Bugfix release version cannot run KaHyPar #100
Conversation
… is left for advanced users
| @@ -531,15 +530,15 @@ def test_circ_approx_explicit_ttn(circuit: Circuit) -> None: | |||
| # Check for TTNxGate | |||
| cfg = Config(truncation_fidelity=0.99, leaf_size=3, float_precision=np.float32) | |||
| ttn_gate = simulate(libhandle, circuit, SimulationAlgorithm.TTNxGate, cfg) | |||
| assert np.isclose(ttn_gate.get_fidelity(), 0.769, atol=1e-3) | |||
| assert np.isclose(ttn_gate.get_fidelity(), 0.751, atol=1e-3) | |||
There was a problem hiding this comment.
Since NetworkX provides a worse partition than KaHyPar, the fidelity in the simulation drops by a bit.
There was a problem hiding this comment.
Are you sure that you want to update the expected value and not the tolerance?
There was a problem hiding this comment.
Hmm, I've been using these tests for a very rudimentary regression/improvement tracking, which is why they have a very particular value for the fidelity. It has happened before that changes that I did not expect that would change the fidelity did. If I made the tolerance margin larger, it's likely some of these wouldn't be caught.
setup.py
Outdated
| @@ -42,7 +42,7 @@ | |||
| license="Apache 2", | |||
| packages=find_namespace_packages(include=["pytket.*"]), | |||
| include_package_data=True, | |||
| install_requires=["pytket ~= 1.26"], | |||
| install_requires=["pytket ~= 1.26", "networkx >= 2.8"], | |||
There was a problem hiding this comment.
| install_requires=["pytket ~= 1.26", "networkx >= 2.8"], | |
| install_requires=["pytket ~= 1.26", "networkx ~= 3.0"], |
There was a problem hiding this comment.
I would be careful with allowing 4.0 here, do you need 2.x here, or should we just ask for 3.x to be installed? (see suggestion)
There was a problem hiding this comment.
Good point. I believe 3.x should work, but I'll quickly check.
There was a problem hiding this comment.
Yep, it works 👍
| @@ -531,15 +530,15 @@ def test_circ_approx_explicit_ttn(circuit: Circuit) -> None: | |||
| # Check for TTNxGate | |||
| cfg = Config(truncation_fidelity=0.99, leaf_size=3, float_precision=np.float32) | |||
| ttn_gate = simulate(libhandle, circuit, SimulationAlgorithm.TTNxGate, cfg) | |||
| assert np.isclose(ttn_gate.get_fidelity(), 0.769, atol=1e-3) | |||
| assert np.isclose(ttn_gate.get_fidelity(), 0.751, atol=1e-3) | |||
There was a problem hiding this comment.
Are you sure that you want to update the expected value and not the tolerance?
| assert ttn_gate.is_valid() | ||
| assert np.isclose(ttn_gate.vdot(ttn_gate), 1.0, atol=cfg._atol) | ||
|
|
||
| # Fixed virtual bond dimension | ||
| # Check for TTNxGate | ||
| cfg = Config(chi=120, leaf_size=3, float_precision=np.float32) | ||
| ttn_gate = simulate(libhandle, circuit, SimulationAlgorithm.TTNxGate, cfg) | ||
| assert np.isclose(ttn_gate.get_fidelity(), 0.857, atol=1e-3) |
There was a problem hiding this comment.
Yeah, I'd rather change the fidelity. The value of the fidelity is consistent between different runs using the same code.
Co-authored-by: cqc-melf <[email protected]>
Description
Now NetworkX is used as the default option for partition, and KaHyPar is left for advanced users. In particular, only those who install from repository will be able to use KaHyPar (if they choose to do so with the appropriate
Configoption). This is something I will continue to play with, this is meant just as a hotfix.Related issues
Fixes #97
Checklist