Preserve i->i initial layout when coupling_map satisfied

[ajavadia]

When a circuit already satisfies the coupling_map on the backend, the swap_mapper should not insert any swaps. This means passing an initial_layout of q[i]->q[i].

PR #478 optimizes the initial layout, but that creates unexpected behavior when the coupling map is already satisfied. Here, I only apply the optimization of #478 if swaps are indeed needed.

The other thing that this PR does is a slight name change: skip_translation -> skip_transpiler, to get ready for the transpiler PR.

Motivation and Context

Issue was raised here:
https://quantumexperience.ng.bluemix.net/qx/community/question?questionId=5b0d17f943fb6e0038a623bf

Screenshots (if appropriate):

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.
• All new and existing tests passed.

[nonhermitian]

'Noy' -> 'Not

[nonhermitian]

Do we really want to remove this kwarg? For backward compatibility, should it not implicitly set skip_transpiler and raise a depreciation warning saying that it will be removed in the future?

[ajavadia]

thanks, good point.

[nonhermitian]

For a two qubit gate between qubits i and j, I thought the check would be if j is in coupling_map[i] or if i in coupling_map[j]. Is that what is going on here, or (more likely) am I missing something?

[ajavadia]

This checks for exact match, meaning if there is a CX(q[i], q[j]) in the circuit, it checks that the device also has CX(q[i], q[j]) in it. The reason is that I want to keep already-matching circuits untouched, but pass any other ones through the mapper. A good mapper may actually find a more compact circuit, but usually a user that writes an already-matching circuit wants to use those exact qubits.

[atilag]

Yeah, I found this confusing at first, because I thought that the coupling_map was an adjacency list, but it's not. It more a list of coupled pairs: coupling_map = [[1, 0], [2, 0], [2, 1], [3, 2], [3, 4], [4, 2]].

[nonhermitian]

Yes, your right. This is what we want.

[CLAassistant]

All committers have signed the CLA.