Understanding Neutral Atom compilers

[quantum-brutus]

Hello,
Hope you are doing well!

I had some questions about the zoned neutral atom compilator.

I looked a bit at QMAP, and it said in the documentation that OPENQASM circuits were accepted as argument of the circ = load(qasm), as well as qiskit.QuantumCircuit objects.

Let's use OPENQASM for a simple circuit:

import stim
print(stim.__version__)

stim_circuit = stim.Circuit('''
            CZ 0 1
            DETECTOR rec[-1] rec[-2]
             ''')

qasm = stim_circuit.to_qasm(open_qasm_version=2, skip_dets_and_obs = True)

My stim version is 1.15.0. QMAP one is 3.3.1.

Let's compile it and see the output using an arbitrary architecture (you can use one from the tutorial):

compiler = RoutingAwareCompiler(arch)
circ = load(qasm)
code = compiler.compile(circ)
print(code)

The output is:

atom (15.000, 3.000) atom0
atom (18.000, 3.000) atom1
@+ load [
atom0
atom1
]
@+ move [
(17.000, 20.000) atom0
(19.000, 20.000) atom1
]
@+ store [
atom0
atom1
]
@+ cz zone_cz0
@+ load [
atom0
atom1
]
@+ move [
(15.000, 3.000) atom0
(18.000, 3.000) atom1
]
@+ store [
atom0
atom1
]

On this, I have a question:

• Why are the qubits going back to the storage place after the CZ instead of just waiting around in the entanglement zone?

When I try to compile this circuit:

stim_circuit = stim.Circuit('''
            CZ 0 1
            SQRT_Y 0 1
            DETECTOR rec[-1] rec[-2]
             ''')

I get an additional

@+ u 1.57080 0.00000 0.00000 atom0
@+ u 1.57080 0.00000 0.00000 atom1

• Why the "@+ u 1.57080 0.00000 0.00000 atom0. "? It is causing the visualization tool, naviz, not to work as these are not supported instructions. Why not "@+ ry 1.57080" as instruction and why not "global" to the target atoms?

Let's try another circuit :

When I try to compile this circuit:

stim_circuit = stim.Circuit('''
            Z 0 1
            DETECTOR rec[-1] rec[-2]
             ''')

The output is

atom (15.000, 3.000) atom0
atom (18.000, 3.000) atom1
@+ rz 3.14159 atom0
@+ rz 3.14159 atom1

• Same question as before, why doesn't the compiler recognize that these a global rotations ?

Another thing that would be nice is for the compiler to handle circuits with resets, and measurements (with a dedicated non-destructive measurement zone).

Circuits like these:

stim_circuit = stim.Circuit('''
            R 0 1
            Z 0 1
            M 0 1
            DETECTOR rec[-1] rec[-2]
             ''')

return

ValueError: Operation type not supported: reset with 0 controls and 1 targets

Otherwise this open-source tool is great. Thanks in advance for the response!