Use this Google Form to register your teams: here
Teams must follow the rules and guidelines mentioned in the Discord Channel
Project 1 - Setting up a new device type for user queries
Go find a device geometry that is not already implemented in SQuADDS, like two coupled differential transmons, a transmon with readout cavity and Purcell filter, two transmons coupled via a tunable coupler, etc., based on a published research article. This article must have enough information published so that you can fully reconstruct the device Hamiltonian--that is, all mode frequencies, anharmonicities, and couplings must be given explicitly or be easily obtainable from other data in the paper. Write code to generate the design (using Qiskit Metal), to vary relevant design geometry parameters (e.g., the dimensions of a coupling capacitor), to simulate it (using PALACE or ANSYS HFSS), and to extract the parameters of an effective lumped-element model that will give the correct effective Hamiltonian. Run quick-and-dirty simulations to see if your simulation results are vaguely close to the published data. Once you've accomplished this, contact the corresponding author of the paper and ask them if you have their permission to contribute the design to SQuADDS--if not, promise to keep it confidential.
The winning project will be given computing time on the Levenson-Falk Lab's workstations to run more accurate simulations, then will be asked to contribute the design if the author approves.
See SQuADDS wish list for detailed ideas: https://github.com/LFL-Lab/SQuADDS/blob/master/wish_list.md
Project 2 - Arbitrary Quantum Operation Construction
Universal gate sets can be used to achieve any desired unitary operator. More general quantum operations, however, can be non-unitary and involve behavior such as dissipation. By using measurements alongside universal gate sets, we should be able to construct arbitrary quantum operations. In this project, students will build a numerical optimizer that, when given a universal gate set and the ability to measure, is able to achieve any desired Kraus map specifying a quantum operation of interest Reference. This project will leverage bosonic and jaxquantum. Later, students can extend the technique to run with closed-loop feedback from an experiment (e.g. with reinforcement learning) to implement desired quantum operations on hardware with high fidelity.
Some Reference: https://arxiv.org/pdf/1611.03463
Project 3 - Measurement-free Topological QEC
In bosonic QEC, we are able to use resets (instead of readout) to dump errors out of the system and stabilize our logical quantum information by cooling back into the logical codespace. Is it possible to do the same for something like a repetition code or surface code? This would allow us to use resets instead of relying on very high-fidelity readout, the latter of which is quite difficult to achieve. This project will use jaxquantum. Initial explorations: note and code
Some References:
- https://github.com/EQuS/jaxquantum/tree/main
- https://mit-equs.notion.site/Autonomous-Register-based-QEC-9d8babd390ea4fffa6b045b669bf783e
- https://github.com/EQuS/jaxquantum/blob/main/tutorials/arbqec/1-rep.ipynb
Project 4 - Design a transmon in a creative way to minimize “antenna modes” from the best values found in the literature
Some References:
- https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.132.017001
- https://arxiv.org/abs/2203.06577
Project 5 - Design a transmon in a creative way to minimize Purcell decay from the best values found in the literature
Some References:
- https://arxiv.org/abs/2405.10107
- https://arxiv.org/abs/2409.04967
- https://journals.aps.org/prx/pdf/10.1103/PhysRevX.14.041007
Project 6 - Contribute to the codebases - JaxQuantum, QCSys, and Bosonic
- Implement GKP qudit support in bosonic, based on https://arxiv.org/abs/2409.15065
- Implement Tesseract code support in bosonic based on https://journals.aps.org/prxquantum/abstract/10.1103/PRXQuantum.3.010335
- Add more superconducting qubits to qcsys.
- Add support to construct and quantize arbitrary superconducting qubits in qcsys.
- Add profiling support to bosonic, qcsys, or jaxquantum to identify and fix bottlenecks.
Some References:
- https://arxiv.org/abs/2409.15065
- https://github.com/EQuS/qcsys
- https://github.com/EQuS/bosonic
- https://github.com/EQuS/jaxquantum
Project 7 - Create machine learning model for generating a device geometry for a given target Hamiltonian, based on data already in SQuADDS
Currently, when a user inputs a target Hamiltonian, SQuADDS returns them the best-matching pre-simulated design and also uses a physics-based algorithm to give them an estimated best-fit design that has not yet been simulated. This algorithm is difficult to derive and does not generalize well. This is an ideal case for machine learning. Using the SQuADDS database on HuggingFace, create a ML model that is capable of generating a geometry that will give the correct behavior based on a user's target, without running an electromagnetic simulation. Ideally, your model should be able to train on some subset of the SQuADDS database and then generate the rest of the database with reasonable accuracy. Bonus points if your model is easily adapted to other types of design geometry.
The winning project will be asked to deploy on the SQuADDS HuggingFace account.
See SQuADDS wish list for detailed ideas: https://github.com/LFL-Lab/SQuADDS/blob/master/wish_list.md
Project 8 - Reliable integration with open-source EM solver
Currently, SQuADDS generates code to simulate designs using ANSYS, and the database was entirely simulated in ANSYS. However, ANSYS is expensive--it would be better to switch to some free / open-source solver. One such solver is AWS PALACE, but others may be available. Write wrapper code to export a design from SQuADDS into one of these solvers, run an accurate simulation, and import the results back into SQuADDS. Validate your simulation pipeline using the experimentally-measured WM1 chip from the SQuADDS database--show that you can match the device parameters using your simulations.
The winning project will be asked to deploy to the SQuADDS package.
See SQuADDS wish list for detailed ideas: https://github.com/LFL-Lab/SQuADDS/blob/master/wish_list.md
Project 9 - Design an optical mask for a simple multi-transmon device where the transmons have optimized Purcell-limited T1
Some Reference: https://qiskit-community.github.io/qiskit-metal/circuit-examples/index.html#small-quantum-chips
Project 10 - Design transmons of varying Ej/Ec and look for strategies to further reduce the surface participation ratio and radiative losses from the best found in the literature
Some References:
- https://www.nature.com/articles/s41534-022-00530-6
- https://sites.google.com/quantala.tech/website/scientific-papers?authuser=0
- https://arxiv.org/abs/2204.07202
Project 11 - Specify two superconducting transmons using Qubit simulator and simulate their energy spectrum
Some Reference: https://github.com/Borealis126/QSM