Skip to content

Kraus Operators

Kraus operators represent quantum operations on density operators. In measurement theory, they describe how the state changes for an outcome or after averaging over outcomes. They are closely related to POVM effects, but they are not the same object.

The most common confusion is:

Kraus operators transform states;effects determine probabilities.\text{Kraus operators transform states;} \qquad \text{effects determine probabilities.}

A finite-dimensional quantum operation has the form

E(ρ)=∑αKαρKα†.\mathcal E(\rho) = \sum_\alpha K_\alpha\rho K_\alpha^\dagger.

The operators KαK_\alpha are Kraus operators. The map E\mathcal E is completely positive by construction.

The trace of the output is

Tr⁡E(ρ)=Tr⁡(ρ∑αKα†Kα).\operatorname{Tr}\mathcal E(\rho) = \operatorname{Tr} \left( \rho\sum_\alpha K_\alpha^\dagger K_\alpha \right).

If the operation is trace preserving, then

∑αKα†Kα=I.\sum_\alpha K_\alpha^\dagger K_\alpha=I.

If the operation is a single outcome branch of a measurement, it is usually trace nonincreasing:

∑αKα†Kα≤I.\sum_\alpha K_\alpha^\dagger K_\alpha\le I.

The missing trace is not lost probability. It is the probability that some other outcome occurred.

For a measurement with outcomes mm, each outcome can be assigned an operation

Im(ρ)=∑αKmαρKmα†.\mathcal I_m(\rho) = \sum_\alpha K_{m\alpha}\rho K_{m\alpha}^\dagger.

The probability of outcome mm is

p(m)=Tr⁡Im(ρ).p(m) = \operatorname{Tr}\mathcal I_m(\rho).

If p(m)≠0p(m)\ne0, the conditional output state is

ρm=Im(ρ)Tr⁡Im(ρ)=∑αKmαρKmα†p(m).\rho_m = \frac{\mathcal I_m(\rho)} {\operatorname{Tr}\mathcal I_m(\rho)} = \frac{ \sum_\alpha K_{m\alpha}\rho K_{m\alpha}^\dagger }{ p(m) }.

The nonselective output is obtained by summing over outcomes:

ρ′=∑mIm(ρ)=∑m,αKmαρKmα†.\rho' = \sum_m\mathcal I_m(\rho) = \sum_{m,\alpha} K_{m\alpha}\rho K_{m\alpha}^\dagger.

If all outcomes are included, the total operation is trace preserving:

∑m,αKmα†Kmα=I.\sum_{m,\alpha} K_{m\alpha}^\dagger K_{m\alpha}=I.

The POVM effect associated with outcome mm is

Fm=∑αKmα†Kmα.F_m = \sum_\alpha K_{m\alpha}^\dagger K_{m\alpha}.

Then

p(m)=Tr⁡(ρFm).p(m) = \operatorname{Tr}(\rho F_m).

This formula explains the relation:

Kraus operators -> operation and state update
Kraus adjoint-products -> POVM effects and probabilities

The arrow goes from a chosen operation to its effect. It does not go uniquely backward from an effect to an operation.

If outcome mm has one measurement operator MmM_m, then

Im(ρ)=MmρMm†,Fm=Mm†Mm.\mathcal I_m(\rho)=M_m\rho M_m^\dagger, \qquad F_m=M_m^\dagger M_m.

The conditional state is

ρm=MmρMm†Tr⁡(MmρMm†).\rho_m = \frac{M_m\rho M_m^\dagger} {\operatorname{Tr}(M_m\rho M_m^\dagger)}.

This common case is useful pedagogically, but realistic detector outcomes often group many microscopic alternatives into one reported outcome. Then several Kraus operators may be needed for a single mm.

An ideal projective measurement is recovered by taking

Ka=Pa.K_a=P_a.

Then

Ia(ρ)=PaρPa,Fa=Pa†Pa=Pa.\mathcal I_a(\rho)=P_a\rho P_a, \qquad F_a=P_a^\dagger P_a=P_a.

The nonselective channel is

ρ′=∑aPaρPa.\rho' = \sum_a P_a\rho P_a.

Projective measurement is therefore not separate from the Kraus formalism. It is the sharp, orthogonal special case.

For a qubit detector intended to click on ∣1⟩|1\rangle with efficiency η\eta, one outcome model is

Kclick=η ∣1⟩⟨1∣.K_{\mathrm{click}} = \sqrt{\eta}\,|1\rangle\langle1|.

The no-click outcome can have two Kraus operators:

Kno,0=∣0⟩⟨0∣,Kno,1=1−η ∣1⟩⟨1∣.K_{\mathrm{no},0} = |0\rangle\langle0|, \qquad K_{\mathrm{no},1} = \sqrt{1-\eta}\,|1\rangle\langle1|.

The effects are

Fclick=η∣1⟩⟨1∣,F_{\mathrm{click}} = \eta|1\rangle\langle1|, Fno=∣0⟩⟨0∣+(1−η)∣1⟩⟨1∣.F_{\mathrm{no}} = |0\rangle\langle0| + (1-\eta)|1\rangle\langle1|.

They sum to II. The two no-click Kraus operators encode two different microscopic histories: the system was in ∣0⟩|0\rangle, or it was in ∣1⟩|1\rangle but the detector missed it. If the detector does not record which history occurred, those histories are grouped into the same reported outcome.

Amplitude damping of a two-level system can be written with

K0=∣0⟩⟨0∣+1−γ ∣1⟩⟨1∣,K_0 = |0\rangle\langle0| + \sqrt{1-\gamma}\,|1\rangle\langle1|, K1=γ ∣0⟩⟨1∣.K_1 = \sqrt{\gamma}\,|0\rangle\langle1|.

The completeness relation is

K0†K0+K1†K1=I.K_0^\dagger K_0+K_1^\dagger K_1=I.

If a photon-emission record is monitored, K1K_1 can represent a jump outcome and K0K_0 a no-jump outcome over the time step. If the record is ignored, the channel is

Φ(ρ)=K0ρK0†+K1ρK1†.\Phi(\rho) = K_0\rho K_0^\dagger + K_1\rho K_1^\dagger.

The same Kraus formula can therefore describe a conditioned measurement update or an unconditional noise channel, depending on whether the record is retained.

A Kraus representation is not unique. If

E(ρ)=∑αKαρKα†,\mathcal E(\rho) = \sum_\alpha K_\alpha\rho K_\alpha^\dagger,

and uβαu_{\beta\alpha} is a unitary matrix mixing the Kraus labels, then

Lβ=∑αuβαKαL_\beta = \sum_\alpha u_{\beta\alpha}K_\alpha

gives another representation of the same operation:

E(ρ)=∑βLβρLβ†.\mathcal E(\rho) = \sum_\beta L_\beta\rho L_\beta^\dagger.

Therefore individual Kraus operators are often representation-dependent. Physical meaning attaches to them only when a measurement record, detector history, environment basis, or unraveling has been specified.

This warning is essential in open-system work. The same channel may have many Kraus decompositions, but not every decomposition corresponds to the same observed measurement record.

The trace condition tells what kind of operation is being represented:

ConditionMeaning
∑αKα†Kα=I\sum_\alpha K_\alpha^\dagger K_\alpha=Itrace-preserving channel or complete nonselective operation
∑αKα†Kα≤I\sum_\alpha K_\alpha^\dagger K_\alpha\le Itrace-nonincreasing operation, usually a selected branch
∑m,αKmα†Kmα=I\sum_{m,\alpha}K_{m\alpha}^\dagger K_{m\alpha}=Icomplete measurement instrument with all outcomes included

If the trace condition fails, either the model is incomplete or the listed operators are not a physical operation.

Kraus operators are a representation of completely positive maps. A quantum instrument is the outcome-indexed collection of those maps:

{Im}m.\{\mathcal I_m\}_m.

The instrument is the operational object. A Kraus representation is one way to write each Im\mathcal I_m. This distinction matters because Kraus representations can change while the instrument remains the same.

For calculations, Kraus operators are indispensable. For specifying the measurement at an invariant level, the instrument is the cleaner object.

  • Confusing KmK_m with the effect Fm=Km†KmF_m=K_m^\dagger K_m.
  • Assuming there is only one Kraus operator per outcome.
  • Forgetting that trace-nonincreasing operations must be normalized after conditioning.
  • Treating a single selected outcome operation as a trace-preserving channel.
  • Reading physical detector histories from a Kraus representation without checking how the record is defined.
  • Forgetting that different Kraus representations can give the same operation.
  • Assuming every decomposition of a channel corresponds to a actually monitored trajectory.
  • K. Kraus, States, Effects, and Operations: Fundamental Notions of Quantum Theory, Springer, 1983.
  • A. S. Holevo, Probabilistic and Statistical Aspects of Quantum Theory, Edizioni della Normale, 2011.
  • M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press, 2010.
  • H. M. Wiseman and G. J. Milburn, Quantum Measurement and Control, Cambridge University Press, 2010.
  • H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems, Oxford University Press, 2002.
  1. Let K0=p IK_0=\sqrt{p}\,I and K1=1−p IK_1=\sqrt{1-p}\,I, with 0≤p≤10\le p\le1. Show that these define a trace-preserving operation and describe the effect of the nonselective channel.
Solution

The completeness relation is

K0†K0+K1†K1=pI+(1−p)I=I.K_0^\dagger K_0+K_1^\dagger K_1 = pI+(1-p)I = I.

The nonselective channel is

Φ(ρ)=pρ+(1−p)ρ=ρ.\Phi(\rho) = p\rho+(1-p)\rho = \rho.

The outcome label is a classical random label independent of the state; ignoring it leaves the state unchanged.

  1. For the amplitude-damping Kraus operators
K0=∣0⟩⟨0∣+1−γ∣1⟩⟨1∣,K1=γ∣0⟩⟨1∣,K_0=|0\rangle\langle0|+\sqrt{1-\gamma}|1\rangle\langle1|, \qquad K_1=\sqrt{\gamma}|0\rangle\langle1|,

verify the completeness relation.

Solution

Compute

K0†K0=∣0⟩⟨0∣+(1−γ)∣1⟩⟨1∣,K_0^\dagger K_0 = |0\rangle\langle0| + (1-\gamma)|1\rangle\langle1|,

and

K1†K1=γ∣1⟩⟨1∣.K_1^\dagger K_1 = \gamma |1\rangle\langle1|.

Adding gives

K0†K0+K1†K1=∣0⟩⟨0∣+∣1⟩⟨1∣=I.K_0^\dagger K_0+K_1^\dagger K_1 = |0\rangle\langle0|+|1\rangle\langle1| =I.
  1. If one outcome has Kraus operators Km1K_{m1} and Km2K_{m2}, what is the associated POVM effect?
Solution

The effect is

Fm=Km1†Km1+Km2†Km2.F_m = K_{m1}^\dagger K_{m1} + K_{m2}^\dagger K_{m2}.

The probability is p(m)=Tr⁡(ρFm)p(m)=\operatorname{Tr}(\rho F_m).

  1. Suppose L0=(K0+K1)/2L_0=(K_0+K_1)/\sqrt2 and L1=(K0−K1)/2L_1=(K_0-K_1)/\sqrt2. Show that {L0,L1}\{L_0,L_1\} gives the same operation as {K0,K1}\{K_0,K_1\}.
Solution

Expand:

L0ρL0†+L1ρL1†=12(K0+K1)ρ(K0†+K1†)+12(K0−K1)ρ(K0†−K1†).L_0\rho L_0^\dagger+L_1\rho L_1^\dagger = \frac{1}{2} (K_0+K_1)\rho(K_0^\dagger+K_1^\dagger) + \frac{1}{2} (K_0-K_1)\rho(K_0^\dagger-K_1^\dagger).

The cross terms cancel, leaving

K0ρK0†+K1ρK1†.K_0\rho K_0^\dagger+K_1\rho K_1^\dagger.

This is a unitary mixing of Kraus labels, so it does not change the operation.