What is a qalgora-Q Kernel?

A kernel is a function that runs on a quantum backend. It is the unit of quantum computation in qalgora-Q.

Defining a kernel

In the full specification, the function body is converted to qalgora IR and dispatched to the target backend; the open reference build mainly converts the core gate-model kernel into a circuit representation executable by the CPU statevector simulator.

import qalgora

@qalgora.kernel
def my_kernel(theta: float):
    q = qalgora.qubit()
    ry(theta, q)
    mz(q)

The gate names ry and mz are auto-injected into the kernel scope by @qalgora.kernel, which is why they appear without a qalgora. prefix.

Allocating qubits

• qalgora.qubit() — a single qubit.
• qalgora.qvector(n) — a register of n qubits, indexed q[i].
• qalgora.qview — a non-owning reference passed between kernels.

Qubits start in the |0⟩ state and are automatically released at the end of the kernel.

The type system

Rules & restrictions

• Kernel arguments must be typed — annotate every one (int, float, list[float], qalgora.qview…).
• Only quantum intrinsics and a restricted classical subset (typed loops, conditionals on measured bits, arithmetic) are allowed inside a kernel.
• Unsupported: Python comprehensions over qubits, exceptions, dynamic typing, most library calls.
• Kernels can call other kernels, enabling modular circuit construction.

Execution model

1. In the full specification, the decorated function is converted to qalgora IR (an MLIR dialect); the open reference build converts the core gate-model kernel to an executable circuit representation.
2. The representation is optimized and lowered (下沉) to the selected target — today the CPU statevector simulator.
3. A primitive (sample/observe) executes it and returns classical results.