KeyError in phased_graph.py during SCMR mode mapping (KeyError: 8)

[teotro]

First, thank you for releasing this great tool -- it has been very helpful for exploring mapping and routing of quantum circuits.
While experimenting with the SCMR mode, I encountered a reproducible KeyError related to the mapping dictionary in phased_graph.py.

Command executed:

wisq /path/to/quantum-compiler-benchmark-circuits/arithmetic/adder_8.qasm --mode scmr -arch compact_layout -tmr 4

Error trace:

Traceback (most recent call last):
  File "/home/george/wisq/.venv/bin/wisq", line 8, in <module>
    sys.exit(main())
  File "/home/george/wisq/src/wisq/__init__.py", line 303, in main
    map_and_route(
  File "/home/george/wisq/src/wisq/__init__.py", line 82, in map_and_route
    map, steps = run_dascot(circ, gates, arch, output_path, timeout)
  File "/home/george/wisq/src/wisq/dascot.py", line 74, in run_dascot
    phased_map, _ = build_phased_map(
  File "/home/george/wisq/src/wisq/phased_graph.py", line 276, in build_phased_map
    final_mapping, cost = sim_anneal(...)
  File "/home/george/wisq/src/wisq/phased_graph.py", line 228, in sim_anneal
    best_overlaps = count_overlapping_fast(mapping, phased_graphs_fast, arch)
  File "/home/george/wisq/src/wisq/phased_graph.py", line 169, in count_overlapping_fast
    closest = min(magic_states_2d, key=lambda p : abs(p[0] - mapping[c][0]) + abs(p[1] - mapping[c][1]) )
KeyError: 8

Are you able to reproduce this issue and be able to share some insights why we observe this behavior? I can also see this behavior for other benchmarks from your quantum-compiler-benchmark-circuits suite and my ow benchmarks as well.

[psdecabooter]

Hi, I have taken a look and I think I may know your issue @teotro . The example gate you gave us uses a lot of 'ccx' gates which are not supported by wisq mapping and routing. We will work on a way to make this error more gracefully, but as a temporary fix for now you should decompose your circuits then run them in wisq.

[amandashoe]

Detect if circuit has gates that are not in {cx, t, tdg, h, x, s, sdg} and then suggest the user invoke full_ft mode with optimization timeout set to 0. Perhaps best to include the exact command to run