Hardware-Aware Quantum Transpiler

Honors Computer Science Research Project

Senior Year | Honors Computer Science Research

Explore how hardware connectivity shapes a quantum circuit.

This research project is a Python prototype for transpiling quantum circuits against different qubit-connectivity layouts. It models mesh and heavy-hex-style architectures and supplies Qiskit with a target's coupling map and native gate set. The goal is to grow this into a clearer, staged pipeline that first lowers circuits into hardware-supported gates, then maps and routes them for the target architecture.

View the project on GitHub

Current Prototype

The prototype builds a graph for the selected layout, then passes that connectivity and the target basis gates to Qiskit's transpiler. Qiskit currently handles gate decomposition and routing together in its transpilation pass. A sample in the project configures a 45-qubit circuit with an H, X, Y, Z, and CNOT gate set.

01

Describe the target

Select a modeled architecture and qubit count, then specify the gates supported by the target.

02

Model connectivity

Represent qubits as graph nodes and allowed two-qubit interactions as edges in a coupling map.

03

Run Qiskit transpilation

Pass the circuit, coupling map, and basis gates to Qiskit, which decomposes and routes operations for the selected target.

Where I Want to Take It

The next step is to make the stages of hardware adaptation more explicit and investigate how the choices at each stage affect the final circuit.

01

Lower to native gates

Add a distinct conversion stage that rewrites circuit operations into the gate set supported by the selected target before routing begins.

02

Map and route for connectivity

Assign logical qubits to physical qubits and handle interactions between qubits that are not directly connected, while preserving the circuit's behavior.

03

Compare transpilation results

Measure outcomes such as circuit depth, gate counts, and added routing operations to compare layouts and transpilation choices.

The longer-term aim is an end-to-end, hardware-aware workflow that makes it easier to see how a circuit is transformed from its input form into a version suited to a particular architecture.

Modeled Architectures

Mesh

A rectangular grid model connecting neighboring qubits horizontally and vertically. The row and column dimensions are derived from the requested qubit count.

Heavy-Hex-Style Lattice

A lattice model that connects qubits horizontally, vertically, and along staggered diagonals. The resulting edges are supplied to Qiskit as the target coupling map.

Tools and Visualization

Qiskit
Transpiles circuits using the selected basis gates and target coupling map.
Python
Implements the layout models and combines them with the circuit transpilation workflow.
NetworkX & Matplotlib
Draw the modeled qubit graph so its connectivity can be inspected visually.