Skip to content

Symmetry Classification Preview

Modern symmetry classification asks a structural question:

Given the symmetries and anti-symmetries of a Hamiltonian, what kinds of spectra, degeneracies, and topological phases are possible?

This page is a preview. It explains the vocabulary that connects ordinary quantum-mechanical discrete symmetries to the tenfold Altland–Zirnbauer classification used for noninteracting fermion systems and Bogoliubov–de Gennes Hamiltonians. Detailed classification by spatial dimension, topology, interactions, and crystalline symmetries belongs to quantum matter and field-theory treatments.

The main message is modest but powerful: once all ordinary commuting unitary symmetries have been used to block-diagonalize the Hilbert space, three remaining symmetry types often organize the problem:

  • time reversal;
  • particle–hole structure;
  • chiral, or sublattice, symmetry.

Their presence or absence, and the signs of their squares, determine the ten Altland–Zirnbauer symmetry classes.

A list of symmetries says which transformations preserve a particular Hamiltonian. A classification scheme goes further: it groups Hamiltonians by the algebraic constraints those transformations impose.

For a Hamiltonian HH, an ordinary unitary symmetry satisfies

UHU−1=H.UHU^{-1}=H.

If UU commutes with HH, one can often decompose the Hilbert space into UU eigenspaces and study each block separately. The Altland–Zirnbauer classification usually assumes this block decomposition has already been done. It then focuses on transformations that relate a block to itself while either preserving or negating the Hamiltonian.

This is why the classification is not just a longer list of parity, rotations, translations, and spin symmetries. It is a normal form for certain remaining internal antiunitary and spectral symmetries.

Time reversal is represented by an antiunitary operator, here denoted T\mathcal T. For a Bloch or momentum-space Hamiltonian, the symmetry condition is

TH(k)T−1=H(−k).\mathcal T H(\mathbf k)\mathcal T^{-1} = H(-\mathbf k).

For a finite Hamiltonian with no momentum label, the same idea is simply

THT−1=H.\mathcal T H\mathcal T^{-1}=H.

The square of time reversal matters:

T2=+IorT2=−I.\mathcal T^2=+I \qquad \text{or} \qquad \mathcal T^2=-I.

Spinless time reversal often has T2=+I\mathcal T^2=+I. A single spin-1/21/2 degree of freedom has T2=−I\mathcal T^2=-I, which leads to Kramers degeneracy when the Hamiltonian is time-reversal invariant.

In band theory, T\mathcal T maps a state at k\mathbf k to a state at −k-\mathbf k. At momenta equivalent to their negatives modulo a reciprocal lattice vector, the T2=−I\mathcal T^2=-I case forces Kramers pairs in the same momentum sector.

Particle–hole symmetry in the Altland–Zirnbauer sense is represented by an antiunitary operator C\mathcal C satisfying

CH(k)C−1=−H(−k).\mathcal C H(\mathbf k)\mathcal C^{-1} = -H(-\mathbf k).

For a finite Hamiltonian, this reads

CHC−1=−H.\mathcal C H\mathcal C^{-1} = -H.

This is not the same as ordinary charge conservation. In many superconducting applications, C\mathcal C is a redundancy created by writing a Bogoliubov–de Gennes Hamiltonian in a Nambu basis containing both particle and hole variables. It guarantees a spectrum symmetric about the chosen zero of energy:

E⟷−E.E \quad\longleftrightarrow\quad -E.

Like time reversal, particle–hole structure has two possible square signs in the tenfold classification:

C2=+IorC2=−I.\mathcal C^2=+I \qquad \text{or} \qquad \mathcal C^2=-I.

The sign depends on the representation and on the physical degrees of freedom included in the block.

Chiral symmetry, also called sublattice symmetry in many tight-binding models, is a unitary transformation S\mathcal S that anticommutes with the Hamiltonian:

SH(k)S−1=−H(k).\mathcal S H(\mathbf k)\mathcal S^{-1} = -H(\mathbf k).

Equivalently,

{S,H}=0.\{\mathcal S,H\}=0.

This also gives a spectrum symmetric around zero energy. If

H∣ψ⟩=E∣ψ⟩,H\lvert\psi\rangle=E\lvert\psi\rangle,

then

HS∣ψ⟩=−E S∣ψ⟩.H\mathcal S\lvert\psi\rangle = -E\,\mathcal S\lvert\psi\rangle.

A simple source of chiral symmetry is a bipartite hopping Hamiltonian with no same-sublattice hopping. In a basis split into two sublattices, it has the block form

H=(0AA†0),S=(I00−I).H = \begin{pmatrix} 0&A\\ A^\dagger&0 \end{pmatrix}, \qquad \mathcal S = \begin{pmatrix} I&0\\ 0&-I \end{pmatrix}.

Then SHS−1=−H\mathcal S H\mathcal S^{-1}=-H. The SSH chain is a standard reference model for this structure.

When both T\mathcal T and C\mathcal C are present, their product can give a chiral symmetry:

S=TC,\mathcal S = \mathcal T\mathcal C,

up to convention and provided the two symmetries act within the same block.

The Altland–Zirnbauer classes are labeled by whether T\mathcal T, C\mathcal C, and S\mathcal S are absent or present. For antiunitary T\mathcal T and C\mathcal C, the table records the square as +1+1 or −1-1. A 00 means the symmetry is absent in that class.

ClassT2\mathcal T^2C2\mathcal C^2S\mathcal STypical description
A000000no time reversal, no particle–hole, no chiral symmetry
AIII000011chiral unitary class
AI+1+10000spinless time reversal
BDI+1+1+1+111time reversal and particle–hole with positive squares
D00+1+100particle–hole structure, common for idealized superconductors
DIII−1-1+1+111spinful time reversal plus particle–hole
AII−1-10000spinful time reversal, Kramers structure
CII−1-1−1-111time reversal and particle–hole with negative squares
C00−1-100particle–hole with negative square
CI+1+1−1-111time reversal positive, particle–hole negative

This table is not yet a table of topological phases. It is a table of symmetry constraints. To get topological classifications one must also specify spatial dimension, locality assumptions, whether interactions are included, and whether crystalline symmetries are being used.

In gapped noninteracting systems, a Hamiltonian is often classified up to continuous deformations that preserve:

  • the energy gap;
  • locality or smoothness in momentum space;
  • the symmetry class.

If two Hamiltonians cannot be connected without closing the gap or breaking the symmetry, they belong to different topological phases.

Examples include:

  • class A in two dimensions, where a Chern number can classify integer quantum Hall bands;
  • class AII topological insulators in two dimensions, where time reversal with T2=−I\mathcal T^2=-I leads to the quantum spin Hall pattern;
  • class D in one dimension, where the Kitaev chain is the standard superconducting model;
  • chiral classes where winding numbers can appear in one-dimensional models such as the SSH chain.

The invariant is not determined by the class name alone. Dimension matters. The same symmetry class can have different possible invariants in different spatial dimensions.

This preview deliberately leaves several major topics to later pages:

  • the periodic table of topological insulators and superconductors;
  • K-theory and homotopy derivations of the tenfold way;
  • interacting reductions of noninteracting classifications;
  • crystalline, magnetic, and spatial symmetry classifications;
  • disorder and localization;
  • edge modes and bulk–boundary correspondence;
  • relativistic charge conjugation and CPT as field-theoretic symmetries.

The preview is here so that the symbols T\mathcal T, C\mathcal C, S\mathcal S, and the labels A, AII, D, BDI, and related classes are not mysterious when they first appear.

  • Treating particle–hole symmetry in a Bogoliubov–de Gennes Hamiltonian as ordinary charge conservation.
  • Forgetting that T\mathcal T and C\mathcal C are antiunitary, so they conjugate complex coefficients.
  • Calling a model class AII just because it has spin; the time-reversal operator must satisfy T2=−I\mathcal T^2=-I and preserve the Hamiltonian.
  • Ignoring ordinary unitary symmetries that should be block-diagonalized before applying the tenfold classification.
  • Assuming the tenfold way includes crystalline symmetries. It does not; crystalline symmetries refine or extend it.
  • Thinking an Altland–Zirnbauer class by itself determines a topological invariant without specifying dimension and gap assumptions.
  • A. Altland and M. R. Zirnbauer, “Nonstandard symmetry classes in mesoscopic normal-superconducting hybrid structures,” Physical Review B 55, 1142, 1997.
  • A. P. Schnyder, S. Ryu, A. Furusaki, and A. W. W. Ludwig, “Classification of topological insulators and superconductors in three spatial dimensions,” Physical Review B 78, 195125, 2008.
  • A. Kitaev, “Periodic table for topological insulators and superconductors,” AIP Conference Proceedings 1134, 22-30, 2009.
  • S. Ryu, A. P. Schnyder, A. Furusaki, and A. W. W. Ludwig, “Topological insulators and superconductors: tenfold way and dimensional hierarchy,” New Journal of Physics 12, 065010, 2010.
  • C.-K. Chiu, J. C. Y. Teo, A. P. Schnyder, and S. Ryu, “Classification of topological quantum matter with symmetries,” Reviews of Modern Physics 88, 035005, 2016.
  • M. Z. Hasan and C. L. Kane, “Colloquium: Topological insulators,” Reviews of Modern Physics 82, 3045, 2010.
  • X.-L. Qi and S.-C. Zhang, “Topological insulators and superconductors,” Reviews of Modern Physics 83, 1057, 2011.
  1. Classify a spinless real Hamiltonian.

Suppose HH is a finite-dimensional real symmetric Hamiltonian with no particle–hole or chiral symmetry. Taking T=K\mathcal T=K, which Altland–Zirnbauer class is suggested?

Solution

Complex conjugation satisfies

KHK−1=HKHK^{-1}=H

for a real Hamiltonian, and

K2=+I.K^2=+I.

There is no particle–hole or chiral symmetry by assumption. The class is therefore AI.

  1. Identify the time-reversal class for a spinful band.

A spin-1/21/2 Bloch Hamiltonian satisfies

TH(k)T−1=H(−k),T2=−I,\mathcal T H(\mathbf k)\mathcal T^{-1} = H(-\mathbf k), \qquad \mathcal T^2=-I,

and has no particle–hole or chiral symmetry. Which class is it?

Solution

The only Altland–Zirnbauer symmetry is time reversal with negative square:

T2=−I.\mathcal T^2=-I.

With C=0\mathcal C=0 and S=0\mathcal S=0, the class is AII.

  1. Check chiral symmetry in a bipartite model.

For

H=(0AA†0),S=(I00−I),H = \begin{pmatrix} 0&A\\ A^\dagger&0 \end{pmatrix}, \qquad \mathcal S = \begin{pmatrix} I&0\\ 0&-I \end{pmatrix},

show that SHS−1=−H\mathcal S H\mathcal S^{-1}=-H.

Solution

Since S−1=S\mathcal S^{-1}=\mathcal S, multiply the matrices:

SHS=(I00−I)(0AA†0)(I00−I).\mathcal S H\mathcal S = \begin{pmatrix} I&0\\ 0&-I \end{pmatrix} \begin{pmatrix} 0&A\\ A^\dagger&0 \end{pmatrix} \begin{pmatrix} I&0\\ 0&-I \end{pmatrix}.

The left multiplication changes the sign of the lower block row, and the right multiplication changes the sign of the right block column. The result is

(0−A−A†0)=−H.\begin{pmatrix} 0&-A\\ -A^\dagger&0 \end{pmatrix} = -H.
  1. Explain why particle–hole symmetry pairs energies.

If CHC−1=−H\mathcal C H\mathcal C^{-1}=-H and H∣ψ⟩=E∣ψ⟩H\lvert\psi\rangle=E\lvert\psi\rangle, show that C∣ψ⟩\mathcal C\lvert\psi\rangle has energy −E-E.

Solution

From CHC−1=−H\mathcal C H\mathcal C^{-1}=-H, one has

HC=−CH.H\mathcal C = -\mathcal C H.

Apply this to ∣ψ⟩\lvert\psi\rangle:

HC∣ψ⟩=−CH∣ψ⟩=−C(E∣ψ⟩).H\mathcal C\lvert\psi\rangle = -\mathcal C H\lvert\psi\rangle = -\mathcal C(E\lvert\psi\rangle).

Because C\mathcal C is antiunitary, the scalar is conjugated:

C(E∣ψ⟩)=E∗C∣ψ⟩.\mathcal C(E\lvert\psi\rangle) = E^*\mathcal C\lvert\psi\rangle.

For a self-adjoint Hamiltonian EE is real, so

HC∣ψ⟩=−E C∣ψ⟩.H\mathcal C\lvert\psi\rangle = -E\,\mathcal C\lvert\psi\rangle.

Thus the spectrum is paired about zero energy.