Controlled-U Powers
Quantum phase estimation (QPE) extracts the eigenphase of a unitary by entangling a register of control qubits with repeated applications of . Each control qubit (indexed ) is coupled to the target through the powered unitary . Understanding how to build and apply these controlled powers is the central programming skill of QPE.
Eigenphases and powers
Suppose where is the unknown phase. Applying times accumulates the phase multiplicatively:
This is why QPE uses : control qubit picks up exactly the bit of the binary expansion through the phase kickback mechanism.
Building in a circuit
The most direct way to realize is to apply exactly times in sequence. Each application is an independent gate invocation on the same wires. For the controlled version, every one of those applications is conditioned on the same control qubit:
Concretely, if your controlled gate is c.cp(lambda, ctrl, target), then
controlled-U^m is obtained by calling that line times with the same arguments.
Example:
The gate is the quarter-turn phase gate , with matrix
Squaring it:
So . You can verify this by applying the controlled- gate twice — the phase angles add: followed by gives .
In circuit notation with qubit 0 as the control and qubit 1 as the target, c.cp(pi/2, 0, 1) is
the controlled- gate, and calling it twice implements controlled-.
Try it
Build the controlled- operator on 2 qubits by applying the controlled- gate twice. The grader checks the full unitary matrix, so an empty circuit (or a single application) will fail.
Sign in on the full site to ask questions and join the discussion.