buQeYes

Inspiration

We learned about the Quantum Advantage Tracker, an open challenge for classical and quantum methods to compete with one another. We were inspired by BlueQubit's extensive contributions to this project, specifically using peaked circuits, and wanted to learn more by implementing some of their methods as well as coming up with some of our own.

Challenges we ran into

The greatest challenge we ran into was certainly implementing IBM's approach, which at a lower level was too challenging to implement exactly, so we had to change some of the steps with a substitute method. Additionally, for the last problem, even the IBM approach was too slow. Although we tried to interface with GPUs, it still did not result in a successful outcome.

Accomplishments that we're proud of

Simply through inspection of the CZ connections of circuit #7, we noticed that it could break down into 2 simpler circuits small enough to be run on a simple statevector simulator. We thought this was the most creative of our solutions.

What we learned

Going into this challenge, we knew very little about approximation techniques such as MPS and Pauli-Path. We certainly improved our understanding by implementing some of these methods on a lower level once the computation was too expensive to use through BlueQubit's SDK. We also gained experience with circuit simplification techniques, such as factorization, consolidation of unitary gates, and removal of redundant operations, things that we have done before but we certainly have not done to this scale.

What's next for buQeYes

What's next for us is to delve deeper into IBM's approach, to ensure its accuracy and optimize its computation speed since that was the bottleneck that we were unable to overcome in determining the bitstring for circuit #10.

Built With

• a100
• bluequbit
• cuda
• python
• qiskit
• quantumrings
• quimb