Controlled-unitary (or controlled-$U$) gate applies a unitary operation $U$ to target qubits if and only if all control qubits are $|1\rangle$. It generalizes controlled single-qubit gates (like CNOT) and is fundamental to quantum algorithms including phase estimation, Shor's algorithm, and variational quantum algorithms.

For a unitary $U$ acting on $m$ target qubits and $k$ control qubits, the controlled-$U$ gate is:

$$C^k(U) = |0\rangle\langle 0|_c \otimes I_t + |1\rangle\langle 1|_c \otimes U$$

for single control qubit, generalizing to multiple controls:

$$C^{c_1 \cdots c_k}(U) = \sum_{x \in \{0,1\}^k} |x\rangle\langle x|_c \otimes (U^{\delta(x)} \otimes I^{1-\delta(x)})$$

where $\delta(x) = 1$ if all controls are $|1\rangle$, else $\delta(x) = 0$.

Single control, single target:

• If control is $|0\rangle$: apply identity to target
• If control is $|1\rangle$: apply $U$ to target

Multiple controls:

• If any control is $|0\rangle$: apply identity to targets
• If all controls are $|1\rangle$: apply $U$ to targets

Controlled-X (CNOT):

• $U = X$ (Pauli-X)
• Flips target if control is $|1\rangle$
• One of the most frequently used two-qubit gates

Controlled-Z:

• $U = Z$ (Pauli-Z)
• Applies phase $-1$ if both qubits are $|1\rangle$

Controlled-phase ($e^{i\theta}$):

• $U = e^{i\theta} I$ (global phase)
• Applies phase to target if control is $|1\rangle$

Controlled-rotation (Controlled-$R_z(\theta)$):

• $U = R_z(\theta) = e^{-i\theta Z/2}$
• Rotates target qubit around Z axis if control is $|1\rangle$

Doubly-controlled-X (Toffoli/CCX):

• $U = X$, with two control qubits
• Flips target if both controls are $|1\rangle$

For controlled-$U$ with single control and single target:

$$C(U) = \begin{pmatrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & u_{00} & u_{01} \\ 0 & 0 & u_{10} & u_{11} \end{pmatrix}$$

where $U = \begin{pmatrix} u_{00} & u_{01} \\ u_{10} & u_{11} \end{pmatrix}$.

For arbitrary unitary $U$, decompose using Euler angle decomposition:

$$U = e^{i\alpha} A X B X C$$

where $ABC = I$ and $A, B, C$ are single-qubit unitaries. Then:

$$C(U) = e^{i\alpha} A \cdot C_X(B) \cdot C_X(C)$$

using controlled-single-qubit gates (easier to implement).

Gate count for controlled-$U$:

• Single-qubit rotations: $O(1)$
• CNOT gates: $O(1)$ to $O(n)$ depending on $U$ and connectivity

For $k$ control qubits and single target, implement controlled-controlled-…-controlled-$U$:

Naive approach: cascade controls

• Total depth: $O(k)$
• Requires ancilla qubits for intermediate phases
• Gate count: $O(k) \times \text{cost}(U)$

Optimized approach (e.g., Toffoli-based):

• Decompose into Toffoli chains
• Requires $O(k)$ Toffoli gates + $O(k)$ cleanup
• Depth: $O(k)$ with careful scheduling

Phase estimation:

• Apply controlled-$U^{2^j}$ for $j = 0, 1, \ldots, n-1$
• Extract eigenvalues via inverse Fourier transform
• Controlled-unitary is essential subroutine

Shor's factoring algorithm:

• Compute controlled modular exponentiation: $|x\rangle \to |x^a \bmod N\rangle$
• Find period via phase estimation
• Controlled-unitary enables quantum speedup

Variational quantum algorithms (VQE, QAOA):

• Construct parameterized ansätze with controlled gates
• Hybrid classical-quantum optimization

Quantum simulation:

• Simulate time evolution: $e^{-iHt}$
• Use Trotter-Suzuki decomposition with controlled-$e^{-iH_j t}$ terms

Controlled-$R_x(\theta)$, Controlled-$R_y(\theta)$, Controlled-$R_z(\theta)$:

• Rotate target qubit around axis if control is $|1\rangle$
• Parametric angle $\theta$ enables tunable interactions
• Used in QAOA and variational circuits

For $U$ acting on multiple target qubits with single control:

$$C(U) = |0\rangle\langle 0|_c \otimes I_t + |1\rangle\langle 1|_c \otimes U_t$$

The unitary $U_t$ acts on all targets simultaneously. Example: controlled-SWAP (Fredkin gate).

• Gate depth: $O(2^k)$ for $k$ control qubits (exponential blowup)
• Ancilla overhead: intermediate states require scratch qubits
• Error accumulation: each added control increases error
• Practical limit: current devices limited to 3–4 control qubits efficiently

Approximate controlled-$U$: drop least significant gates to reduce depth

• Trade accuracy for circuit size
• Useful for near-term devices

• CNOT: simplest controlled gate ($U = X$)
• Toffoli/CCX: doubly-controlled-X
• Fredkin/CSWAP: controlled-SWAP
• Quantum Fourier Transform: uses controlled-rotations
• Phase oracle: marks solutions via controlled-phase