Cogitan

Research

What we are trying
to find out.

Open questions, each with the result that would settle it. We state the settling condition before we start, because it is how we decide what to stop.

Open threads

3

What must a device paper report for its claims to be independently checkable?

The median published device reaches 7–8 rules of our set. Saturated with what its own field already knows how to publish, it reaches 28 — so the binding constraint is reporting convention, not physics or tooling. Growing the database from 16 to 60 superconducting rules barely moved the number.

Corpus intake gate, 18 published devices

Settled if

A minimal reporting set lifts median reach from ~8 toward 28 across a large corpus — and closing it changes verdicts, not just counts.

How tight is the structural floor on logical error for dissipative cats?

The ideal-code floor is a valid lower bound but runs 6–14× low against the dissipative cat. The missing factor is set by the confinement ratio κ₂/γ and buffer adiabaticity g₂/κ_b.

qudit-lab, exact Liouvillian numerics

Settled if

An effective gap for the dissipative cat closes the floor-to-rate distance across the κ₂/γ sweep — or is shown not to exist.

Does the readability limit extend to codes read by canonical phase measurement?

Our preprint bounds logical decoherence from a code's own structure, but the framework cannot express canonical phase measurement — which is how number-phase codes are conventionally read. So the result carries a scope restriction we state rather than hide.

readability.tex:472 — an open question the paper leaves, not work in progress

Settled if

An extension of the bound to phase measurement, or a proof that the restriction is essential.

Programmes

2

Directions, not projects. A finding on its own is a fact; inside a programme it is evidence for a direction, and the direction is what says why anyone went looking.

Construct validity

8 published · 1 open

Does the standard method measure what it claims?

The programme →

Closed

2

Directions that ended, with the reason. They stay here.

Published

How gently can a protected quantum memory be read? Rate limits on logical readout from the structure of the code

Separates the logical decoherence induced by weak continuous readout into a conjugate cost and a self cost, and lower-bounds the self cost from the code's own Knill–Laflamme structure. Verified cell-by-cell across cat and GKP codes.

Where it stops

Leading order in the measurement rate, single-mode Markovian limit, and the framework cannot express canonical phase measurement — which is how number-phase codes are conventionally read.

PDF →Code →Code & data (Zenodo) →

Instruments we gave away

Each one published with the limit it runs into. A limitation you have to go looking for is not really published.

RSFQ-JEPA

Neural surrogate predicting SPICE-level electrical properties and design-rule violations of RSFQ logic cells from layout and circuit graph, with calibrated uncertainty across 17 properties.

Where it stops

Ships with a gradient-boosted-tree baseline that beats it on Ic margin — 0.006 against 0.093. Validated against an oracle, not against silicon.

GitHub →

margo-sfq

Margin analysis and inverse design for SFQ logic cells — a JoSIM-verified oracle paired with a learned netlist surrogate for design-space search.

Where it stops

Margins are computed against idealised junction models, so they are simulator-level and relative. Useful for ranking and search; not fab-calibrated.

GitHub →

Work that does not need a team, we open to anyone.

Questions too small to staff still deserve answering. Those go to Cogitan Commons, where anyone can propose a group — and where every result gets published, including the ones that end the project.

commons.cogitan.ai →