# Gate not supported in default.mixed device

**URL:** <https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201>\
**Category:** PennyLane Help\
**Created:** [March 1, 2024, 3:25pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201 "2024-03-01T15:25:34Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![sdas](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/sdas/32/1615_2.png) [@sdas](https://discuss.pennylane.ai/u/sdas)\
**Post date:** [March 1, 2024, 3:25pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201/1 "2024-03-01T15:25:34Z")

</div>

I am using default.mixed device but getting the following error

pennylane.\_device.DeviceError: Gate C(CNOT) not supported on device default.mixed

Is there any way to tackle this?  
My code is the following:

```auto
import pennylane as qml
from pennylane import numpy as np
import math

# Generate density matrices and arbitrary labels
density_matrix = []
labels = []
for i in range(20):
    state = np.random.randn(16,1) + 1j*np.random.randn(16, 1)
    state_normlized = np.array(state/np.linalg.norm(state))
    density_matrix.append(np.dot(state_normlized, state_normlized.conj().T))
    if state_normlized[0].real > 0:
        labels.append(1)
    else:
        labels.append(0)

def cross_entropy(labels, predictions):
    loss = 0
    for l, p in zip(labels, predictions):
        c_entropy = l * (anp.log(p[l])) + (1 - l) * anp.log(1 - p[1 - l])
        loss = loss + c_entropy
    return -1 * loss
    

def cost(params, X, Y, cost_fn):
    predictions = np.array([circuit(x, params) for x in X])
    if cost_fn == 'cross_entropy':
        loss = cross_entropy(Y, predictions)
    return loss
    

dev = qml.device('default.mixed', wires = 7)
@qml.qnode(dev)
def circuit(features, params):
    ancilla_state = [1/math.sqrt(6), 1/math.sqrt(6), 1/math.sqrt(6), 1/math.sqrt(6), 1/math.sqrt(6), 1/math.sqrt(6), 0, 0]
    # Embedding of data and initialization of control register
    qml.QubitDensityMatrix(features, wires = [0, 1, 2, 3])
    qml.AmplitudeEmbedding(ancilla_state, wires = [4, 5, 6])
    
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (0, 0, 0))(params[0], wires=(0,1))
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (0, 0, 1))(params[1], wires=(0,2))
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (0, 1, 0))(params[2], wires=(0,3))
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (0, 1, 1))(params[3], wires=(1,2))
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (1, 0, 0))(params[4], wires=(1,3))
    qml.ctrl(qml.IsingXX, control=(4,5,6), control_values = (1, 0, 1))(params[5], wires=(2,3))
       
    result = qml.probs(wires=0)
    return result
    
batch_size = 6
learning_rate = 0.005
total_params = 6  
params = np.random.randn(total_params, requires_grad=True)
opt = qml.NesterovMomentumOptimizer(stepsize=learning_rate)
cost_fn = "cross_entropy"

for it in range(10):
    batch_index = np.random.randint(0, len(density_matrix), (batch_size,))
    X_batch = [density_matrix[i] for i in batch_index]
    Y_batch = [labels[i] for i in batch_index]
    params, cost_new = opt.step_and_cost(lambda v: cost(v, X_batch, Y_batch, cost_fn), params)
    

```

I get the following error message:  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/qnode.py”, line 1027, in **call**  
res = qml.execute(  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/interfaces/execution.py”, line 616, in execute  
results = inner\_execute(tapes)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/interfaces/execution.py”, line 249, in inner\_execute  
return cached\_device\_execution(tapes)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/interfaces/execution.py”, line 371, in wrapper  
res = list(fn(tuple(execution\_tapes.values()), \*\*kwargs))  
File “/home/sreetamadas/anaconda3/lib/python3.10/contextlib.py”, line 79, in inner  
return func(\*args, \*\*kwds)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_qubit\_device.py”, line 460, in batch\_execute  
res = self.execute(circuit)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/devices/default\_mixed.py”, line 685, in execute  
return super().execute(circuit, \*\*kwargs)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_qubit\_device.py”, line 276, in execute  
self.check\_validity(circuit.operations, circuit.observables)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_device.py”, line 984, in check\_validity  
raise DeviceError(  
pennylane.\_device.DeviceError: Gate C(CNOT) not supported on device default.mixed

---

<div class="post-metadata">

**Author:** ![CatalinaAlbornoz](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/catalinaalbornoz/32/1196_2.png) [@CatalinaAlbornoz](https://discuss.pennylane.ai/u/CatalinaAlbornoz)\
**Post date:** [March 4, 2024, 3:34pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201/2 "2024-03-04T15:34:11Z")

</div>

Hi @sdas,

Thank you for your question!

We will release a new version of PennyLane tomorrow which should fix your problem. Please let us know if the problem persists after upgrading to PennyLane v0.35 😃.

---

<div class="post-metadata">

**Author:** ![sdas](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/sdas/32/1615_2.png) [@sdas](https://discuss.pennylane.ai/u/sdas)\
**Post date:** [March 5, 2024, 10:48pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201/3 "2024-03-05T22:48:03Z")

</div>

Hi @CatalinaAlbornoz , as suggested I upgraded pennylane. Now when I run the above code, I get the following error:

Traceback (most recent call last):  
File “/home/sreetamadas/mixed\_state\_simulation.py”, line 62, in   
params, cost\_new = opt.step\_and\_cost(lambda v: cost(v, X\_batch, Y\_batch, cost\_fn), params)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/optimize/gradient\_descent.py”, line 64, in step\_and\_cost  
g, forward = self.compute\_grad(objective\_fn, args, kwargs, grad\_fn=grad\_fn)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/optimize/nesterov\_momentum.py”, line 76, in compute\_grad  
grad = g(\*shifted\_args, \*\*kwargs)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_grad.py”, line 165, in **call**  
grad\_value, ans = grad\_fn(\*args, \*\*kwargs) # pylint: disable=not-callable  
File “/home/sreetamadas/.local/lib/python3.10/site-packages/autograd/wrap\_util.py”, line 20, in nary\_f  
return unary\_operator(unary\_f, x, \*nary\_op\_args, \*\*nary\_op\_kwargs)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_grad.py”, line 183, in \_grad\_with\_forward  
vjp, ans = \_make\_vjp(fun, x) # pylint: disable=redefined-outer-name  
File “/home/sreetamadas/.local/lib/python3.10/site-packages/autograd/core.py”, line 10, in make\_vjp  
end\_value, end\_node = trace(start\_node, fun, x)  
File “/home/sreetamadas/.local/lib/python3.10/site-packages/autograd/tracer.py”, line 10, in trace  
end\_box = fun(start\_box)  
File “/home/sreetamadas/.local/lib/python3.10/site-packages/autograd/wrap\_util.py”, line 15, in unary\_f  
return fun(\*subargs, \*\*kwargs)  
File “/home/sreetamadas/mixed\_state\_simulation.py”, line 62, in   
params, cost\_new = opt.step\_and\_cost(lambda v: cost(v, X\_batch, Y\_batch, cost\_fn), params)  
File “/home/sreetamadas/mixed\_state\_simulation.py”, line 27, in cost  
predictions = np.array([circuit(x, params) for x in X])  
File “/home/sreetamadas/mixed\_state\_simulation.py”, line 27, in   
predictions = np.array([circuit(x, params) for x in X])  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/workflow/qnode.py”, line 1048, in **call**  
res = qml.execute(  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/workflow/execution.py”, line 684, in execute  
results = inner\_execute(tapes)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/workflow/execution.py”, line 283, in inner\_execute  
return cached\_device\_execution(tapes)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/workflow/execution.py”, line 361, in wrapper  
res = list(fn(tapes, \*\*kwargs))  
File “/home/sreetamadas/anaconda3/lib/python3.10/contextlib.py”, line 79, in inner  
return func(\*args, \*\*kwds)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_qubit\_device.py”, line 459, in batch\_execute  
res = self.execute(circuit)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/devices/default\_mixed.py”, line 685, in execute  
return super().execute(circuit, \*\*kwargs)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/\_qubit\_device.py”, line 277, in execute  
self.apply(circuit.operations, rotations=self.\_get\_diagonalizing\_gates(circuit), \*\*kwargs)  
File “/home/sreetamadas/anaconda3/lib/python3.10/site-packages/pennylane/devices/default\_mixed.py”, line 693, in apply  
raise DeviceError(  
pennylane.\_device.DeviceError: Operation StatePrep cannot be used after other Operations have already been applied on a default.mixed device.

---

<div class="post-metadata">

**Author:** ![isaacdevlugt](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/isaacdevlugt/32/1159_2.png) [@isaacdevlugt](https://discuss.pennylane.ai/u/isaacdevlugt)\
**Post date:** [March 7, 2024, 7:45pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201/4 "2024-03-07T19:45:52Z")

</div>

Hey @sdas!

The issue here is that you’re using two instances of state preparation (`Operation StatePrep cannot be used after other Operations have already been applied on a default.mixed device.`)

> [@sdas](#):
>
> ```auto
> qml.QubitDensityMatrix(features, wires = [0, 1, 2, 3])
> qml.AmplitudeEmbedding(ancilla_state, wires = [4, 5, 6])
> 
> ```

The way around this is to mush it all together into a 7-qubit density matrix. Mathematically, you want to do:

\rho \otimes \vert \psi \rangle \langle \psi \vert

and plug that into `QubitDensityMatrix`. The code would look like this:

```auto
state_as_density_matrix = np.outer(ancilla_state, np.conj(ancilla_state))
features = np.kron(features, state_as_density_matrix)

```

Let me know if that helps!

---

<div class="post-metadata">

**Author:** ![sdas](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/sdas/32/1615_2.png) [@sdas](https://discuss.pennylane.ai/u/sdas)\
**Post date:** [March 16, 2024, 2:54pm UTC](https://discuss.pennylane.ai/t/gate-not-supported-in-default-mixed-device/4201/5 "2024-03-16T14:54:00Z")

</div>

Hi @isaacdevlugt , thanks for the reply. Yes this indeed works.