Connectivity & Topology
Track: Quantum Hardware & Providers · Difficulty: Beginner–Intermediate · Est: 12 min
Connectivity & Topology
Overview
This page answers the question: “Which qubits can interact directly, and what happens when they can’t?”
Quantum algorithms are usually described as if any two qubits can be entangled on demand. Real devices have constraints: only certain pairs of qubits interact easily. Those constraints shape performance and the kinds of circuits that run well.
Intuition
Think of qubits as cities and two-qubit gates as direct flights.
- In an all-to-all network, every city has a direct flight to every other city.
- In a local network, cities mostly connect to nearby neighbors.
If you need to move a passenger from one far city to another:
- all-to-all: one flight
- local: multiple connecting flights
In circuits, the “connecting flights” are extra operations inserted to make a desired interaction possible. Those extra operations increase circuit length and expose the computation to more noise.
What This Metric Captures
“Connectivity” and “topology” capture:
- Which qubit pairs can perform a two-qubit gate directly (native interactions).
- The interaction graph shape (line, grid, star-like, near-all-to-all, etc.).
- The routing difficulty for circuits that require non-local interactions.
Sometimes specs also describe directionality (whether a two-qubit interaction works equally well both ways) and available couplers (how interactions are mediated).
What This Metric Misses
Connectivity alone still doesn’t tell you how well the device performs. It misses:
- Gate quality (Noise & Errors connection): extra routing steps only matter because each step has error.
- Compilation strategy: the same circuit can be mapped to hardware in different ways with very different overhead.
- Non-uniform links: some connections are “stronger” or more reliable than others.
- Crosstalk and scheduling constraints: even if two pairs are connected, doing operations simultaneously may be limited.
Most importantly, a topology that looks restrictive can still be workable if gates are high quality, while a rich topology can underperform if gates are noisy.
Turtle Tip
When you see a connectivity diagram, translate it into a question about routing: “How many extra operations will I need when my circuit wants distant qubits to interact?” Those extra operations turn topology into real cost.
Common Pitfalls
- Assuming all-to-all connectivity automatically means better results for every workload.
- Ignoring routing overhead when reading circuit depth or algorithm descriptions.
- Forgetting that routing overhead multiplies noise exposure: more steps means more chances for errors.
- Treating connectivity as a single number; the graph structure matters.
Quick Check
- What does “connectivity” mean in the context of a quantum device?
- Why can local connectivity increase circuit depth?
- What is the intuitive reason that SWAP-like routing overhead hurts more when noise is present?
What’s Next
Connectivity tells you how much routing you’ll need. Next we look at gate fidelity and error rates, which determine how costly those extra operations are in practice.