Skip to article frontmatterSkip to article content

The Bloch Sphere

Abstract

The Bloch sphere representation of 2 dimensional quantum systems.

Keywords:Bloch VectorsPure StateMixed StatesSphere.

Definition

The Bloch sphere is a geometrical way to represent both d=2d=2 dimensional quantum systems (qubits) and the unitary dynamics of those d=2d=2 dimensional systems.

All possible qubit states, both pure and mixed, can be represented as a point on or inside of a unit sphere. Qubit dynamics can then be represented as rotations on the unit sphere.

States

Any qubit density operator ρH2\rho \in \mathcal{H}^2 has a Bloch sphere representation as

ρ=12(I+nσ)\rho = \frac{1}{2} (\mathbb{I} + \bm{n} \cdot \bm{\sigma})

where I\mathbb{I} is the d=2d=2 dimensional identity operator, nR3\bm{n} \in \mathbb{R}^{3} is the Bloch vector, and σ\bm{\sigma} is a vector of Pauli operators, such that

n=(nxnynz),   σ=(XYZ),\bm{n} = \begin{pmatrix} n_x \\ n_y \\ n_z \end{pmatrix}, ~ ~ ~ \bm{\sigma} = \begin{pmatrix} X \\ Y \\ Z \end{pmatrix},

where

X=(0110),  Y=(0ii0),  Z=(1001).X = \begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix}, ~~ Y = \begin{pmatrix} 0 & -i \\ i & 0 \end{pmatrix}, ~ ~ Z = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}.

The elements of the Bloch vector can be found as

nx=tr[ρX],ny=tr[ρY],nz=tr[ρZ].\begin{align*} n_x = \textrm{tr} \big[ \rho X \big], \\ n_y = \textrm{tr} \big[ \rho Y \big], \\ n_z = \textrm{tr} \big[ \rho Z \big]. \\ \end{align*}

The 12I\frac{1}{2} \mathbb{I} terms fixes the normalisation, such that tr[ρ]=1\textrm{tr}\big[ \rho \big] = 1.

Alternatively, in terms of spherical polar coordinates, the Bloch vector elements are given by

nx=nsin(θ)cos(ϕ),ny=nsin(θ)sin(ϕ),nz=ncos(θ),\begin{align*} n_x &= \vert n \vert \sin (\theta) \cos (\phi), \\ n_y &= \vert n \vert \sin (\theta) \sin (\phi), \\ n_z &= \vert n \vert \cos (\theta), \\ \end{align*}

where the norm of the Bloch vector is the taken as the radius. Each set of three coordinates (θ,ϕ,n)( \theta, \phi, \vert n \vert) represent a uniquely qubit state.

Properties

  1. n=1 \vert \bm{n} \vert = 1 if and only if ρ \rho is a pure state. If n<1 \vert \bm{n} \vert < 1 then ρ \rho is a mixed state. Hence, points on the surface of the sphere are pure states, whilst points in the interior of the sphere are mixed states. The smaller the norm of the Bloch vector, the more mixed a qubit state is.
  2. The qubit maximally mixed state, ρ=12I \rho = \frac{1}{2} \mathbb{I}, has Bloch vector nmm=(0,0,0)t\bm{n}_{mm} = (0,0,0)^{t}. With nmm=0\vert \bm{n}_{mm} \vert = 0 it can be seen that the qubit maximally mixed state sits at the centre of the unit circle.
Derivation of Bloch Sphere

Initially, pure qubit states will be considered.

Firstly, a general form for pure qubit states is found. A pure qubit state can be written with respect to the standard basis as

ψ=α0+β1,\ket{\psi} = \alpha \ket{0} + \beta \ket{1},

where α,β  C1\alpha, \beta ~\in~\mathbb{C}^{1}. By writing the complex numbers in polar coordinates, the qubit state can equivalently be written as

ψ=rαeiϕα0+rβeiϕβ1.\ket{\psi} = r_\alpha e^{i \phi_\alpha} \ket{0} + r_\beta e^{i \phi_\beta} \ket{1}.

From here, it looks as if four real numbers are needed to specify a general qubit state: rα,ϕα,rβ,ϕβr_\alpha, \phi_\alpha, r_\beta, \phi_\beta.

However, all quantum states can be multiplied by a global phase without it effecting the measurement outcomes (the density operator, which holds all information about the possible measurement outcomes, of ψ\ket{\psi} and eiωψe^{i \omega}\ket{\psi} are the same). Hence, ψ \ket{\psi} can be multiplied by eiϕαe^{-i \phi_\alpha} without loss of generality, giving

ψ=rα0+rβei(ϕβϕα)1.\ket{\psi} = r_\alpha \ket{0} + r_\beta e^{i (\phi_\beta - \phi_\alpha)} \ket{1}.

It now looks as if three real numbers are needed to specify a general qubit state: rα,rβ,(ϕβϕα)r_\alpha, r_\beta, (\phi_\beta-\phi_\alpha).

Although, valid quantum states are also normalised,

ψψ=1,\braket{\psi|\psi} = 1,

giving

rα2+rβ2=1,r_\alpha^2 + r_\beta^2 = 1,

which looks like the equation of a unit circle in 2d plane. Hence, we can let rα=cos(θ),rβ=sin(θ)r_\alpha = \cos(\theta), r_\beta = \sin(\theta) and ϕ=ϕβϕα \phi = \phi_\beta - \phi_\alpha, giving

ψ=cos(θ)0+eiϕsin(θ)1,\ket{\psi} = \cos(\theta) \ket{0} + e^{i \phi} \sin(\theta) \ket{1},

where it can now be seen that only two real numbers are need to specify a pure qubit state: θ,ϕ\theta, \phi.

However, it can be seen that if θ=0\theta = 0 then ψ=0\ket{\psi} = \ket{0} and that if θ=π/2\theta = \pi/2 then ψ=eiϕ1\ket{\psi} = e^{-i \phi} \ket{1}, where, as before, the global phase can be ignored. All qubit states can therefore be generated via 0θπ/20 \leq \theta \leq \pi/2 and 0ϕ2π0 \leq \phi \leq 2 \pi. By convention, a general pure qubit state is considered to be

ψ=cos(θ2)0+eiϕsin(θ2)1:  0θπ,  0ϕ2π.\ket{\psi} = \cos\bigg(\frac{\theta}{2}\bigg) \ket{0} + e^{i \phi} \sin\bigg(\frac{\theta}{2}\bigg) \ket{1}: ~ ~ 0 \leq \theta \leq \pi, ~ ~ 0 \leq \phi \leq 2 \pi.

The density matrix of this general pure qubit state is then found with respect to the standard basis,

ρ=ψψ=(cos2(θ2)eiϕcos(θ2)sin(θ2)eiϕcos(θ2)sin(θ2)sin2(θ2)),\begin{align*} \rho = \ket{\psi}\bra{\psi} &= \begin{pmatrix} \cos^{2}\bigg(\frac{\theta}{2}\bigg) & e^{-i \phi} \cos\bigg(\frac{\theta}{2}\bigg) \sin\bigg(\frac{\theta}{2}\bigg) \\ e^{i \phi} \cos\bigg(\frac{\theta}{2}\bigg) \sin\bigg(\frac{\theta}{2}\bigg) & \sin^2\bigg(\frac{\theta}{2}\bigg) \end{pmatrix}, \end{align*}

which, using the trigonometric double angle formula and Euler’s formula, can be rewritten as

=12 (1+cos(θ)(cos(ϕ)isin(ϕ))sin(θ)(cos(ϕ)+isin(ϕ))sin(θ)1cos(θ)).\begin{align*} = \frac{1}{2}~\begin{pmatrix} 1 + \cos(\theta) & \big( \cos(\phi) - i \sin(\phi)\big) \sin(\theta) \\ \big( \cos(\phi) + i \sin(\phi)\big) \sin(\theta) & 1 - \cos(\theta) \end{pmatrix}. \end{align*}

Finally, it can be seen that this can be written as a sum of Pauli-operators as

=12(I+(sin(θ)cos(ϕ))X+(sin(θ)sin(ϕ))Y+(cos(θ))Z),=12(I+nσ).\begin{align*} &= \frac{1}{2} \biggl( \mathbb{I} + \big(\sin(\theta) \cos(\phi) \big) X + \big(\sin(\theta) \sin(\phi) \big) Y + \big( \cos(\theta) \big) Z \biggl), \\ &= \frac{1}{2} (\mathbb{I} + \bm{n} \cdot \bm{\sigma}). \end{align*}

The qubit state is therefore uniquely specified by n\bm{n}, which can be seen to be spherical polar coordinates. For these pure qubit states, the radius is always one. Each n\bm{n} therefore specifics a point on the surface of the unit sphere - there is a one to one mapping between point on the surface of the unit sphere and pure qubit states.

To show that each Bloch vector element is given by the expectation value of the respective Pauli operator one just needs to sub in the Bloch vector form of ρ \rho into the expectation value equation,

tr[ρX]=12 tr[(I+nσ)X],=12 tr[X+nxI+nyYX+nzZX],=12 nx tr[I],=nx,\begin{align*} \textrm{tr} \big[ \rho X \big] &= \frac{1}{2}~\textrm{tr} \big[ \big( \mathbb{I} + \bm{n} \cdot \bm{\sigma} \big) X \big], \\ &= \frac{1}{2}~\textrm{tr} \big[ X + n_x \mathbb{I} + n_y YX + n_z ZX \big], \\ &= \frac{1}{2}~n_x~\textrm{tr}\big[\mathbb{I}\big], \\ &= n_x, \end{align*}

where the fact that the Pauli operators are traceless has been used.

Finally, mixed states are considered. All mixed states can be written as a mixture of pure states, which each have their own Bloch form,

ρ=ipiψiψi,=12 ipi(I+niσ),=12(I+(ipini)σ),=12(I+nσ),\begin{align*} \rho &= \sum_{i} p_i \ket{\psi_i}\bra{\psi_i}, \\ &= \frac{1}{2} ~ \sum_{i} p_i \big( \mathbb{I} + \bm{n}_i \cdot \bm{\sigma} \big), \\ &= \frac{1}{2} \big( \mathbb{I} + \bigg(\sum_{i} p_i \bm{n}_i \bigg) \cdot \bm{\sigma} \big), \\ &= \frac{1}{2} \big( \mathbb{I} + \bm{n} \cdot \bm{\sigma} \big), \end{align*}

where n\bm{n} is now take to be the mixture of Bloch vectors of each of the pure states in the description of ρ \rho . The norm of this Bloch vector is given by

n=nn,=ijpipjninj,1,\begin{align*} \vert \bm{n} \vert &= \bm{n} \cdot \bm{n}, \\ &= \sum_{i} \sum_j p_i p_j \bm{n}_i \cdot \bm{n}_j, \\ & \leq 1, \end{align*}

as ninj1  i,j \bm{n}_i \cdot \bm{n}_j \leq 1~\forall~i,j and ijpipj1\sum_{i} \sum_j p_i p_j \leq 1 as ipi=1\sum_i p_i =1, and similarly for jj. Hence, the Bloch vectors of mixed states have a norm less than one. When considered with respect to the unit sphere, this means that mixed state sit inside the sphere. As with pure qubit states, there is a one to one mapping between points inside the sphere and mixed qubit states.

Orthogonality

Given a pure quantum state ψ\ket{\psi} with Bloch vector n \bm{n} , the orthogonal state ψ \ket{\psi'}, such that ψψ=0 \braket{\psi|\psi'}=0, has Bloch vector n-\bm{n}.

Hence, orthogonal pure state correspond to the opposite point on the bloch sphere. Perpendicular Bloch vectors do not mean orthogonal quantum states.

Proof

Let

ψ=cos(θ2)0+eiϕsin(θ2)1.\ket{\psi} = \cos\bigg(\frac{\theta}{2}\bigg) \ket{0} + e^{i \phi} \sin\bigg(\frac{\theta}{2}\bigg) \ket{1}.

The state on the opposite of the Bloch sphere has

θπθ,  ϕϕ±π,\theta \rightarrow \pi - \theta, ~ ~ \phi \rightarrow \phi \pm \pi,

giving the state

ψ=cos(πθ2)0+eiϕ+iπsin(πθ2)1,=cos(πθ2)0eiϕsin(πθ2)1.\begin{align*} \ket{\psi'} &= \cos\bigg(\frac{\pi - \theta}{2}\bigg) \ket{0} + e^{i \phi + i \pi} \sin\bigg(\frac{\pi - \theta}{2}\bigg) \ket{1}, \\ &= \cos\bigg(\frac{\pi - \theta}{2}\bigg) \ket{0} - e^{i \phi} \sin\bigg(\frac{\pi - \theta}{2}\bigg) \ket{1}. \end{align*}

The inner product between these state is then

ψψ=(cos(πθ2)0eiϕsin(πθ2)1)×(cos(θ2)0+eiϕsin(θ2)1),=cos(θ2)cos(πθ2)sin(θ2)sin(πθ2),=cos(θ2)[cos(π2)cos(θ2)+sin(π2)sin(θ2)]sin(θ2)[sin(π2)cos(θ2)cos(π2)sin(θ2)],=0,\begin{align*} \braket{\psi'|\psi} = &\biggl( \cos\bigg(\frac{\pi - \theta}{2}\bigg) \bra{0} - e^{-i \phi} \sin\bigg(\frac{\pi - \theta}{2}\bigg) \bra{1} \biggl) \\ &\times \biggl( \cos\bigg(\frac{\theta}{2}\bigg) \ket{0} + e^{i \phi} \sin\bigg(\frac{\theta}{2}\bigg) \ket{1} \biggl), \\ &= \cos\bigg(\frac{\theta}{2}\bigg)\cos\bigg(\frac{\pi - \theta}{2}\bigg) - \sin\bigg(\frac{\theta}{2}\bigg) \sin\bigg(\frac{\pi - \theta}{2}\bigg), \\ &= \cos\bigg(\frac{\theta}{2}\bigg) \biggl[ \cos\bigg(\frac{\pi}{2}\bigg)\cos\bigg(\frac{\theta}{2}\bigg) + \sin\bigg(\frac{\pi}{2}\bigg)\sin\bigg(\frac{\theta}{2}\bigg) \biggl] \\ & - \sin\bigg(\frac{\theta}{2}\bigg) \biggl[ \sin\bigg(\frac{\pi}{2}\bigg)\cos\bigg(\frac{\theta}{2}\bigg) - \cos\bigg(\frac{\pi}{2}\bigg)\sin\bigg(\frac{\theta}{2}\bigg) \biggl], \\ &= 0, \end{align*}

as

sin(π2)=1  and  cos(π2)=0,cos(xy)=cos(x)cos(y)+sin(x)sin(y),sin(xy)=cos(x)sin(y)cos(y)sin(x).\begin{align*} \sin\bigg(\frac{\pi}{2}\bigg) &= 1 ~ ~ \textrm{and} ~ ~ \cos\bigg(\frac{\pi}{2}\bigg) = 0, \\ \cos(x-y) &= \cos(x)\cos(y) + \sin(x)\sin(y), \\ \sin(x-y) &= \cos(x)\sin(y) - \cos(y)\sin(x). \end{align*}

Higher Dimensions

There exists generalisations of the Bloch vector form to higher dimensional states, but the nice geometrical interpretation of points in a unit sphere is lost.

In order to get a higher dimensional Bloch vector form, an operator basis is needed. Typically, in a dd dimensional space, an operator basis is a set of operators d21d^2-1 operators {Ai}\{ A_i \} such that

  1. I{Ai}\mathbb{I} \in \{ A_i \}.
    • The set contains the identity operator
  2. tr[AiAj]=Nδij\textrm{tr} \big[ A_i^{\dagger}A_j \big] = N \delta_{ij}, where NRN \in \mathbb{R} and δij\delta_{ij} is the delta function.
    • The set of operators is orthogonal according to the Hilbert-Schmit inner product.

In the qubit case, these set of operators were the identity and Pauli operators, {I,X,Y,Z}\{ \mathbb{I}, X, Y, Z \}.

Given {Ai}\{ A_i \} forms a basis, any matrix can be decomposed into {Ai}\{ A_i \}. For a density operators ρHd \rho \in \mathcal{H}^{d} , it’s density matrix with respect to {Ai}\{ A_i \} can be written as

ρ=1d(I+bΣ),\rho = \frac{1}{d} \bigg( \mathbb{I} + \bm{b} \cdot \bm{\Sigma} \bigg),

where bRd21\bm{b} \in \mathbb{R}^{d^2-1} and Σ \bm{\Sigma} is a vector whose elements are the operators in the basis {Ai}\{ A_i \}. Hence, ρ \rho is written as linear combination of operators in the basis.

As before, the 1dI\frac{1}{d} \mathbb{I} term fixes the normalisation, such that tr[ρ]=1\textrm{tr}\big[ \rho \big] = 1, and the components of the Bloch vector can be found as bi=tr[ρAi]b_i = \textrm{tr} \big[ \rho A_i \big].

However, not all Bloch vectors, b\bm{b}, correspond to valid density operators, as they did in the qubit case.

There are multiple different operator basis one could choose and hence no unique Bloch vector for a given state. See the following for more details on different operator basis one could choose: Bertlmann & Krammer (2008), Asadian et al. (2016).

Dynamics

Unitaries

Unitary operators cause rotations of points on the Bloch sphere.

The Pauli operators, {X,Y,Z}\{ X, Y, Z \}, give rise to rotations around the x,yx, y, and zz axis respectively when exponentiated,

Rx(Θ)=eiΘX/2=cos(Θ2)Iisin(Θ2)X,Ry(Θ)=eiΘY/2=cos(Θ2)Iisin(Θ2)Y,Rz(Θ)=eiΘZ/2=cos(Θ2)Iisin(Θ2)Z.\begin{align*} R_x(\Theta) &= e^{-i \Theta X/2} = \cos \bigg(\frac{\Theta}{2} \bigg)\mathbb{I} - i \sin \bigg( \frac{\Theta}{2} \bigg) X , \\ R_y(\Theta) &= e^{-i \Theta Y/2} = \cos \bigg(\frac{\Theta}{2} \bigg)\mathbb{I} - i \sin \bigg( \frac{\Theta}{2} \bigg) Y, \\ R_z(\Theta) &= e^{-i \Theta Z/2} = \cos \bigg(\frac{\Theta}{2} \bigg)\mathbb{I} - i \sin \bigg( \frac{\Theta}{2} \bigg) Z. \end{align*}
Proof

The exponential of an operator XX is given by

eX=k=0Xkk!=I+X+X22!+X33!+e^{X} = \sum_{k=0}^{\infty} \frac{X^k}{k!} = \mathbb{I} + X + \frac{X^2}{2!} + \frac{X^3}{3!} + \ldots

Therefore,

eiΘA=I+iΘA+(iΘA)22!+(iΘA)33!+(iΘA)44!+e^{i\Theta A} = \mathbb{I} + i\Theta A + \frac{\big(i\Theta A\big)^2}{2!} + \frac{\big(i\Theta A \big)^3}{3!} + \frac{\big(i\Theta A \big)^4}{4!} + \ldots

If A2=IA^2 = \mathbb{I} then

eiΘA=I+iΘAΘ22!IiΘ3A3!+Θ44!+,=(1Θ22!+Θ44!+)I+i(1Θ33!+)A,=cos(Θ)I+isin(Θ)A,\begin{align*} e^{i\Theta A} &= \mathbb{I} + i\Theta A - \frac{\Theta^2}{2!} \mathbb{I} - i \frac{\Theta^3 A}{3!} + \frac{\Theta^4}{4!} + \ldots, \\ &= \biggl( 1 - \frac{\Theta^2}{2!} + \frac{\Theta^4}{4!} + \ldots \biggl) \mathbb{I} + i \biggl(1 - \frac{\Theta^3}{3!} + \ldots \biggl) A, \\ &= \cos\big(\Theta\big) \mathbb{I} + i \sin\big( \Theta \big) A, \end{align*}

where the taylor series of cos()\cos(\cdot) and sin()\sin(\cdot) have been used. Taking ΘΘ/2\Theta \rightarrow -\Theta/2 and noting that X2=Y2=Z2=IX^2=Y^2=Z^2=\mathbb{I} completes the proof.

General rotations around an axis defined by the Bloch vector n=(nx,ny,nz)t\bm{n} = (n_x, n_y, n_z )^{t} are given by

Rn(Θ)=eiΘnσ/2,=cos(Θ2)Iisin(Θ2)(nxX+nyY+nzZ),\begin{align*} R_{\bm{n}}(\Theta) &= e^{-i \Theta \bm{n} \cdot \bm{\sigma}/2}, \\ &= \cos \bigg(\frac{\Theta}{2} \bigg)\mathbb{I} - i \sin \bigg( \frac{\Theta}{2} \bigg) \big( n_x X + n_y Y + n_z Z \big), \end{align*}

as (nσ)2=I\big(\bm{n} \cdot \bm{\sigma} \big)^2=\mathbb{I}.

Using this general rotation, any single qubit unitary can be created by adding an additional phase factor,

U=eiaRn(Θ),  UU=UU=I,U = e^{-i a} R_{\bm{n}}(\Theta), ~ ~ UU^{\dagger}=U^{\dagger}U=\mathbb{I},

where a,ΘR1a, \Theta \in \mathbb{R}^{1}.

Channels

Quantum channels that are not unitary will act to take points on the surface of the Bloch sphere into the interior - pure states are taken to mixed states.

More generally, the set of output states from a channel can be visualised on the Bloch sphere. Consider the quantum depolarising channel,

Dppol(ρ)=pρ+(1p)I/2,\mathcal{D}^{\textrm{pol}}_p(\rho) = p \rho + (1-p)\mathbb{I}/2,

which can be interpreted as returning the state unchanged with probability pp, and returning noise with probability 1p1-p. The action of this channel is a uniform contraction of the Bloch sphere, with the radius of the new sphere depending on pp. If p=0p=0, the set of possible output states becomes just a point as the channel will output a single state - the maximally mixed state - for all input states.

Consider instead the dephasing channel,

Dpph(ρ)=pρ+(1p)nnnρnn,\mathcal{D}^{\textrm{ph}}_p(\rho) = p \rho + (1-p) \sum_n \ket{n}\bra{n} \rho \ket{n}\bra{n},

which acts to dampen the off-diagonal terms of the density matrix with respect to some basis. If the off-diagonal terms with respect to the standard basis are dampened, the set of possible output states is an ellipsoid, a sphere contracted along the xx and yy axis wih the complete zz axis contained. This highlights the fact that any state that is diagonal in the standard basis is left unchanged by the dephasing channel. If the dephasing is performed with respect to a different basis, then the ellipsoid with be rotated compared to the above.

References
  1. Bertlmann, R. A., & Krammer, P. (2008). Bloch vectors for qudits. Journal of Physics A: Mathematical and Theoretical, 41(23), 235303. 10.1088/1751-8113/41/23/235303
  2. Asadian, A., Erker, P., Huber, M., & Klöckl, C. (2016). Heisenberg-Weyl Observables: Bloch vectors in phase space. Physical Review A, 94(1). 10.1103/physreva.94.010301