Practical distributed quantum information processing with LOCCNet - Paper Implementation Project

[FGranda2]

Selected Paper for Implementation

The original publication highlights quantum entanglement as a fundamental resource in quantum communication, quantum computing, and other quantum technologies. However, real-world imperfections—such as preparation errors and noise in transmission channels—gradually degrade entanglement over distance. Entanglement distillation seeks to counteract this degradation by consuming multiple noisy entangled pairs and applying Local Operations and Classical Communication (LOCC) to restore highly entangled states.

Several entanglement distillation protocols have been proposed, including:

• BBPSSW Protocol (1996): Designed for isotropic states.
• DEJMPS Protocol (1996): Optimized for Bell-diagonal states.

While effective for specific noise models, these protocols are not generalizable to arbitrary noise. Moreover, due to the complexity of LOCC operations, manually designing new distillation protocols remains a significant challenge.

To overcome this limitation, LOCCNet—a machine learning framework for LOCC protocols—offers a promising approach to studying and optimizing entanglement distillation.

Technical Approach for Implementation

From the publication and some LOCCNet examples found on GitHub, the following key steps were identified:

• State Preparation - Preparation of isotropic states given by:

$$ \rho_{\text{iso}}(p) = p \lvert \Phi^+ \rangle \langle \Phi^+ \rvert + (1 - p) \frac{I}{4}, \quad p \in [0, 1] $$

• Create the platform so that these states can be distributed to Alice and Bob.
• Use the platform to apply the QNN circuits as LOCC operations on Qubits held by Alice and Bob.
• Perform proper measurements and post processing for successful distillation.
• Implement an optimization scheme for maximizing state fidelity using classical optimizers (e.g., COBYLA, ADAM) .
• Add implementation for distributed quantum state discrimination as part of the platform.
• Add implementation for quantum channel simulation as part of the platform.

High-Level Example

For this specific publication, I already completed an initial implementation of the methodology for distillation using IBM's Qiskit and classical Scipy optimizers which provided similar results to the example code found in the Paddle Quantum repo in terms of state fidelity.

This implementation can be found in my personal repo.

Personal Motivation

While exploring and working through the methodology presented in this paper, I found it fascinating to observe and implement these concepts, ultimately obtaining meaningful results. I also believe that establishing functional platforms for experimentation and analysis is crucial. However, I was unable to find a similar tool in other quantum software libraries or SDKs comparable to the one available in Paddle Quantum. Since no such tool existed, I decided to try and build it myself to bridge this gap and enable further research in this area.

Additionally, I have been actively working with Classiq, gaining valuable experience since last year's hackathons and the Classiq Mega Challenge. This new opportunity excites me, and I look forward to further collaboration with the Classiq Team.

Please don't hesitate to reach out for further discussion of this project.

Cheers.

[NadavClassiq]

Hello @FGranda2!

First of all, thank you! We appreciate your interest in implementing and collaborating with Classiq!

From a brief look at the paper, it seems interesting. CLassiq's main power and advantage is in large-scale implementation (big circuits), which doesn't seem to be the case here, but we are still curious about where you will take it.

It would be really interesting if you could make a larger implementation and demonstrate the same concepts in the paper. For the beginning, I recommend following the paper and starting small.

Please let me know if you have further questions.

[FGranda2]

Hi @NadavClassiq

Thank you for your feedback! I agree that following the paper is a good starting point, as it provides a solid foundation for building a platform that can later handle more complex and larger implementations.

The number of shared entangled pairs can be scaled to explore its effects on fidelity and the success rate of distillation. Additionally, the LOCC operations will function as a generic ansatz, so even a small-scale implementation should enable the exploration of larger circuits with more layers and parameters.

I'll begin with the basic implementation and follow up with ideas for scaling the problem.

Thanks again!

[NadavClassiq]

Sounds great @FGranda2!

Are you familiar with our PyTorch integration? it might be handy. Good luck!

[TaliCohn]

Hi @FGranda2, what is the status of this? Are you still working on the implementation?

[FGranda2]

Hi @TaliCohn.

Currently, I am still working on the implementation. However, due to work and academic commitments, I won’t be able to complete the project by the original deadline. I plan to continue working on it as a side project in the coming months when I have more time available.

Thank you.