Craig Gidney

> Would making the second one an "@" would be confusing, since it might make someone think it actually is a CNOT? But it *is* a controlled not. It's a...

You definitely wouldn't want to remove Rz (since people will indeed want to use it). But the internal constructions can prefer ZPow over Rz and thereby gain free precision. There'd...

Having rotation gates support `__pow__`, with the gate as either the base or the exponent, would subsume this idea.

Some followup on what this would look like: ```python def _exp_i_pi(exponent): exponent %= 2 exponent = float(exponent) # At half-steps, give results without floating point error. if exponent % 0.5...

This is a qcircuit bug. That being said, I don't think it's going to be fixed there and we probably should work around it somehow..

I don't actually know if there's anyone maintaining it...

> Do you have an example of when the cost of key lookup would be significant compared to the cost of doing the simulation itself? It's not so much that...

I agree that a QASM export would be good. However, note that this functionality would be a lot of work. For example, QASM doesn't have built-in operations for modular exponentiation...

Cirq has an import-from-quirk feature that you might be able to model this after. It defines custom gates in cirq for gates that otherwise wouldn't be present (like modular exponentiation)....

@mpharrigan @dabacon You don't find that the amount of dot output was becoming problematically large? When a test fails, the output is hidden way down instead of being right in...