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
For example, it would be great to test fully-rewinding multireweighting, but there's no way that case will be reached just by setting up a simple supervisor test. This is because a zero-weight edge will be seen earlier in the multireweighting process, causing it to abort. We would need to simulate until reaching MULTIREWEIGHT-BROADCAST-REWEIGHT, then reach in and modify the weight of a nearby node so that there is then a weightless edge between it and the HOLD cluster, and then simulate until dead. I think this feature will also allow us to write tests for SCANing -- we could inject the START-SCAN command and simulate until IDLE.
For example, it would be great to test fully-rewinding multireweighting, but there's no way that case will be reached just by setting up a simple supervisor test. This is because a zero-weight edge will be seen earlier in the multireweighting process, causing it to abort. We would need to simulate until reaching
MULTIREWEIGHT-BROADCAST-REWEIGHT
, then reach in and modify the weight of a nearby node so that there is then a weightless edge between it and theHOLD
cluster, and then simulate until dead. I think this feature will also allow us to write tests forSCAN
ing -- we could inject theSTART-SCAN
command and simulate untilIDLE
.The text was updated successfully, but these errors were encountered: