About theory-formalism-translator
A mathematical reasoning skill for researchers translating concepts between formalisms. It maps structural objects, distinguishes equivalence from analogy, and identifies what is preserved, lost, or newly tractable.
Key features
- 01Precise description of source formalism
- 02Mapping of structural objects
- 03Separation of equivalence and analogy
- 04Analysis of preserved and lost structure
- 05Identification of newly available tools
Use cases
- 01Translating a model into another mathematical framework
- 02Checking whether two concepts share real structure
- 03Finding analysis tools enabled by a translation
Install
Copy the command and run it in a terminal.
npx skills add https:/ / github.com/ jurgendn/ agent-skills --skill theory-formalism-translatorAfter installing, type/theory-formalism-translatorin Claude Code to confirm it loaded.
Details
From the project READMEFormalism Translator
Look for structural equivalence, not superficial analogy.
The goal is not to rename concepts.
The goal is to determine whether:
- two formalisms encode the same invariant structure;
- one formalism exposes stronger tools;
- or the similarity is only metaphorical.
Procedure
1. State the Original Formalism
Describe the concept in its native language.
Specify:
- objects;
- operators;
- constraints;
- dynamics;
- optimization target;
- probabilistic interpretation if applicable.
Example:
Original formalism:
Random walk on a graph with transition matrix P,
analyzed through mixing time and spectral gap.
Avoid vague summaries such as:
Diffusion process on networks.
2. Identify Core Structural Objects
Extract the mathematical structure underneath the notation.
Look for:
- state space;
- energy or objective function;
- conservation law;
- invariant measure;
- transition operator;
- geometry;
- symmetry;
- equilibrium condition;
- entropy;
- curvature;
- flow;
- distance metric.
Example:
Core objects:
- Markov operator P
- stationary distribution π
- spectral gap γ
- diffusion dynamics
- local-to-global information propagation
This step matters more than terminology.
3. Define the Translation Target
State the target formalism explicitly.
Possible targets:
- optimization;
- probability;
- information theory;
- statistical physics;
- dynamical systems;
- control theory;
- graph signal processing;
- variational inference;
- optimal transport;
- spectral theory.
Example:
Target formalism:
Optimal transport interpretation of graph diffusion.
4. Map Objects Across Formalisms
Construct an explicit correspondence.
Example:
| Original Formalism | Target Formalism |
|---|---|
| Random walk transition | Transport kernel |
| Stationary distribution | Equilibrium measure |
| Spectral gap | Contraction / mixing rate |
| Diffusion distance | Wasserstein-like geometry |
| Community boundary | Transport bottleneck |
Do not skip mismatches.
If the mapping is only partial, say so.
5. Identify Preserved Structure
State what survives the translation.
Examples:
Preserved:
- equilibrium structure
- diffusion locality
- entropy decay
- conservation of probability mass
Distinguish:
- exact equivalence;
- asymptotic equivalence;
- heuristic similarity.
6. Identify What Is Lost
Every translation discards information.
Examples:
Lost:
- discrete graph combinatorics
- exact path interpretation
- local neighborhood semantics
- node-level interpretability
Do not present the target formalism as strictly superior.
7. Identify What Becomes Easier
Ask whether the new formalism provides:
- stronger theorems;
- cleaner optimization;
- better geometry;
- sharper concentration;
- stability analysis;
- computational advantages;
- alternative intuition.
Example:
Optimal transport formalism may provide:
- metric geometry tools;
- contraction analysis;
- curvature interpretation;
- continuous relaxation methods.
8. Check Whether the Translation Is Genuine
Distinguish between:
- structural equivalence;
- reusable machinery;
- notation change only.
Bad translation:
Graph diffusion is basically attention.
Better:
Both can be interpreted as weighted information propagation operators,
but attention is data-adaptive and non-Markovian,
while graph diffusion is typically fixed and stochastic.
9. Suggest New Tools Enabled by the Translation
If the translation is useful, identify concrete tools now available.
Examples:
| New Formalism | Potential Tools |
|---|---|
| Optimal transport | Wasserstein geometry, Sinkhorn |
| Dynamical systems | Stability analysis, Lyapunov functions |
| Information theory | Entropy bounds, mutual information |
| Spectral theory | Eigenvalue perturbation |
| Control theory | Controllability, observability |
If no genuinely new tool appears, state that explicitly.
Rules
- Do not claim equivalence unless the mapping is explicit.
- Surface mismatches, not just analogies.
- Distinguish:
- exact equivalence;
- asymptotic similarity;
- heuristic resemblance.
- Preserve assumptions during translation.
- Do not confuse shared notation with shared structure.
- If the translation only renames concepts, say so.
- Prefer explicit operator/object mappings over narrative analogy.
Output Format
# Original Formalism
State the original framing precisely.
# Core Structural Objects
List the key operators, invariants, constraints, and dynamics.
# Target Formalism
State the destination framework.
# Structural Mapping
| Original | Target | Notes |
|---|---|---|
| ... | ... | ... |
# Preserved Structure
- ...
# Lost Structure
- ...
# What Becomes Easier
- ...
# Mismatches / Non-Equivalences
- ...
# New Tools Enabled
- ...
# Verdict
Choose one:
- Deep structural equivalence
- Useful reframing
- Partial analogy
- Mostly notation change
Reason:
...