Closed Strilanc closed 1 year ago
Thanks for these ideas.
Thanks, that solves my questions.
We don't yet ZX calculus because I don't know where it's useful for QEC
This is kind of a weird thing to hear as the reason to not use ZX. ZX is incredibly ridiculously useful for QEC. Here are a few examples of papers that fruitfully use the ZX calculus in some way, as part of researching or tooling for QEC circuits:
I literally use the ZX calculus every day as part of my own research. It is an incredibly useful bag of concepts and diagrams. Not using it would be like not using quantum circuits. For me the most useful things are the following:
In my opinion, anyone and everyone working in QEC should know ZX or be planning to learn ZX. It should be one of the first things taught in university courses on quantum computing, alongside quantum circuits and honestly maybe even INSTEAD of quantum circuits. Not knowing ZX is like trying to do math the way they did it in the 500s, before algebraic notation was invented. Sure you can do it, but it kinda sucks.
Thanks, it was good to know this. I've peppered it many of those refs into the surface code entry, which is our most popular one. This should help make more people aware.
Turns out I already had one ref where I wrote ZX-calculus with the dash; right now, it's written without.
Recent work could be of interest to you; it feels like ZX should be useful for such circuit codes: https://errorcorrectionzoo.org/c/spacetime_circuit
I have a [5,1,3] encoder with fewer two-qubit-gates than the referenced papers. How do I add it to the [5,1,3] page? Specifically, how do I add images and external URLs and code blocks?
Quirk link
There are also many highly symmetrical ZX graphs of the encoder (up to permutation and single qubit basis change of the outputs). Examples:
Presumably there are ways to add linkable versions of these things, instead of just references to books and papers? Books and papers are great and all, but it's often difficult to find the specific figure and the representation is usually not computer parseable.