Some of the device coupling maps in benchmark/topology seem to be inconsistent with the actual

[The-Sunspot]

I wanted to customize some quantum devices for experiments, so I checked benchmark/topology to learn how to define quantum devices.

Here I found the device linear_5.layout, which from the naming seems to be a linear device of length five, and its coupling map looks very intuitive: qubits are labeled from 0, and a two-dimensional list is used to record each edge from smallest to largest (including reversed edges on an undirected graph).

Further, I found the example device example.layout, which looks linear in length 3. Also the comments on the couplingmap fit my guesses perfectly, and I think I've got a full understanding of how to customise the coupling map of a device.

But when I check guadalupe_16.layout, I have some problems. As I understand, this file should correspond to the IBMQ guadalupe device, but in its coupling map, a 15 to 1 edge is recorded, which is supposed to be 15 to 12, and at the same time this results in no reverse edge for the 12 to 15 edge.

The same problem occurs in tokyo_20.layout, I haven't checked the other devices yet, and I'm not sure if I'm wrong about the rules for custom coupling maps, or if there's some error in these files themselves?