Checkpoint: Estimate Simulability
This checkpoint pulls the module together: given the entanglement structure of a state, estimate the bond dimension an exact matrix product state would need — the single number that decides whether a classical tensor-network simulation is feasible.
The rule you will apply
The exact MPS bond dimension across a cut equals the Schmidt rank at that cut: the number of non-zero Schmidt coefficients when you write
A coefficient that is (numerically) zero contributes nothing and does not count. The overall bond dimension a faithful MPS needs is the largest Schmidt rank over all cuts of the chain. If that number stays small as the system grows, the state is efficiently simulable; if it grows exponentially, it is not.
The state in this problem
Consider a 4-qubit state cut down the middle, into the left pair and the right pair . Suppose its Schmidt decomposition across that cut has the squared Schmidt coefficients (the eigenvalues of the reduced density matrix)
Two of the four possible Schmidt terms carry weight; the other two vanish. So the Schmidt rank — and hence the exact bond dimension across this cut — is . The entanglement entropy is bit, consistent with .
Your task
The reduced-density-matrix eigenvalues across the middle cut are given to you as an array. Count how many are non-negligible (above a tolerance such as ) and return that count — the exact bond dimension for this cut. For the spectrum above the answer is .
Try it
Compute and return the bond dimension. The grader checks the returned number.
After running, you should return the integer : a state with two non-zero Schmidt coefficients across a cut needs an MPS bond dimension of exactly there.
Sign in on the full site to ask questions and join the discussion.