Building your first CUDA-Q Program

Python

We can define our quantum kernel as a typical Python function, with the additional use of the @cudaq.kernel decorator. Let’s begin with a simple GHZ-state example, producing a state of maximal entanglement amongst an allocated set of qubits.

import cudaq


# Define our kernel.
@cudaq.kernel
def kernel(qubit_count: int):
    # Allocate our qubits.
    qvector = cudaq.qvector(qubit_count)
    # Place the first qubit in the superposition state.
    h(qvector[0])
    # Loop through the allocated qubits and apply controlled-X,
    # or CNOT, operations between them.
    for qubit in range(qubit_count - 1):
        x.ctrl(qvector[qubit], qvector[qubit + 1])
    # Measure the qubits.
    mz(qvector)

C++

We can define our quantum kernel as we do any other typed callable in C++, through the use of the __qpu__ annotation. For the following example, we will define a kernel for a simple GHZ-state as a standard free function.

#include <cudaq.h>
#include <iostream>

__qpu__ void kernel(int qubit_count) {
  // Allocate our qubits.
  cudaq::qvector qvector(qubit_count);
  // Place the first qubit in the superposition state.
  h(qvector[0]);
  // Loop through the allocated qubits and apply controlled-X,
  // or CNOT, operations between them.
  for (auto qubit : cudaq::range(qubit_count - 1)) {
    x<cudaq::ctrl>(qvector[qubit], qvector[qubit + 1]);
  }
  // Measure the qubits.
  mz(qvector);
}

This kernel function can accept any number of arguments, allowing for flexibility in the construction of the quantum program. In this case, the qubit_count argument allows us to dynamically control the number of qubits allocated to the kernel. As we will see in further examples, we could also use these arguments to control various parameters of the gates themselves, such as rotation angles.