Protein folding & medicine
A protein’s shape determines much of what it can do. We study the physics of folding and binding to understand disease at the molecular level and guide the design of treatments.
Recognition Physics Institute / Science from first principles
Build a world
of abundance.
Abundant energy, a world free of disease and a new kind of intelligence.
Understanding nature should change what we can build. We develop the mathematics, experiments and engineering behind that ambition, with a long-term goal: to make matter designable, down to the atom.
Each circle is linked to all the others. Choose a pair to see how the connection holds.
Why these circles?
A connection that holds as the shape changes.
Every circle links every other. You can bend or stretch the loops without breaking those connections; separating them would require cutting one or passing it through another. These lasting relationships illustrate the loop record in Recognition Science’s account of why space has three dimensions.
The curves are exact Hopf fibres, with an artistic choice of arrangement, color and lighting.
Read the paper: From one generator to loop order on the three-cube
Our scientific foundation
Start with how nature works.
A detector absorbs a photon. A molecule changes shape. A plasma carries a current. Recognition Science seeks a common mathematical account of physical change, beginning with how events are distinguished, compared and recorded.
We derive the consequences of that account, check its proofs and test it against observation. Physicists, mathematicians and computational scientists work together to find where it explains a result and where more work is needed.
Explore Recognition ScienceApplied research
Fusion energy
Fusion power depends on controlling a hot, moving plasma. We study its behavior and develop the physics and engineering needed to turn fusion into a practical source of abundant energy.
Noa / Native Intelligence
An intelligent system should learn from a new situation and use that understanding in the next one. Noa explores how the principles of Recognition Science could support that lasting, adaptable understanding.
Quantum computing
We use Recognition Science to develop quantum algorithms and study how quantum states maintain the relationships that make computation possible, including in processors that work with light.
Selected research
What we are
discovering.
A theorem tells us what follows from its premises. An experiment tells us what happened. These three projects show how we use both, with AI helping to discover and verify new mathematics.
Mathematics · Published
A consistent way to measure difference.
If one length is twice another, the second is half the first. Our formula treats those two views equally. The paper proves that this formula is uniquely selected by its rules: a nonnegative measure, zero at a match, a law for combining comparisons and a fixed scale.
The result supplies a common measure of difference for the framework.
Experiment · Preprint
Measuring correlations in light.
In eight sequential passes on Quandela’s photonic processor, our team measured correlations in selected photon detections. The counts, analysis and code let other researchers inspect and reproduce the calculation.
The chart shows recorded scores and their sensitivity to readout imbalance. This work establishes a measurement reference; it does not certify entanglement.
Recorded Readout sensitivity
Bars show counting uncertainty.
AI discovery · Proof checked
More detail without losing the guarantee.
Physics calculations often divide space into a grid. Cambrian, our AI research system, produced a machine-checked proof of a bound on a cube that holds however fine the grid becomes.
Researchers can add detail to this calculation without weakening the bound, which depends on the cube’s size rather than the number of grid points.
Try the comparison ↓Close the comparison ↑
Recognition Science / 90 seconds
How a comparison
becomes a record.
When a detector absorbs light or a clock advances, something changes. Recognition Science begins with these physical events and asks how their differences can be compared and carried forward.
Notice a physical change.
A light detector changes when it absorbs a photon, leaving a difference between its earlier and later states. In Recognition Science, “recognition” means a physical process that makes such a difference available for comparison, without requiring a conscious observer.
Measure the difference.
If one length is twice another, reversing the comparison gives one half. The formula assigns both the same score, called the comparison cost. Move the slider to see how the score changes as the lengths move closer together or farther apart.
Carry the information forward.
To compare a new event with an earlier one, some information about the earlier event must remain. Recognition Science studies the rules these records must follow and uses them to develop arithmetic, geometry and models of the physical world.
The same difference, either way
A
B
Ratio A / B2
Comparison cost0.25
Twice as much and half as much both have a cost of 0.25. At a perfect match, the cost is zero.
J(x) = ½(x + 1/x) − 1
The published proof explains why the rules for combining comparisons and setting their scale select this formula. Here we compare lengths; applying it elsewhere means identifying the quantities being compared and how they are measured.
See how these ideas develop into the wider theory.
Explore the full introductionHow our work fits together
The scientific foundation
Recognition Science
Recognition Science is our unified mathematical account of nature. It starts by asking how physical events can be distinguished, compared and recorded, then develops the consequences for mathematics and physics.
Explore the scienceMathematical discovery
Cambrian
Cambrian helps researchers discover new mathematics. Its AI agents develop results and proofs, which are checked by Lean, software designed to verify each logical step.
See the discoveriesNative Intelligence
Noa
Noa explores a new kind of intelligence built from the principles of Recognition Science. We call it Native Intelligence: a system designed to learn from experience and build a lasting understanding of the world.
Explore our research directionsResearch
Explore the research.
How do we compare quantities? What follows for gravity, matter and quantum behavior? These papers develop the answers. Follow an entry to its publication details, or explore the library by question.
A consistent way to compare
Shows why the paper’s rules for combining comparisons and setting their scale select a unique formula. It treats twice as much and half as much as the same degree of difference.
Building numbers from first principles
Develops counting, arithmetic and richer number systems from the act of making a distinction and keeping a record. Each stage makes clear which logical rules it needs.
The geometry of comparison
Compares several quantities at once and derives the resulting geometry. The measure of difference depends on one underlying direction in the space of comparisons.
Gravity from discrete information
Derives Newton’s law of gravity, with a small correction, from a model built from discrete information. It compares the result with galaxy rotation without adjusting a separate parameter for each galaxy.
Patterns in liquid water
Uses the golden ratio to organize the timescales and vibrations of water’s hydrogen bonds. A measured timescale sets the scale, and a second measured relationship connects the model to the liquid.
Testing quantum correlations with light
Reports photon correlations measured in eight sequential passes on a commercial processor. The archived counts and code provide an inspectable measurement reference, with an analysis of readout imbalance and the limits of the experiment.
From the institute
Latest discoveries.
New papers, experiments and verified mathematical results from across our research.
- arXiv preprint
Why the order of steps matters.
Two journeys along a cube’s edges can have identical transition tallies yet leave different loop records. The paper identifies the information these tallies lose, finds a shortest 14-step loop whose signed counts cancel but whose order survives, and shows how an unbounded stack preserves the reduced loop word.
- Proof verified
A finite bound for an infinite calculation.
A theorem produced by Cambrian is used in a checked proof that bounds accumulated approximation error across an infinite series, under stated size and error conditions.
- Preprint submitted
Testing how atomic patterns carry into molecules.
A study of 52 molecules and ions tests where a geometric description of atoms can predict molecular charge, and where it needs more structure.
- Preprint submitted
A reference experiment with light.
Measurements from eight sequential passes on a cloud photonic processor are released with the counts, code and analysis needed to inspect the result.
Our people
Different disciplines.
A shared ambition.
Our work connects mathematical foundations with molecules, plasmas, light and computation. The team brings the expertise needed to move between a formal result and a physical experiment.
Collaborate with us
Bring a question worth solving.
We welcome researchers and partners with a problem our science could help address, or an experiment that could put it to the test.
jon@recognitionphysics.orgResearch together
Bring a question, a mathematical idea or an experiment that could advance the work.
Discuss a research ideaExplore a partnership
Find where our work in energy, medicine, intelligence or quantum computing could connect with yours.
Start a conversation