Calling the same quantum kernel in cudaq.control and cudaq.adjoint in the main kernel give error.

Required prerequisites

[X] Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
[X] Make sure you've read the documentation. Your issue may be addressed there.
[X] Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.
[ ] If possible, make a PR with a failing test to give us a starting point to work on!

Describe the bug

If I run the code in the example 1 below with both

cudaq.control(my_func,ancilla,q,theta)
cudaq.adjoint(my_func,q,theta)

I got this error: error: 'func.call' op '__nvqpp__mlirgen__my_func.adj' does not reference a valid function

However if I comment one of them, for example:

cudaq.control(my_func,ancilla,q,theta)
#cudaq.adjoint(my_func,q,theta)

It works.

In Example 2, I am trying first to apply a cudaq.control on a given quantum kernel and then apply control-adjoint of the quantum kernel, but it gives error. error: 'func.call' op '__nvqpp__mlirgen__my_func.adj.ctrl' does not reference a valid function

A work around it is shown in example 3. Example 3 works.

This does not make sense. Because, I have to explicitly rewrite the kernel using adj() for each applied gate instead of using the cudaq.adjoint() and then cudaq.control().

Steps to reproduce the bug

Example 1:

import cudaq

@cudaq.kernel
def my_func(q:cudaq.qubit, theta:float):
    ry(theta,q)
    rz(theta,q)

@cudaq.kernel
def kernel(theta:float):
    ancilla=cudaq.qubit()
    q=cudaq.qubit()

    h(ancilla)
    cudaq.control(my_func,ancilla,q,theta)
    cudaq.adjoint(my_func,q,theta)

theta=1.5
count=cudaq.sample(kernel,theta)
print(count)

Example 2:

import cudaq

@cudaq.kernel
def my_func(q:cudaq.qubit, theta:float):
    ry(theta,q)
    rz(theta,q)

@cudaq.kernel
def adj_func(q:cudaq.qubit, theta:float):
    cudaq.adjoint(my_func,q,theta)

@cudaq.kernel
def kernel(theta:float):
    ancilla=cudaq.qubit()
    q=cudaq.qubit()

    h(ancilla)
    cudaq.control(my_func,ancilla,q,theta)
    cudaq.control(adj_func,ancilla,q,theta)

theta=1.5
count=cudaq.sample(kernel,theta)
print(count)

Example 3:

import cudaq

@cudaq.kernel
def my_func(q:cudaq.qubit, theta:float):
    ry(theta,q)
    rz(theta,q)

@cudaq.kernel
def adj_func(q:cudaq.qubit, theta:float):
    #cudaq.adjoint(my_func,q,theta)
    ry.adj(theta,q)
    rz.adj(theta,q)

@cudaq.kernel
def kernel(theta:float):
    ancilla=cudaq.qubit()
    q=cudaq.qubit()

    h(ancilla)
    cudaq.control(my_func,ancilla,q,theta)
    cudaq.control(adj_func,ancilla,q,theta)

theta=1.5
count=cudaq.sample(kernel,theta)
print(count)

Expected behavior

I expect to be able to call the same quantum kernel and use it with both cudaq.control() and cudaq.adjoint().

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

CUDA Quantum version: latest
Python version: 3.10
C++ compiler:
Operating system: linux

Suggestions

No response

NVIDIA / cuda-quantum