|q⟩ Bad Qubits

intermediate · Physics · Spin-½ Systems & Pauli Algebra

Pauli Algebra and Anticommutators

The Pauli matrices obey two complementary product rules: a commutator relation that encodes how rotations about different axes fail to commute, and an anticommutator relation that encodes their orthogonality. Together they generate the entire algebra of spin-½.

The commutator relation

The commutator of two operators is [A,B]=ABBA[A, B] = AB - BA. For the Pauli matrices,

[σx,σy]=2iσz,[σy,σz]=2iσx,[σz,σx]=2iσy.[\sigma_x, \sigma_y] = 2i\,\sigma_z, \qquad [\sigma_y, \sigma_z] = 2i\,\sigma_x, \qquad [\sigma_z, \sigma_x] = 2i\,\sigma_y.

These are cyclic: advancing xyzxx \to y \to z \to x reproduces the next line. Compactly, with the Levi-Civita symbol εjkl\varepsilon_{jkl} and the Einstein summation convention,

[σj,σk]=2iεjklσl.[\sigma_j, \sigma_k] = 2i\,\varepsilon_{jkl}\,\sigma_l.

Let us verify the first relation directly. Using σxσy=iσz\sigma_x\sigma_y = i\sigma_z and σyσx=iσz\sigma_y\sigma_x = -i\sigma_z,

[σx,σy]=σxσyσyσx=iσz(iσz)=2iσz.[\sigma_x, \sigma_y] = \sigma_x\sigma_y - \sigma_y\sigma_x = i\sigma_z - (-i\sigma_z) = 2i\,\sigma_z.

The fact that σx\sigma_x and σy\sigma_y do not commute is the algebraic origin of the uncertainty relation between spin components: you cannot have a definite value of SxS_x and SyS_y simultaneously.

The anticommutator relation

The anticommutator is {A,B}=AB+BA\{A, B\} = AB + BA. The Paulis satisfy

{σj,σk}=2δjkI.\{\sigma_j, \sigma_k\} = 2\,\delta_{jk}\,I.

So distinct Paulis anticommute ({σx,σy}=0\{\sigma_x, \sigma_y\} = 0), while each squares to the identity ({σx,σx}=2I\{\sigma_x, \sigma_x\} = 2I, i.e. σx2=I\sigma_x^2 = I). Anticommutation is the matrix analogue of two unit vectors being orthogonal.

The master product identity

Adding half the commutator and half the anticommutator gives a single formula for any product of two Paulis:

σjσk=δjkI+iεjklσl.\sigma_j \sigma_k = \delta_{jk}\,I + i\,\varepsilon_{jkl}\,\sigma_l.

For example σxσy=iσz\sigma_x\sigma_y = i\sigma_z (here jkj\neq k, so only the ε\varepsilon term survives), and σzσz=I\sigma_z\sigma_z = I (here j=kj = k, so only the δ\delta term survives). This one identity lets you multiply Paulis without ever writing out a matrix.

Try it

Compute the commutator [σx,σy]=σxσyσyσx[\sigma_x, \sigma_y] = \sigma_x\sigma_y - \sigma_y\sigma_x as a complex 2×22\times 2 matrix and return the imaginary part of its (0,0)(0,0) entry. Since [σx,σy]=2iσz[\sigma_x,\sigma_y] = 2i\sigma_z and (σz)00=1(\sigma_z)_{00} = 1, the answer should be 22.

Run your code to see the quantum state.

Sign in on the full site to ask questions and join the discussion.