Skip to content

Spin-Half Matrices

Spin-1/21/2 calculations reduce many measurement, rotation, and evolution problems to 2×22\times2 matrix algebra. In the ordered SzS_z eigenbasis, the physical spin operators are

Si=ℏ2σi.S_i = \frac{\hbar}{2}\sigma_i.

This card is organized around four recurring tasks:

  1. construct the operator and eigenstates for a chosen measurement axis;
  2. turn a spinor or density matrix into outcome probabilities;
  3. rotate a spin state through a specified angle;
  4. diagonalize and exponentiate a general Hermitian two-level Hamiltonian.

The conceptual explanation belongs at Pauli Matrices. For an exhaustive multiplication, trace, eigenvector, and Pauli-string lookup, use the Pauli Matrices Table.

Use

{∣+z⟩,∣−z⟩},∣+z⟩=(10),∣−z⟩=(01).\lbrace \lvert+z\rangle, \lvert-z\rangle \rbrace, \qquad \lvert+z\rangle = \begin{pmatrix}1\\0\end{pmatrix}, \qquad \lvert-z\rangle = \begin{pmatrix}0\\1\end{pmatrix}.

These satisfy

Sz∣±z⟩=±ℏ2∣±z⟩.S_z\lvert\pm z\rangle = \pm\frac{\hbar}{2} \lvert\pm z\rangle.

In this basis,

σx=(0110),σy=(0−ii0),σz=(100−1).\sigma_x = \begin{pmatrix} 0&1\\ 1&0 \end{pmatrix}, \qquad \sigma_y = \begin{pmatrix} 0&-i\\ i&0 \end{pmatrix}, \qquad \sigma_z = \begin{pmatrix} 1&0\\ 0&-1 \end{pmatrix}.

Changing the ordered basis changes the displayed matrices. The abstract operator and all physical probabilities remain unchanged if states and operators are transformed consistently.

TaskFormula
Physical spin operatorSi=ℏσi/2S_i=\hbar\sigma_i/2
Pauli productσiσj=δijI+iϵijkσk\sigma_i\sigma_j=\delta_{ij}I+i\epsilon_{ijk}\sigma_k
Directional observableSn=ℏn⋅σ/2S_{\boldsymbol n}=\hbar\boldsymbol n\cdot\boldsymbol\sigma/2
Directional projectorsP±(n)=[I±n⋅σ]/2P_\pm(\boldsymbol n)=[I\pm\boldsymbol n\cdot\boldsymbol\sigma]/2
Outcome probabilityp±=[1±r⋅n]/2p_\pm=[1\pm\boldsymbol r\cdot\boldsymbol n]/2
Active rotationUn(θ)=cos⁡(θ/2)I−isin⁡(θ/2)n⋅σU_{\boldsymbol n}(\theta)=\cos(\theta/2)I-i\sin(\theta/2)\boldsymbol n\cdot\boldsymbol\sigma
General Hermitian matrixH=c0I+c⋅σH=c_0I+\boldsymbol c\cdot\boldsymbol\sigma
Two-level energiesE±=c0±∥c∥E_\pm=c_0\pm\lVert\boldsymbol c\rVert

The compact multiplication rule is

σiσj=δijI+iϵijkσk.\sigma_i\sigma_j = \delta_{ij}I + i\epsilon_{ijk}\sigma_k.

Its symmetric and antisymmetric parts are

{σi,σj}=2δijI,[σi,σj]=2iϵijkσk.\lbrace\sigma_i,\sigma_j\rbrace = 2\delta_{ij}I, \qquad [\sigma_i,\sigma_j] = 2i\epsilon_{ijk}\sigma_k.

Consequently,

[Si,Sj]=iℏϵijkSk[S_i,S_j] = i\hbar\epsilon_{ijk}S_k

and

S2=Sx2+Sy2+Sz2=34ℏ2I.S^2 = S_x^2+S_y^2+S_z^2 = \frac{3}{4}\hbar^2I.

For three-component vectors a\boldsymbol a and b\boldsymbol b,

(a⋅σ)(b⋅σ)=(a⋅b)I+i(a×b)⋅σ.(\boldsymbol a\cdot\boldsymbol\sigma) (\boldsymbol b\cdot\boldsymbol\sigma) = (\boldsymbol a\cdot\boldsymbol b)I + i(\boldsymbol a\times\boldsymbol b) \cdot\boldsymbol\sigma.

This identity is usually faster and less error-prone than multiplying entries. For a real unit vector n\boldsymbol n,

(n⋅σ)2=I.(\boldsymbol n\cdot\boldsymbol\sigma)^2=I.

That single result controls directional eigenvalues, spectral projectors, and matrix exponentials.

Parameterize a unit vector by

n=(sin⁡θcos⁡ϕ,sin⁡θsin⁡ϕ,cos⁡θ).\boldsymbol n = (\sin\theta\cos\phi, \sin\theta\sin\phi, \cos\theta).

The dimensionless directional operator is

σn=n⋅σ=(cos⁡θe−iϕsin⁡θeiϕsin⁡θ−cos⁡θ).\sigma_{\boldsymbol n} = \boldsymbol n\cdot\boldsymbol\sigma = \begin{pmatrix} \cos\theta & e^{-i\phi}\sin\theta\\ e^{i\phi}\sin\theta & -\cos\theta \end{pmatrix}.

It has eigenvalues +1+1 and −1-1. A convenient phase convention for normalized eigenstates is

∣n,+⟩=(cos⁡(θ/2)eiϕsin⁡(θ/2)),\lvert\boldsymbol n,+\rangle = \begin{pmatrix} \cos(\theta/2)\\ e^{i\phi}\sin(\theta/2) \end{pmatrix}, ∣n,−⟩=(−e−iϕsin⁡(θ/2)cos⁡(θ/2)).\lvert\boldsymbol n,-\rangle = \begin{pmatrix} -e^{-i\phi}\sin(\theta/2)\\ \cos(\theta/2) \end{pmatrix}.

They satisfy

σn∣n,±⟩=±∣n,±⟩.\sigma_{\boldsymbol n} \lvert\boldsymbol n,\pm\rangle = \pm \lvert\boldsymbol n,\pm\rangle.

Their phases are conventional. Multiplying either eigenvector by its own constant phase changes no projector or probability.

Common Cartesian eigenstates in the stated basis are

StateColumn vector
∣+x⟩\lvert+x\rangle2−1/2(1,1)T2^{-1/2}(1,1)^{\mathsf T}
∣−x⟩\lvert-x\rangle2−1/2(1,−1)T2^{-1/2}(1,-1)^{\mathsf T}
∣+y⟩\lvert+y\rangle2−1/2(1,i)T2^{-1/2}(1,i)^{\mathsf T}
∣−y⟩\lvert-y\rangle2−1/2(1,−i)T2^{-1/2}(1,-i)^{\mathsf T}
∣+z⟩\lvert+z\rangle(1,0)T(1,0)^{\mathsf T}
∣−z⟩\lvert-z\rangle(0,1)T(0,1)^{\mathsf T}

The spectral projectors of σn\sigma_{\boldsymbol n} are

P±(n)=12(I±n⋅σ).P_\pm(\boldsymbol n) = \frac12 \left( I\pm\boldsymbol n\cdot\boldsymbol\sigma \right).

They obey

P±2=P±,P+P−=0,P++P−=I.P_\pm^2=P_\pm, \qquad P_+P_-=0, \qquad P_++P_-=I.

For a state ρ\rho, the probabilities of measuring Sn=±ℏ/2S_{\boldsymbol n}=\pm\hbar/2 are

p±=tr⁡[ρP±(n)].p_\pm = \operatorname{tr} \left[ \rho P_\pm(\boldsymbol n) \right].

If

ρ=12(I+r⋅σ),∥r∥≤1,\rho = \frac12 \left( I+\boldsymbol r\cdot\boldsymbol\sigma \right), \qquad \lVert\boldsymbol r\rVert\le1,

then

p±=12(1±r⋅n).p_\pm = \frac12 \left( 1\pm\boldsymbol r\cdot\boldsymbol n \right).

The directional mean and variance are

⟨Sn⟩=ℏ2r⋅n,\langle S_{\boldsymbol n}\rangle = \frac{\hbar}{2} \boldsymbol r\cdot\boldsymbol n, (ΔSn)2=ℏ24[1−(r⋅n)2].(\Delta S_{\boldsymbol n})^2 = \frac{\hbar^2}{4} \left[ 1-(\boldsymbol r\cdot\boldsymbol n)^2 \right].

For a pure state pointing along unit vector m\boldsymbol m, r=m\boldsymbol r=\boldsymbol m. The conditional probability for a second ideal analyzer is therefore

Pr⁡(+n∣+m)=1+m⋅n2=cos⁡2γ2,\Pr(+\boldsymbol n\mid+\boldsymbol m) = \frac{1+\boldsymbol m\cdot\boldsymbol n}{2} = \cos^2\frac{\gamma}{2},

where γ\gamma is the angle between the axes. The half-angle is a spinor feature; replacing it by cos⁡2γ\cos^2\gamma is incorrect.

Let

∣ψ⟩=α∣+z⟩+β∣−z⟩,∣α∣2+∣β∣2=1.\lvert\psi\rangle = \alpha\lvert+z\rangle + \beta\lvert-z\rangle, \qquad \lvert\alpha\rvert^2+\lvert\beta\rvert^2=1.

The pure-state Bloch vector is

r=(2Re⁡(α∗β)2Im⁡(α∗β)∣α∣2−∣β∣2).\boldsymbol r = \begin{pmatrix} 2\operatorname{Re}(\alpha^*\beta)\\ 2\operatorname{Im}(\alpha^*\beta)\\ \lvert\alpha\rvert^2-\lvert\beta\rvert^2 \end{pmatrix}.

Conversely, up to global phase, a pure state with polar angles (θ,ϕ)(\theta,\phi) is

∣ψ⟩=cos⁡θ2∣+z⟩+eiϕsin⁡θ2∣−z⟩.\lvert\psi\rangle = \cos\frac{\theta}{2} \lvert+z\rangle + e^{i\phi} \sin\frac{\theta}{2} \lvert-z\rangle.

The relative phase controls the transverse components. A global phase eiχe^{i\chi} changes neither r\boldsymbol r nor any measurement probability. Mixed-state geometry and positivity belong at the Bloch Sphere canonical page.

An active right-handed rotation of a spin-1/21/2 state through angle ϑ\vartheta about unit axis n\boldsymbol n is

Un(ϑ)=exp⁡ ⁣(−iϑℏn⋅S).U_{\boldsymbol n}(\vartheta) = \exp\!\left( -\frac{i\vartheta}{\hbar} \boldsymbol n\cdot\mathbf S \right).

Since S=ℏσ/2\mathbf S=\hbar\boldsymbol\sigma/2,

Un(ϑ)=cos⁡ϑ2,I−isin⁡ϑ2n⋅σ.U_{\boldsymbol n}(\vartheta) = \cos\frac{\vartheta}{2},I - i\sin\frac{\vartheta}{2} \boldsymbol n\cdot\boldsymbol\sigma.

Special cases are

Uz(ϑ)=(e−iϑ/200eiϑ/2)U_z(\vartheta) = \begin{pmatrix} e^{-i\vartheta/2}&0\\ 0&e^{i\vartheta/2} \end{pmatrix}

and

Ux(ϑ)=(cos⁡(ϑ/2)−isin⁡(ϑ/2)−isin⁡(ϑ/2)cos⁡(ϑ/2)).U_x(\vartheta) = \begin{pmatrix} \cos(\vartheta/2)&-i\sin(\vartheta/2)\\ -i\sin(\vartheta/2)&\cos(\vartheta/2) \end{pmatrix}.

The spinor obeys

Un(2π)=−I,Un(4π)=I.U_{\boldsymbol n}(2\pi)=-I, \qquad U_{\boldsymbol n}(4\pi)=I.

The minus sign after 2π2\pi is a global phase for an isolated spinor, but relative signs can be observed interferometrically. The associated Bloch vector has already returned after 2π2\pi.

With the active convention above,

Un(ϑ)(a⋅σ)Un†(ϑ)=[Rn(ϑ)a]⋅σ,U_{\boldsymbol n}(\vartheta) (\boldsymbol a\cdot\boldsymbol\sigma) U_{\boldsymbol n}^\dagger(\vartheta) = \left[ \mathcal R_{\boldsymbol n}(\vartheta) \boldsymbol a \right] \cdot\boldsymbol\sigma,

where Rn\mathcal R_{\boldsymbol n} is the ordinary right-handed three-dimensional rotation. Reversing the conjugation order produces the inverse rotation.

Every Hermitian 2×22\times2 Hamiltonian can be written uniquely as

H=c0I+c⋅σ,c0∈R,c∈R3.H = c_0I+\boldsymbol c\cdot\boldsymbol\sigma, \qquad c_0\in\mathbb R, \qquad \boldsymbol c\in\mathbb R^3.

For c≠0\boldsymbol c\ne\boldsymbol0, define

c=∥c∥,c^=cc.c=\lVert\boldsymbol c\rVert, \qquad \widehat{\boldsymbol c} = \frac{\boldsymbol c}{c}.

The energies and projectors are

E±=c0±c,P±=12(I±c^⋅σ).E_\pm = c_0\pm c, \qquad P_\pm = \frac12 \left( I\pm \widehat{\boldsymbol c}\cdot\boldsymbol\sigma \right).

The exact propagator is

U(t)=e−iHt/ℏ=e−ic0t/ℏ[cos⁡ ⁣(ctℏ)I−isin⁡ ⁣(ctℏ)c^⋅σ].\begin{aligned} U(t) &= e^{-iHt/\hbar} \\ &= e^{-ic_0t/\hbar} \left[ \cos\!\left(\frac{ct}{\hbar}\right)I - i\sin\!\left(\frac{ct}{\hbar}\right) \widehat{\boldsymbol c}\cdot\boldsymbol\sigma \right]. \end{aligned}

The scalar c0Ic_0I contributes a global phase to closed-system state evolution. The vector c\boldsymbol c fixes the energy axis, gap 2c2c, and Bloch-vector precession axis.

For a physical magnetic moment μ=γS\boldsymbol\mu=\gamma\mathbf S,

H=−μ⋅B=−ℏγ2B⋅σ.H = -\boldsymbol\mu\cdot\mathbf B = -\frac{\hbar\gamma}{2} \mathbf B\cdot\boldsymbol\sigma.

The sign of the gyromagnetic ratio γ\gamma matters. Do not replace this by a memorized electron formula when the two-level degree of freedom is a pseudospin, hyperfine doublet, or effective qubit.

The dimensionless matrices

σ+=σx+iσy2=(0100),σ−=σx−iσy2=(0010)\sigma_+ = \frac{\sigma_x+i\sigma_y}{2} = \begin{pmatrix}0&1\\0&0\end{pmatrix}, \qquad \sigma_- = \frac{\sigma_x-i\sigma_y}{2} = \begin{pmatrix}0&0\\1&0\end{pmatrix}

are related to physical spin ladders by

S+=ℏσ+,S−=ℏσ−.S_+=\hbar\sigma_+, \qquad S_-=\hbar\sigma_-.

The general spin-jj action and endpoint rules are collected on Ladder-Operator Action.

  1. Write the ordered basis before writing a matrix.
  2. Decide whether the symbols are dimensionless Pauli matrices or physical spin operators carrying ℏ\hbar.
  3. Convert a measurement axis to n⋅σ\boldsymbol n\cdot\boldsymbol\sigma and use spectral projectors rather than re-diagonalizing it each time.
  4. Convert a state to its Bloch vector when several directional measurements are needed.
  5. For rotations or exponentials, use (n⋅σ)2=I(\boldsymbol n\cdot\boldsymbol\sigma)^2=I.
  6. For a generic Hermitian matrix, separate the trace term c0Ic_0I from the traceless vector term.
  7. Check Hermiticity, normalization, and limiting cases such as U(0)=IU(0)=I.
  • Forgetting the factor ℏ/2\hbar/2 between σi\sigma_i and SiS_i.
  • Reversing the sign of the upper-right entry of σy\sigma_y.
  • Copying matrices from a source with the opposite basis order.
  • Treating a generic two-level pseudospin as literal spatial angular momentum.
  • Dropping relative phase when constructing the Bloch vector.
  • Using cos⁡2γ\cos^2\gamma instead of cos⁡2(γ/2)\cos^2(\gamma/2) for sequential spin analyzers.
  • Confusing active UOU†UOU^\dagger and passive U†OUU^\dagger OU conventions.
  • Treating the 2π2\pi spinor sign as a different ray in a noninterferometric measurement.
  • Using σ±\sigma_\pm and S±S_\pm interchangeably even though their dimensions differ.
  • Assuming the basis displayed here is automatically the energy basis.
  • J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
  • R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
  • L. E. Ballentine, Quantum Mechanics: A Modern Development, World Scientific, 1998.
  • M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, 10th anniversary ed., Cambridge University Press, 2010.
  1. Construct the spin-up eigenstate along n=(sin⁡θ,0,cos⁡θ)\boldsymbol n=(\sin\theta,0,\cos\theta) and verify its eigenvalue.
Solution

Here ϕ=0\phi=0, so

∣n,+⟩=(cos⁡(θ/2)sin⁡(θ/2)).\lvert\boldsymbol n,+\rangle = \begin{pmatrix} \cos(\theta/2)\\ \sin(\theta/2) \end{pmatrix}.

The directional matrix is

σn=(cos⁡θsin⁡θsin⁡θ−cos⁡θ).\sigma_{\boldsymbol n} = \begin{pmatrix} \cos\theta&\sin\theta\\ \sin\theta&-\cos\theta \end{pmatrix}.

Using angle-addition identities in the matrix product gives

σn∣n,+⟩=∣n,+⟩.\sigma_{\boldsymbol n} \lvert\boldsymbol n,+\rangle = \lvert\boldsymbol n,+\rangle.

Therefore the physical spin eigenvalue is +ℏ/2+\hbar/2.

  1. A beam prepared in ∣+z⟩\lvert+z\rangle enters an analyzer whose axis makes angle γ\gamma with zz. Find both outcome probabilities and the variance of the measured spin component.
Solution

The prepared Bloch vector is r=z^\boldsymbol r=\hat{\mathbf z}. Hence

p+=1+cos⁡γ2=cos⁡2γ2,p_+ = \frac{1+\cos\gamma}{2} = \cos^2\frac{\gamma}{2},

and

p−=1−cos⁡γ2=sin⁡2γ2.p_- = \frac{1-\cos\gamma}{2} = \sin^2\frac{\gamma}{2}.

The variance is

(ΔSn)2=ℏ24(1−cos⁡2γ)=ℏ24sin⁡2γ.(\Delta S_{\boldsymbol n})^2 = \frac{\hbar^2}{4} \left(1-\cos^2\gamma\right) = \frac{\hbar^2}{4}\sin^2\gamma.
  1. Apply an active rotation by angle π\pi about the xx axis to ∣+z⟩\lvert+z\rangle.
Solution

The rotation operator is

Ux(π)=cos⁡π2I−isin⁡π2σx=−iσx.U_x(\pi) = \cos\frac{\pi}{2}I - i\sin\frac{\pi}{2}\sigma_x = -i\sigma_x.

Since σx∣+z⟩=∣−z⟩\sigma_x\lvert+z\rangle=\lvert-z\rangle,

Ux(π)∣+z⟩=−i∣−z⟩.U_x(\pi)\lvert+z\rangle = -i\lvert-z\rangle.

The phase −i-i does not affect a subsequent projective spin measurement, so the Bloch vector points along −z-z.

  1. For
H=(Δgg−Δ),Δ,g∈R,H = \begin{pmatrix} \Delta&g\\ g&-\Delta \end{pmatrix}, \qquad \Delta,g\in\mathbb R,

find the energies and the exact propagator.

Solution

Write

H=gσx+Δσz.H = g\sigma_x+\Delta\sigma_z.

Thus c0=0c_0=0, c=(g,0,Δ)\boldsymbol c=(g,0,\Delta), and

Ω=g2+Δ2.\Omega = \sqrt{g^2+\Delta^2}.

The energies are

E±=±Ω.E_\pm=\pm\Omega.

The propagator is

U(t)=cos⁡ ⁣(Ωtℏ)I−isin⁡ ⁣(Ωtℏ)gσx+ΔσzΩ.U(t) = \cos\!\left(\frac{\Omega t}{\hbar}\right)I - i\sin\!\left(\frac{\Omega t}{\hbar}\right) \frac{g\sigma_x+\Delta\sigma_z}{\Omega}.

At g=0g=0 this reduces to diagonal phase evolution; at Δ=0\Delta=0 it becomes rotation about the xx axis.