grothendieck-rising-sea
Build an extensive mathematical theory around a hard conjecture: develop the objects, generalizations, constructions, examples, invariants, and connections until the conjecture can become a natural co
SKILL.md.claude/skills/grothendieck-rising-sea/SKILL.md
GitHubGrothendieck Rising Sea
Treat the conjecture as an invitation to understand a mathematical world. Develop that world far enough that the conjecture can eventually be situated in a network of structural results and become a natural consequence. The immediate research question is what theory these objects deserve, including questions the original formulation does not yet make visible.
This is a sustained theory-building mode. It can be chosen from the outset; it need not wait for direct proof attempts to fail. Substantial periods may produce no identifiable reduction of the conjecture. A direct return can remain unknown while the surrounding mathematics becomes richer and better understood.
A fruitful new theory and its unexpected results can be as important as, or more important than, the initiating conjecture. Contributions beyond its original scope are desirable outcomes in their own right, including when the conjecture remains open. Actively look for them as the theory develops.
Organizing Principle
Let the objects and their natural relationships guide the development. Seek a coherent picture: central examples, natural operations, general classes, structural theorems, exceptional cases, comparisons, and questions that connect them. Allow the theory to broaden and reorganize its own foundations as this picture emerges.
The useful abstraction may be algebraic, geometric, analytic, probabilistic, combinatorial, or categorical. Categories and universal properties are possible tools, not required destinations. Abstract to expose and develop structure; a change of vocabulary alone supplies no mathematical understanding.
Do not require a candidate master theorem, a known implication to the conjecture, a cheapest decisive test, or a near-term application before substantial theory development. The right definitions and central theorems may themselves be among its eventual discoveries. Foundational lemmas can be essential infrastructure whose uses become apparent only after other parts of the theory have matured.
Develop The Surrounding Theory
Use the following as a recurring research cycle, not a sequence of gates that must be passed before work may continue.
- Map the mathematical territory. Start from the objects, their examples and sources, natural questions, known theories, and neighboring subjects. Include the initiating conjecture, but do not let its latest auxiliary obstruction determine the whole agenda. Search the shared extracted literature before web searches, and distinguish existing foundations from prospective additions.
- Enlarge the setting. Vary assumptions, parameters, ambient structures, and notions of equivalence. Include natural boundary cases and degenerations. Seek classes closed under meaningful constructions and families that explain several phenomena together. There need not be a single smallest framework.
- Build foundations in depth. Develop definitions, examples, morphisms where useful, constructions, invariants, existence results, and basic laws. Work out restrictions, extensions, products, quotients, dualities, limits, or gluing as the objects warrant. Prove the foundational results needed to make these tools reliable, even when they do not improve a bound in the conjecture.
- Follow the theory's own questions. Study representation, classification, decomposition, obstruction, comparison, and realization problems. Ask which hypotheses are essential, which phenomena extend, and what an exceptional example teaches about the general setting. A question can deserve serious development because it explains the objects, without supplying a proof route.
- Connect and extend. Develop relationships with other theories, including translations of objects, invariants, and constructions in both directions. A connection may require substantial work on both sides before it transports a theorem. Let several interdependent parts of the theory mature together; do not force them into a single sequence of short sufficient lemmas.
- Consolidate the picture. Integrate results into a coherent account, expose dependencies and missing foundations, reconcile definitions, and generalize proofs when their mechanism suggests it. Revisit earlier examples with the new tools. A local result may acquire significance through these later connections. Periodically assess independent contributions and opportunities for a standalone paper or a distinct exploration branch, as described below.
- Revisit the conjecture at natural milestones. Ask whether it now fits a classification, becomes a comparison statement, or follows from several independently developed results. Prove any claimed implication carefully. If no such connection is known, record that fact and continue the worthwhile theory work within the user's scope and resources.
For analogies and a worked research pattern, read references/examples.md.
Use alon-method when an external technique helps; in this mode, developing the
bridge to another subject can itself be a substantial mathematical project.
Aim For A Coherent And Extensible Picture
Keep asking what is still missing from an understanding of the objects: which constructions are unexplained, which examples lack a common account, which invariants omit important information, and which generalizations reveal a new phenomenon. The theory should develop its own landscape of results and questions, not just collect lemmas adjacent to the initial statement.
Useful directions include:
- enlarging an object class to accommodate natural operations and limiting cases;
- finding representations, dual descriptions, or universal constructions;
- developing invariants together with their behavior under those operations;
- explaining local and global phenomena, including failures of assembly;
- relating previously separate examples, classifications, or mathematical subjects;
- creating tools that support further questions not anticipated at the outset.
Pursue completeness relative to an evolving mathematical scope. This does not mean proving every statement about every generalization. Depth, breadth, and integration matter together; an elaborate formalism without explanatory or mathematical substance is not the intended outcome.
Recognize Independent Contributions
At consolidation reviews and natural milestones, ask what the theory now makes possible beyond the initiating problem: new structural results, useful constructions, connections between subjects, or a family of interesting questions with its own mathematical motivation. Assess their significance independently of whether they help prove the original conjecture.
When a contribution warrants it, propose a standalone paper or a distinct theory branch. State its central question, mathematical scope, strongest results and their proof status, relationship to existing literature, and natural next questions. Check the shared literature and relevant prior work before claiming novelty. A promising branch need not already have a publication-ready theorem; a paper proposal should distinguish established contributions from work still needed. Record these opportunities in the project artifacts so they survive handoffs.
Favor branches that develop explanatory principles, reusable tools, or independently interesting mathematics. A growing census of cases or a case-by-case decomposition of the original conjecture is not, by itself, a fruitful new theory or a reason to open another branch. Enumeration can inform the theory, and a structural classification can be a major contribution; assess what unifies or explains the cases and what new understanding follows. Redirect expanding casework when it supplies neither such understanding nor a substantive independent question.
Assess Progress On The Scale Of Theory Development
Maintain two distinct accounts:
- Development of the theory: what can now be constructed, proved, compared, classified, or understood; which examples or phenomena have become connected; and which foundations or important questions are now accessible.
- Status of the initiating conjecture: what is actually implied, which implications are conditional, and where the connection remains unknown.
Assess development cumulatively. A foundation, example, or generalization can matter through its place in the emerging theory rather than through immediate standalone novelty. Do not require each turn to yield a theorem, each definition to remove an old obligation, or each branch to show a short path to the target. An increasingly coherent theory may justify sustained investment despite an unchanged conjecture status.
Use reviews to synthesize, redirect, and decide where depth is needed. A review
is not an automatic expiry date. Generic short discovery allowances or counts
of unsuccessful direct attempts must not terminate a theory-building program
merely because a payoff for the conjecture has not appeared. If using
honest-conjecture-resolution, retain its correctness and claim-accounting
standards, while using these criteria for the explicitly chosen theory program.
Respect actual user budgets, deadlines, and scope; this mode does not authorize
unlimited execution or guarantee eventual resolution.
Redirect when there are mathematical reasons: constructions remain artificial, examples do not support the proposed generality, proofs repeatedly recover only what was encoded in the definitions, or disconnected elaboration fails to deepen the picture. Locate the issue and improve the theory accordingly. Lack of a known return to the conjecture, by itself, is not such a reason.
Preserve Mathematical Honesty
Distinguish proved results, conjectural structure, instructive examples, and analogies. The ambition to build an extensive theory does not relax proof standards. Test definitions on varied examples, prove claimed laws, and use counterexamples to refine scope. An elegant proposed formalism may fail.
If an invariant cannot distinguish two relevant phenomena, identify what this rules out about that invariant's proposed use. A failed comparison can reveal missing structure without invalidating the rest of the theory. An equivalent reformulation is not automatically a simpler proof. Never claim the original conjecture is closer to resolution merely because the exposition has grown.
Working Output
When asked to pursue this approach, actually develop the theory within the available work period: construct examples, formulate definitions and natural questions, prove foundational or structural results, and integrate connections. Do not stop at a list of proposed abstractions or a programmatic manifesto.
Maintain an evolving theory document with motivations, definitions, examples, proved results, open structural questions, and connections to existing work. Organize it by mathematical dependencies, not the chronology of attempts. It may grow into several coherent parts or papers as the mathematics warrants. Keep research-specific content in the project's artifacts, not in this skill.
At a handoff, state what was developed, why it contributes to the surrounding picture, the natural next work, and what is known or unknown about its relation to the conjecture. Include any promising independent contributions and concrete paper or branch proposals. A valuable theory is a valid outcome even if the initiating problem remains open; resolving that problem still requires its own proof.