# Platform for mixed-state quantum machine learning?

I have been using PennyLane to run numerical QML simulations but it now seems to only support backprop on pure state simulations. Does anyone familiar with other packages know if there exists one that supports quantum machine learning on mix-states?

Thank you.

You can use default.mixed on PennyLane for simulating mixed states. However, instead of using backpropagation you can set diff_method="best" when defining your qnode, which in this case will use the parameter-shift rule for differentiation.

EG:

@qml.qnode(dev, diff_method="best")