Routing #6
Routing #6
Conversation
…ial-routing' branch
|
Here's a little tip in case you weren't aware: Use |
|
|
||
| class TransportError(Exception): | ||
| def __init__(self, a: list[int], b: list[int]): | ||
| super().__init__(f"Traps different sizes: {len(a)} vs. {len(b)}") |
There was a problem hiding this comment.
Rather than different sizes, this is more telling that placement was done improperly. If the current state does not have the same number of qubits as the goal state, it implies that not every qubit was placed properly. It is a side effect of improper placement.
tests/jonhas_test.py
Outdated
|
|
||
| m = TestMachine(12, {0,2,4,6,8,10}, 10, 2, 2) | ||
|
|
||
| def parse_circuit(file): |
There was a problem hiding this comment.
Remove as we should only be working with pytket circuits now that Neal's sharding looks good
tests/test.py
Outdated
| assert transport_cost(init, goal, swap_cost) == 2 | ||
|
|
||
| #many | ||
| init = [0,1,2,3,4,5,6,7,8,9] |
There was a problem hiding this comment.
If we are looking at distance based, the number of swaps isn't 9 here is it, but rather 11?
| # Because SQ zones are offsets of TQ zones, each tq op covers 2 sq zones | ||
| raise GateOpportunitiesError | ||
|
|
||
| # place the tq ops |
There was a problem hiding this comment.
You may want to break the tq placement into a method to help make this function less long.
| placed_qubits.add(q2) | ||
|
|
||
| # place the sq ops | ||
| for op in sq_ops: |
There was a problem hiding this comment.
Similarly to above a method for placing sq ops seems natural.
pytket/phir/placement.py
Outdated
| raise GateOpportunitiesError | ||
|
|
||
| # place the tq ops | ||
| for op in tq_ops_sorted: |
There was a problem hiding this comment.
This looks quite similar to the tq placement code in the previous method. I didn't run a full diff, but it seems like there's a lot of shared elements, which makes me want to double down on the idea that the tq placement can be broken down into a helper method.
pytket/phir/machine_class.py
Outdated
| """ | ||
| self.size = size | ||
| self.tq_options = tq_options | ||
| self.sq_options = set() |
There was a problem hiding this comment.
I think the argument types make the types of the various fields obvious, but we may still want to either add type annotations to each of the fields in the class, or at the least add the annotation self.sq_options: set[int] = set(), so it's clear what thet set is supposed to hold.
| tq_options = {4, 8, 12, 16, 20, 24, 28} | ||
| sq_options = set(range(32)) | ||
| trap_size = 32 | ||
| expected = [ |
There was a problem hiding this comment.
how did you arrive at this expected test value? The way the tq gates map to tq_indices seems straightforward, but the rest does not.
No description provided.