Skip to content

Bloch sphere geometry is the coordinate geometry of a two-dimensional complex Hilbert space after normalization and global phase have been removed. It is special to qubits: the pure-state projective space CP1\mathbb{CP}^1 is a two-sphere, and every one-qubit density matrix is a point in the corresponding three-dimensional ball.

This page gives the linear-algebra formulas. The spinor interpretation is Bloch Sphere, and the density-operator introduction is Bloch Sphere in Core Formalism.

Helpful background. Pauli matrices for the two-by-two operator basis; Hermitian operators for adjoints and real spectra.

The matrices

I,σx,σy,σzI,\sigma_x,\sigma_y,\sigma_z

form a real basis for Hermitian 2×22\times2 matrices. Therefore every Hermitian trace-one matrix can be written uniquely as

ρ(r)=12(I+r⋅σ),\rho(\mathbf r) = \frac12 \left( I+\mathbf r\cdot\boldsymbol\sigma \right),

where

σ=(σx,σy,σz),r=(rx,ry,rz)∈R3.\boldsymbol\sigma = (\sigma_x,\sigma_y,\sigma_z), \qquad \mathbf r=(r_x,r_y,r_z)\in\mathbb R^3.

The coordinates are Pauli expectation values:

ri=Tr⁡(ρσi),i=x,y,z.r_i = \operatorname{Tr}(\rho\sigma_i), \qquad i=x,y,z.

Thus the Bloch vector is not an additional physical object. It is the density matrix written in the Pauli basis.

Using

(r⋅σ)2=∥r∥2I,(\mathbf r\cdot\boldsymbol\sigma)^2 = \lVert\mathbf r\rVert^2 I,

the eigenvalues of ρ(r)\rho(\mathbf r) are

λ±=12(1±∥r∥).\lambda_\pm = \frac12 \left( 1\pm\lVert\mathbf r\rVert \right).

Therefore

ρ(r)≥0⟺∥r∥≤1.\rho(\mathbf r)\ge0 \quad \Longleftrightarrow \quad \lVert\mathbf r\rVert\le1.

The allowed one-qubit density matrices fill the unit ball. Surface points are pure states; interior points are mixed states; the center r=0\mathbf r=\mathbf0 is the maximally mixed state I/2I/2.

Cross-section of the Bloch ball at ry=0, with pure states on the boundary, the maximally mixed state at the center, and a mixed state halfway up the z axis.

This ry=0r_y=0 cross-section shows the distinction between the pure-state sphere and the mixed-state interior. The point (0,0,1/2)(0,0,1/2) has σz\sigma_z probabilities 3/43/4 and 1/41/4; the center gives equal probabilities along every axis.

A normalized qubit ray can be represented as

∣ψ(θ,ϕ)⟩=cos⁡θ2 ∣0⟩+eiϕsin⁡θ2 ∣1⟩,\lvert\psi(\theta,\phi)\rangle = \cos\frac{\theta}{2}\,\lvert0\rangle + e^{i\phi}\sin\frac{\theta}{2}\,\lvert1\rangle,

where

0≤θ≤π,0≤ϕ<2π.0\le\theta\le\pi, \qquad 0\le\phi<2\pi.

The associated unit Bloch vector is

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

The rank-one projector onto the ray is

∣ψ(θ,ϕ)⟩⟨ψ(θ,ϕ)∣=12(I+n⋅σ).\lvert\psi(\theta,\phi)\rangle \langle\psi(\theta,\phi)\rvert = \frac12 \left( I+\mathbf n\cdot\boldsymbol\sigma \right).

The global phase of ∣ψ⟩\lvert\psi\rangle has disappeared. Away from the poles, the relative phase ϕ\phi remains as the azimuthal angle. At either pole one coefficient vanishes, so ϕ\phi no longer labels distinct rays.

Opposite points on the Bloch sphere are orthogonal pure states, not the same state. If n\mathbf n is a unit Bloch vector, then the projectors for the two antipodal directions are

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

They satisfy

P+P−=0,P++P−=I.P_+P_-=0, \qquad P_++P_-=I.

Do not confuse ∣ψ⟩↦−∣ψ⟩|\psi\rangle\mapsto-|\psi\rangle with n↦−n\mathbf n\mapsto-\mathbf n. The first leaves the projector unchanged; the second changes it to the orthogonal projector I−PI-P.

Let n\mathbf n and m\mathbf m be two unit Bloch vectors, with angle γ\gamma between them:

n⋅m=cos⁡γ.\mathbf n\cdot\mathbf m=\cos\gamma.

The transition probability between the corresponding pure states is

∣⟨ψ(n)∣ψ(m)⟩∣2=12(1+n⋅m)=cos⁡2γ2.\lvert \langle\psi(\mathbf n)\vert\psi(\mathbf m)\rangle \rvert^2 = \frac12(1+\mathbf n\cdot\mathbf m) = \cos^2\frac{\gamma}{2}.

The angle between rays, defined by arccos⁡∣⟨ψ∣ϕ⟩∣\arccos|\langle\psi|\phi\rangle|, is therefore γ/2\gamma/2. Orthogonal states have γ=π\gamma=\pi, while identical rays have γ=0\gamma=0.

A unitary generated by a Pauli direction has the form

U(a^,α)=exp⁡(−iα2 a^⋅σ).U(\hat{\mathbf a},\alpha) = \exp \left( -\frac{i\alpha}{2}\, \hat{\mathbf a}\cdot\boldsymbol\sigma \right).

Acting on a density matrix,

ρ(r)↦Uρ(r)U†,\rho(\mathbf r) \mapsto U\rho(\mathbf r)U^\dagger,

rotates the Bloch vector:

r↦Ra^(α)r,\mathbf r\mapsto R_{\hat{\mathbf a}}(\alpha)\mathbf r,

where Ra^(α)R_{\hat{\mathbf a}}(\alpha) is the ordinary SO(3)SO(3) rotation by angle α\alpha around a^\hat{\mathbf a}.

The map from SU(2)SU(2) spinor unitaries to SO(3)SO(3) rotations is two-to-one: UU and −U-U rotate the Bloch vector in the same way. The group-theory background is SU(2), and the physical spin version is Spin Rotations.

The projective measurement of a Pauli component along a unit direction a^\hat{\mathbf a} has effects

P±(a^)=12(I±a^⋅σ).P_\pm(\hat{\mathbf a}) = \frac12 \left( I\pm\hat{\mathbf a}\cdot\boldsymbol\sigma \right).

For a state ρ(r)\rho(\mathbf r),

p±=Tr⁡(ρ(r)P±(a^))=12(1±r⋅a^).p_\pm = \operatorname{Tr}\bigl(\rho(\mathbf r)P_\pm(\hat{\mathbf a})\bigr) = \frac12 \left( 1\pm\mathbf r\cdot\hat{\mathbf a} \right).

Geometrically, measuring along a^\hat{\mathbf a} reads the projection of the Bloch vector onto the measurement axis. A pure state at a^\hat{\mathbf a} gives the plus outcome with probability one; a state at the center gives equal probabilities in every direction.

Small changes of a pure-state Bloch direction are tangent directions on S2S^2, not arbitrary independent changes in R3\mathbb R^3. The local tangent-space language is summarized in Tangent and Cotangent Spaces.

The Bloch-ball formula depends on the Pauli basis and on the special identity

(r⋅σ)2=∥r∥2I.(\mathbf r\cdot\boldsymbol\sigma)^2 = \lVert\mathbf r\rVert^2 I.

Higher-dimensional density matrices can be expanded in generalized generator bases, but the allowed set is not simply a Euclidean ball, and pure states do not form an ordinary two-sphere. The qubit case is unusually visual.

  • Treating the Bloch vector as a two-component spinor.
  • Confusing the surface sphere of pure states with the full Bloch ball of density matrices.
  • Calling antipodal points the same state; they represent orthogonal rays.
  • Forgetting that global phase is removed but relative phase is visible as azimuth.
  • Assuming the Bloch-ball picture generalizes directly to qutrits or larger systems.
  • Missing the half-angle relation between spinors and Bloch-sphere directions.
  • Treating UU and −U-U as different rotations of the Bloch vector.
  • F. Bloch, “Nuclear induction,” Physical Review 70, 460-474, 1946.
  • B. C. Hall, Quantum Theory for Mathematicians, Springer, 2013.
  • M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press, 2010.
  • J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
  1. Find the Bloch vector for
∣+⟩=12(∣0⟩+∣1⟩).\lvert+\rangle = \frac{1}{\sqrt2} \bigl( \lvert0\rangle+\lvert1\rangle \bigr).
Solution

The projector is

∣+⟩⟨+∣=12(1111)=12(I+σx).\lvert+\rangle\langle+\rvert = \frac12 \begin{pmatrix} 1 & 1\\ 1 & 1 \end{pmatrix} = \frac12(I+\sigma_x).

Therefore the Bloch vector is

r=(1,0,0).\mathbf r=(1,0,0).
  1. Show that ρ(r)\rho(\mathbf r) is pure exactly when ∥r∥=1\lVert\mathbf r\rVert=1.
Solution

The eigenvalues of ρ(r)\rho(\mathbf r) are

λ±=12(1±∥r∥).\lambda_\pm = \frac12 \left( 1\pm\lVert\mathbf r\rVert \right).

A density matrix is pure exactly when one eigenvalue is 11 and the other is 00. This happens exactly when ∥r∥=1\lVert\mathbf r\rVert=1.

  1. Two pure states have Bloch vectors separated by angle γ=π/3\gamma=\pi/3. What is their transition probability?
Solution

Use

∣⟨ψ(n)∣ψ(m)⟩∣2=cos⁡2γ2.\lvert \langle\psi(\mathbf n)\vert\psi(\mathbf m)\rangle \rvert^2 = \cos^2\frac{\gamma}{2}.

For γ=π/3\gamma=\pi/3,

cos⁡2π6=(32)2=34.\cos^2\frac{\pi}{6} = \left(\frac{\sqrt3}{2}\right)^2 = \frac34.
  1. A state has Bloch vector r=(0,0,1/2)\mathbf r=(0,0,1/2). What are the probabilities for measuring σz\sigma_z?
Solution

For measurement direction z^\hat{\mathbf z},

p±=12(1±r⋅z^)=12(1±12).p_\pm = \frac12 \left( 1\pm\mathbf r\cdot\hat{\mathbf z} \right) = \frac12 \left( 1\pm\frac12 \right).

Therefore

p+=34,p−=14.p_+=\frac34, \qquad p_-=\frac14.
  1. Why do UU and −U-U give the same Bloch-vector rotation?
Solution

The density matrix transforms by conjugation:

ρ↦UρU†.\rho\mapsto U\rho U^\dagger.

Replacing UU by −U-U gives

ρ↦(−U)ρ(−U)†=UρU†.\rho \mapsto (-U)\rho(-U)^\dagger = U\rho U^\dagger.

Thus the Bloch vector changes in the same way. This is the two-to-one relation between SU(2)SU(2) and SO(3)SO(3) rotations.