Entanglement Entropy
The entanglement entropy of a bipartite pure state is the von Neumann entropy of a reduced state, $S(\rho_A)=-\mathrm{Tr}\,\rho_A\log_2\rho_A$; it is zero for product states and 1 bit for Bell states.
For a pure state $|\psi\rangle_{AB}$ we take the partial trace over $B$ to get $\rho_A=\mathrm{Tr}_B|\psi\rangle\langle\psi|$. The Von Neumann Entropy is $$S(\rho_A)=-\mathrm{Tr}\,\rho_A\log_2\rho_A=-\sum_k\lambda_k\log_2\lambda_k,$$ where $\lambda_k$ are the eigenvalues of $\rho_A$. For pure states $S(\rho_A)=S(\rho_B)$ (from the Schmidt decomposition), and this value is an established measure of entanglement: it is zero if and only if the state is a product state, and its maximum is $\log_2d$, where $d$ is the dimension of the smaller subsystem.
Schmidt Decomposition: $|\psi\rangle=\sum_k\sqrt{\lambda_k}|a_k\rangle|b_k\rangle$, and the coefficients $\lambda_k$ are exactly the eigenvalues of the two reduced states. The number of non-zero coefficients (the Schmidt rank) is greater than 1 for entangled states.
Caution: for mixed states the subsystem entropy is not a valid entanglement measure, because classical mixing also raises it; other measures are used instead.
In simulation, entanglement entropy is linked to how hard a state is to simulate classically with tensor-network methods: states with low entanglement across every bipartition can be represented efficiently, while high entanglement requires resources that grow exponentially. Low entanglement is not, however, the only condition for efficient simulation.
Mathematical formula
Worked example
For the state $|\psi\rangle=\cos\alpha|00\rangle+\sin\alpha|11\rangle$: $\rho_A=\mathrm{diag}(\cos^2\alpha,\sin^2\alpha)$. At $\alpha=\pi/4$ (a Bell state): $S=-2\cdot\tfrac12\log_2\tfrac12=1$ bit. At $\alpha=\pi/8$: $\lambda=(0.854,0.146)$ and $S\approx-0.854\log_20.854-0.146\log_20.146\approx0.194+0.405=0.60$ bits. At $\alpha=0$: $S=0$.
Test your understanding
The quiz needs JavaScript.
Related topics
References
- Michael A. Nielsen; Isaac L. Chuang (2010). Quantum Computation and Quantum Information (10th Anniversary Edition). Cambridge University Press doi:10.1017/CBO9780511976667The standard reference for quantum computing; the Crossref record gives 2012 for the Cambridge Core online edition.
