Structuresin development
VSEPR as computable geometry
A model in which shape is not an illustration added after the argument, but the outcome of constraints, competition, and admissible configurations.
What can be tested →An independent research project
We are not looking for one more theory of everything.
We study connections.
UCMT is the hypothesis that a single compressible medium underlies physical phenomena, while known laws describe different modes of its behaviour. We ask whether different laws of physics can be reduced to one property of that medium: its compressibility. The project is open to physicists, mathematicians, and anyone curious about how different parts of the physical world may be connected.
If everything is indeed connected, what rules make some structures possible, others impossible, and still others observable?
01
The project’s turn
From a medium to rules of connectedness
Where we started
Can light, gravity, particles, and other phenomena be read as different regimes of one medium?
What we study now
Which constraints, admissible structures, and transitions recur across different problems — and can we make them computable?
The original idea has not been discarded. It has become a starting hypothesis rather than a ready-made answer. What matters now is not the elegance of a single explanation, but whether the same network of relations survives different tests.
02
Computable objects
Models and sandboxes
This is where the theory stops being only a conversation. Each sandbox is a way to ask a system a concrete question — and receive an answer that does not always agree with our expectations.
Dynamicssandbox
The three-body problem
Not a promise of a universal solution, but a study of which signatures of a regime are genuinely stable and observable.
Open the model card →Morphologyworking model
Structural sandbox
A move from pictures to graphs, state classes, and reproducible corpora of transitions.
See how it works →03
Method and instrument
CQT: testing the preservation of meaning, not the resemblance of pictures
CQT grew out of an internal research discipline. It helps us state explicitly what counts as observable, which transformations are admissible, and when a comparison must honestly return INVALID.
- 01Define the observable
Do not hide the choice of features inside an attractive visualisation.
- 02Fix the admissible transformations
What can be discarded without destroying the subject of comparison?
- 03Check with an independent oracle
A classification should not be allowed to confirm itself.
- 04Preserve failure
INVALID is not an interface error. It is a legitimate scientific result.
The project does not need to look infallible. It matters more to keep the path visible: useful guesses, dead ends, changes of language, and the moment when a beautiful idea first becomes testable.
That is why working notes, model histories, and honest limitations remain beside the publications. Theory here is not a monument. It is a workshop.
04
Read and verify