 0.45, 0.42. All parameters are adjustable with the labeled controls below.xyz|0⟩|1⟩Basis probabilitiesP(0)71.1%P(1)28.9%
Bloch vector / P(0)P(1)Phase latitude
Probability |0⟩
71.1%
Probability |1⟩
28.9%
Bloch (x, y, z)
0.78, 0.45, 0.42
t = 0 s

A normalized two-level pure state, not an interacting field simulation. Playback adds one radian per second to the relative phase; basis probabilities stay fixed. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forCoherences can rotate while diagonal probabilities stay fixed, but a modular-flow claim needs its algebra, faithful state, and time convention.

Check your understanding

Which condition makes Ω separating in the GNS representation used for this modular construction?

Try it yourself

For M₂(C) with ρ=I₂/2, calculate the modular group and explain why its triviality does not imply that the algebra is commutative.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Scalar powers of ρ commute with every matrix, and normalized matrix trace is cyclic.

Compare with the solution

The factors ρ^(it) are scalar multiples of I, so σ_t(a)=a for every matrix and real t. The state is normalized trace, hence φ(ab)=φ(ba), making the KMS boundary function constant under the stated ordering. Nevertheless e₁₂e₂₁=e₁₁ and e₂₁e₁₂=e₂₂ differ. Trivial modular flow diagnoses traciality of this faithful normal state, not commutativity of the algebra.

Take it with you

  • Modular theory requires a cyclic separating vector or a faithful normal state in a suitable representation.
  • KMS identities need a declared boundary ordering and analytic-domain hypotheses.
  • Finite faithful density matrices make modular flow explicit; nonfaithful inverse formulas require support changes.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.