Quantum Computing
Coin Tossing paper
I published this paper 15 years ago. Back then, the world of Quantum Information was quite different, and Quantum Computing was more of an hypothetical long-term scenario.
The paper was a practical demonstration of what one could nowadays be tempted to call 'quantum supremacy'. For a specific problem -the tossing of a single coin- we could build as a tabletop experiment a protocol which in some way was more secure than what any classical protocol could do.
The coin-tossing problem is simple: Alice and Bob live far away, they don't trust each other (nor any third party), but they want to generate a random classical bit together. They need a protocol which guarantees that none of them can cheat, or at least not too much.
This was implemented with optical fiber, a technology which already allowed to generate extremely easy some quantum states, make them interfere which each other, entangle them if we want, transport them far away, store them for a relatively long time if needed, and of course measure them. The main technical limitation was the low efficiency of the single-photon detectors.
Today, I decided -mostly for fun- to reproduce this protocol using a well-known Python package for Quantum Computing: Qiskit. The notebook can be found below. There isn't a perfect mapping between the paper and the notebook. In particular, in our protocol it's quite important that the laboratories of the two players are separated (even if the proof of security does not really rely on any relativistic argument), and that was something easy to reproduce in optical fiber. However it is obviously not the case in this Qiskit notebook, which is an illustration rather than a real-world application. On the other hand, Qiskit allows to do things very fast and to explore the effects of changes here and there.
The main results of this notebook would be the Kitaev biases, since my explicit computation with specific strategies does not go very well with the merit criterion we developped in the paper, which was not directly measured but instead reconstructed from the characteristics of the system.