You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently some gate operations are done by combining several other gate operations. For example, the swap is done by using three cx operations. There is nothing wrong with that functionally, of course, but instead of going separately with each over all stabilizers & destabilizers, a single loop could apply all such operations. This should increase the speed a little due of cache locality.
Informations
What is the current behavior?
Currently some gate operations are done by combining several other gate operations. For example, the swap is done by using three cx operations. There is nothing wrong with that functionally, of course, but instead of going separately with each over all stabilizers & destabilizers, a single loop could apply all such operations. This should increase the speed a little due of cache locality.
What is the expected behavior?
Something like this implementation I've done: https://github.com/aromanro/QCSim/blob/4b2589cb50e2724fbd2da2b18c57001595b3f3f8/QCSim/Clifford.h#L288
Currently aer is doing this:
qiskit-aer/src/simulators/stabilizer/stabilizer_state.hpp
Line 381 in 269c26f
which results in three loops over the stabilizers & destabilizers.
This is only an example, there are many gate operations that could be changed like that.
The text was updated successfully, but these errors were encountered: