Oopsy, a latex error snuck through. Too late to fix now, but
thanks for pointing it out!
18.02.2026 08:29
π 2
π 0
π¬ 0
π 0
Oopsy, a latex error snuck through. Too late to fix now, but
thanks for pointing it out!
Despite this austerity, the model is universal: it densely approximates all unitaries and extends to measurement and classical control. Same computational power as the standard model, but with cleaner foundations.
The model is fully discrete: finite generators, finite equations, no continuous parameters. That enables exact symbolic reasoning. Crucially, the axioms are complete: all valid equalities are derivable.
The qualitative gap between classical and quantum computing is isolated sharply: the ability to take well-behaved square roots; operationally speaking, the ability to stop some computations halfway.
The answer is a free model. "Free" means minimal and universal: any quantum model satisfying the axioms factors uniquely through it. Nothing extra is assumed beyond what is forced.
Quantum computing is usually framed in complex linear algebra. This paper asks: what is the least you need to add to reversible classical computing to recover quantum computation?
www.pnas.org/doi/10.1073/...
@manchegobaby.bsky.social
Want to shape the global direction of quantum software? We're open to new member organisations! See quantumsoftwarealliance.org/documents/ch...
One specific goal is to keep an unbiased Global Quantum Software Research Agenda: quantumsoftwarealliance.org/documents/gl...
Quantum Software Alliance logo
At the closing of the UN International Year of Quantum today, the Quantum Software Alliance was born! Founded by institutions from 15 countries over 4 continents, QSA will champion and co-ordinate the development of #quantumsoftware: quantumsoftwarealliance.org
Research Associate position available at the University of Edinburgh Quantum Software Lab to work on a joint project on orchestration of heterogeneous quantum computers with Fujitsu. Apply here: elxw.fa.em3.oraclecloud.com/hcmUI/Candid...
Six weeks left to apply: 16 fully funded Quantum Informatics PhDs
quantuminformatics-cdt.ac.uk
The number of low-quality or fraudulent publications is rising to hundreds of thousands per year. It is time to reevaluate current publishing models and outline a global plan. Read the 'Reformation of science publishing: the Stockholm Declaration': royalsocietypublishing.org/doi/10.1098/... #RSOS π§ͺ
The equations governing controlled circuits.
Our notation implicitly assumed one more equation, (h) below, now added. Thanks to Peter Selinger for catching this!
Sheaves over such causal coverages now describe how to assign data to your spacetime such that "past data evolves continuously and deterministically into future data". For much more, see Nesta's great PhD thesis, with many fun ideas and pictures: arxiv.org/abs/2406.15406.
Illustration of a path of points regarded as an infinitesimally thin approximation by sequences of opens that causally precede each other.
That means you need a way to think about paths in terms of open sets rather than points. The idea: approximate a path by finer and finer finite sequences of opens that are causally before each other.
Well, you don't need points to describe that, and can do it purely with open sets. This new paper with Nesta van der Schaaf describes how: arxiv.org/abs/2510.17417. The idea is in the picture: region U is covered by collection A of regions if any causal path that ends up in U must pass through A.
Illustration of causal coverage: region U is covered by regions A because any path that ends in U must pass through A.
Coverages, or Grothendieck topologies, describe when a collection of regions covers another region in a space. But what if your space is a spacetime? The topological space may have a causal order saying which (spacetime) points 'come before' which other points, for example.
Also about circuits: the physics Nobel prize just announced!
So much material for titles here, too! "If you liked it you shoulda put a rig on it", "The rig is up", "Taking back control". What's your favourite?
Theorem: these two ways to build a theory of controlled circuits are the same! Practically, this means the equations are complete, and we can use them to manipulate and optimise controlled circuits. Foundationally, this pins down what this not-quite-data-or-control-flow control really is about.
Another thing you can do is make a new theory where 'controlled gates' are, roughly, matrices of base gates. Technically, you freely adjoin sums to the base (tensor) prop, universally giving it rig structure.
The equations all have natural interpretations. For example, the complementarity equation (e) says that a gate on the target wire is the same as a positively controlled and then a negatively controlled version of it: the control bit is either on or off.
The equations governing controlled gates.
Start with any base circuit theory, in terms of tensors only; technically, as a prop. Build a new controlled circuit theory that has controlled versions of the base gates, subject to these equations.
The key is that circuits contain controlled gates. It's not quite data flow, nor control flow, in the usual computer science sense. But it is clearly important. Can we take back this control, and separate this control from a base theory of uncontrolled circuits? Yes we can!
Have you ever wondered why we describe circuits with matrices? I mean, circuits are about tensor products, while matrices are about direct sums. This new paper with Louis Lemonnier and @manchegobaby.bsky.social gives a practically useful explanation: arxiv.org/abs/2510.05032.
A new quantum programming construct: more abstract than circuits, capturing important examples elegantly and in fact universal, yet simple and intuitive. With prototype compiler and clean categorical semantics to boot!
With Alex Rice, Chris McNally, and Louis Lemonnier:
arxiv.org/abs/2507.11676
I presented a taster poster on our upcoming work in distributed quantum compilation at the Uni of Edinburghβs Informatics internal research showcase this Tuesday.
Focus: our upcoming model/hardware-agnostic intermediate representation for compilation of quantum programs (quite a mouthful I know!)
A discussion of the success of Bell Labs as a research organization and why we have nothing like it anymore.
A sound and complete finite system of equations to manipulate Toffoli+Hadamard quantum circuits you say, so you can automate circuit optimisation, you say? Why of course, here you go: arxiv.org/abs/2506.06835, with Wang Fang and @manchegobaby.bsky.social.
90% of doing science is being open to new ideas.