TypeError: ParameterExpression with unbound parameters ({Parameter(w_112)}) cannot be cast to a float.

[dhamotharnaf]

Environment

• Qiskit version: 0.44.1
• Python version: 3.10.12
• Operating system: Linux (Distro-Ubuntu)

What is happening?

I have a variational circuit, and I'm also employing the MCMT method to construct a generalized circuit. My circuit is dependent on an input Hermitian matrix. However, when I attempt to work with a larger-sized matrix, I encounter a significant increase in the number of variational parameters. Consequently, I face an issue where the parameters cannot be cast to a float. Strangely, this problem is not present for smaller parameter sets. I suspect that the problem might be related to the complex numbers I am using. I've been attempting to resolve this issue for a while. I'll attach a screenshot of the error for reference.

How can we reproduce the issue?

    def make_Ai(self, key, control_list):
        ''' Applies gates coherently using multi-controlled gates
            Args : key --> Gate name
            control_list --> index of ancilla qubits to be used as control'''
        Ai = QuantumCircuit(self.tot_qubits)
        key = key[::-1]
        for _ in range(len(key)):
            if key[_] == 'I':
                continue
            '''
            MCMT is multi controlled multi target gate. Here we define MCX, MCY, MCZ from base X,Y Z gates
            '''
            if key[_] == 'X':
                MCX = MCMT(Ai.x, len(control_list), 1)
                apply_bits = control_list + [self.ancilla_size + _]
                Ai.append(MCX, apply_bits)

            if key[_] == 'Z':
                MCZ = MCMT(Ai.z, len(control_list), 1)
                apply_bits = control_list + [self.ancilla_size + _]
                Ai.append(MCZ, apply_bits)

            if key[_] == 'Y':
                MCY = MCMT(Ai.y, len(control_list), 1)
                apply_bits = control_list + [self.ancilla_size + _]
                Ai.append(MCY, apply_bits)
        return Ai.decompose()
    def var_circuit(self, parameters):
        '''
        Creates a variational quantum circuit.
        This is analogous to CA function but here we apply multi controlled ry gates coherently

        Args : parameters --> variable names for rotation gates

        '''
        vc = QuantumCircuit(self.tot_qubits, name='var_circuit')
        for idx in range(1, len(self.keys) + 1):
            control_bin = self.convert_bin(idx, self.ancilla_size)
            control_bits = []
            for j in range(len(control_bin)):
                if control_bin[j] == '1':
                    control_bits.append(j)
            for n_idx in range(self.n_qubits):
                q = QuantumCircuit(1, name='cry')
                q.ry(parameters[idx - 1][n_idx], 0)
                working_bits = control_bits + [n_idx + self.ancilla_size]
                mcry = MCMT(q.to_gate(), len(control_bits), 1)
                vc.append(mcry, working_bits)
        return vc.decompose()

def full_circuit(self, parameters):
        '''
        Putting all quantum sub circuits into one full circuit.
        This is used for evaluating cost funtions
        '''
        FC = QuantumCircuit(self.tot_qubits)
        tot_qubits_list = [_ for _ in range(self.tot_qubits)]
        ancilla_list = [_ for _ in range(self.ancilla_size)]

        FC.append(self.UC(), ancilla_list)
        FC.barrier()

        FC.append(self.var_circuit(parameters), tot_qubits_list)
        FC.barrier()

        FC.append(self.CA(self.keys), tot_qubits_list)
        FC.barrier()

        FC.append(self.UCD(), ancilla_list)

        return FC

    def global_ground(self, weights):
        '''
        Cost function circuit. 
        '''
        num_parameters = len(self.keys) * self.n_qubits
        parameters = []
        for i in range(num_parameters):
            parameters.append(Parameter('w_' + str(i)))

        value_dict = {}
        for i in range(len(parameters)):
            value_dict[parameters[i]] = float(weights[i])
        print(value_dict)
        parameters = np.array(parameters).reshape(
            len(self.keys), self.n_qubits)
        circuit = QuantumCircuit(self.tot_qubits, self.n_qubits)
        circuit.append(self.full_circuit(parameters), [
                       _ for _ in range(self.tot_qubits)])
        circuit = circuit.bind_parameters(value_dict)
        circuit.measure([self.ancilla_size + _ for _ in range(self.n_qubits)],
                        [_ for _ in range(self.n_qubits)]) 
n_qubits = 10 
# assume that i got 500 as my matrice's coefficients so that i have 500 X 10 of parameters 
weights = 500 * (n_qubits)                  
global_ground(weights) ```       

### What should happen?

It shouldn't throw this error instead it should continue to perform normally like it did for smaller matrices 

### Any suggestions?

Could be an issue of complex number trying to be converted into a float. 

[jakelishman]

Please could you:

• try to reproduce this failure on a more recent version of Qiskit? There have been quite a few bugfix releases since Qiskit 0.44.1.
• try to narrow down the reproducing example to avoid extra code around the edges, and make sure that all imports and other setups are included

If those still don't help, please can you provide any error message in text - screenshots are very hard to read and can't be searched.

[dhamotharnaf]

@jakelishman, thank you for your suggestions, and I apologize for the delay. After following your advice, I experimented with two different approaches while creating virtual environments. I am utilizing a GPU EC2 Machine, so instead of qiskit-aer, I am using qiskit-aer-gpu.

Approach 1 (Virtual Environment 1)

• Version details: qiskit -> 0.45.3 | qiskit-aer-gpu -> 0.13.2 | qiskit-terra -> 0.45.3
• I attempted to run the same problem in this environment, encountering the same error. Thus, it seems that updating the version does not resolve the issue.

Approach 2 (Virtual Environment 2)

• Version details: qiskit -> 0.46.0 | qiskit-aer-gpu -> 0.13.2 | qiskit-terra -> 0.45.3
• Unfortunately, I faced the same error in this environment as well. Despite being a more recent version, I still encountered the issue.

Note: I cannot use qiskit 1.0.0 due to a conflict between qiskit-terra and qiskit>=1.0.

I hope this clarifies the steps I took, and I appreciate your ongoing support in resolving this issue.

Error Message

Traceback (most recent call last):
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTestMain.py", line 76, in <module>
    qx  = solver.Solve()
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTest.py", line 376, in Solve
    w = opt.minimize(fun=self.cost_fn,x0=weights,jac=num_parameters,bounds=[(0, 2 * np.pi) for i in range(num_parameters)])
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/algorithms/optimizers/scipy_optimizer.py", line 154, in minimize
    raw_result = minimize(
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_minimize.py", line 719, in minimize
    res = _minimize_cobyla(fun, x0, args, constraints, callback=callback,
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_cobyla_py.py", line 35, in wrapper
    return func(*args, **kwargs)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_cobyla_py.py", line 278, in _minimize_cobyla
    sf = _prepare_scalar_function(fun, x0, args=args, jac=_jac)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_optimize.py", line 402, in _prepare_scalar_function
    sf = ScalarFunction(fun, x0, args, grad, hess,
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_differentiable_functions.py", line 166, in __init__
    self._update_fun()
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_differentiable_functions.py", line 262, in _update_fun
    self._update_fun_impl()
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_differentiable_functions.py", line 163, in update_fun
    self.f = fun_wrapped(self.x)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/scipy/optimize/_differentiable_functions.py", line 145, in fun_wrapped
    fx = fun(np.copy(x), *args)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTest.py", line 260, in cost_fn
    p_gg = self.global_ground(weights)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTest.py", line 218, in global_ground
    circuit.append(self.full_circuit(parameters), [
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTest.py", line 172, in full_circuit
    FC.append(self.var_circuit(parameters), tot_qubits_list)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/QTest.py", line 156, in var_circuit
    mcry = MCMT(q.to_gate(), len(control_bits), 1)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/library/generalized_gates/mcmt.py", line 85, in __init__
    self._build()
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/library/generalized_gates/mcmt.py", line 98, in _build
    mcmt_gate = broadcasted_gate.control(self.num_ctrl_qubits)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/gate.py", line 129, in control
    return add_control(self, num_ctrl_qubits, label, ctrl_state)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/add_control.py", line 61, in add_control
    cgate = control(operation, num_ctrl_qubits=num_ctrl_qubits, label=label, ctrl_state=ctrl_state)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/add_control.py", line 143, in control
    controlled_circ.mcry(
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/library/standard_gates/multi_control_rotation_gates.py", line 331, in mcry
    RYGate(theta).to_matrix(),
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/gate.py", line 61, in to_matrix
    return self.__array__(dtype=complex)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/library/standard_gates/ry.py", line 106, in __array__
    cos = math.cos(self.params[0] / 2)
  File "/home/aakif.akhtar/CVQLS/CVQLS_Qiskit/qiskit_gpu_v3/lib/python3.10/site-packages/qiskit/circuit/parameterexpression.py", line 493, in __float__
    raise TypeError(
TypeError: ParameterExpression with unbound parameters (dict_keys([Parameter(w_112)])) cannot be cast to a float.