mathematical-framework-design
Construct a new mathematical setting or a substantial two-way correspondence when existing objects, morphisms, or foundations obstruct useful constructions. Use for diagnosing structural failures, des
SKILL.md.claude/skills/mathematical-framework-design/SKILL.md
GitHubMathematical Framework Design
Make the setting itself a research object. Determine what objects, morphisms, notions of equivalence, and constructions would support the mathematics being sought. A replacement setting or a bridge between two theories can be a major contribution independently of the problem that motivated it.
In a campaign, honest-conjecture-resolution selects the objective and integrates
the results; this skill owns framework construction and comparison. Return the
mathematical output and transfer status to the existing campaign state.
Use grothendieck-rising-sea for sustained development of the resulting theory.
Use alon-method when an existing theorem and a one-way encoding suffice. This
skill addresses construction of the setting or correspondence itself, including
the substantial mathematics needed to make it work.
Identify The Structural Question
Inspect concrete examples, source proofs, and relevant existing theories in the campaign, sibling repositories, and shared literature. Describe the construction that is unavailable, fails, or repeatedly demands extra data. Distinguish:
- a mathematical obstruction in the chosen objects or morphisms;
- a representation that forgets essential information;
- inconvenient notation or a missing implementation;
- a promising correspondence whose mathematical content is still unknown.
Failure of a desirable categorical property is relative to a purpose, not proof that an entire subject has defective foundations. A category need not be abelian or Cartesian closed to support the intended mathematics. State why the missing property would help and what would remain difficult even if it were available.
Use representation-refinement for a focused loss-of-information diagnosis or
finite sufficient quotient. A repaired statistic need not become a new framework;
use this skill when constructing the setting or comparison is itself substantial.
Build A Setting Or A Bridge
Choose the direction suggested by the examples; these may develop together.
Repair or enlargement. Explore extra structure, a changed morphism class, quotients, completions, localization, test objects, or an internal language. Define the construction and its intended universal or comparison property. Build concrete models. Determine what original information is retained, what is identified or added, and what new operations become available. Simple changes may suffice; a sheaf category or new foundation is an option, not a destination.
Two-way correspondence. Use matching examples and invariants as clues, then construct maps on objects and morphisms with the correct variance. Work out what happens to composition, constructions, and relevant quantitative data. Specify whether the proposed result is an equivalence, duality, adjunction, embedding, correspondence on a restricted class, or only a useful translation. Matching invariants do not establish any of these. Prove the laws appropriate to the claim: for example, functoriality and natural isomorphisms for quasi-inverses, or an adjunction with its compatibility laws. The bridge may deserve extensive research before it can transport a theorem.
Read comparison obligations when choosing what must be proved for completions, categorical bridges, or new internal languages.
Prove The Comparison
Track separately the existence of the proposed setting, its internal structural theorems, the construction of intended models, and the transfer back to the original mathematics. Distinguish preservation from reflection: proving something about an image need not prove it about the original object.
Check behavior under the operations that motivated the change. Specify domains, size restrictions, regularity, exactness, and quantitative constants where relevant. Treat singular objects, degeneration, or failure of a proposed law as mathematical information about the framework's scope. Do not repair every counterexample by adding the desired conclusion as another axiom without explaining the loss.
A more convenient language can move an obligation to a comparison theorem; it does not remove it. Sheafification alone does not prove an analytic estimate, faithful recovery of arbitrary original objects, or exactness of a chosen completed tensor product. A qualitative compactness argument does not by itself supply effective finite bounds.
When a new logic or type theory is genuinely relevant, distinguish a notation or library extension from new axioms or kernel rules. Give semantics and identify the claimed soundness or conservativity theorem precisely. Existence of a model is not automatically conservativity over an earlier theory. Check the particular translation needed for the original statement. Proposed formal sketches must be labelled as such; do not change a proof assistant's trusted foundation as an incidental implementation step.
Develop And Assess The Mathematics
Work on actual definitions, examples, constructions, comparison proofs, and obstructions. A list of desirable axioms is an agenda, not a completed framework. It can nonetheless be a useful starting point for serious foundational work.
At consolidation points, explain which phenomena now have a common account, which constructions are justified, what original information is recoverable, and what remains conjectural. Account for the work of realizing models and proving comparisons, even if downstream proofs become short. Syntax length, graph centrality, and matching invariant data are discovery heuristics rather than evidence of correctness, equivalence, or significance.
Assess progress cumulatively and respect the user's scope and resources. Do not terminate a coherent theory merely because its return to the initiating conjecture is unknown. Seek independent structural results, useful constructions, and new questions; propose a standalone paper or exploration branch when they warrant it. A proliferation of names or case lists without deeper explanation is a reason to revise the direction. Record mathematical output and comparison status in the owning campaign's artifacts, keeping reusable guidance generic.