Stationary sets and club filters
Use regular uncountable cardinals to distinguish unboundedness, closed-unbounded sets, and stationary intersection principles.
On this page
The question for today
What you’ll understand
- Define clubs and stationary subsets at a regular uncountable cardinal.
- Explain intersection and diagonal-intersection closure of the club filter.
- Apply Fodor’s lemma with its stationarity and regressiveness hypotheses.
Unboundedness alone does not make a subset of an uncountable ordinal unavoidable. Closed-unbounded sets provide a stronger notion of largeness, and stationary sets are those that meet every such test set. Their intersection principles use regularity in an essential way. Working in ZFC at a regular uncountable cardinal keeps the hypotheses visible and prevents a finite cyclic picture or a countable-limit example from being mistaken for the same combinatorics.
Closed and unbounded impose different conditions
Fix a regular uncountable cardinal κ. A subset C is unbounded if above every ordinal below κ there is a larger member of C. It is closed if every nonzero limit δ<κ for which C∩δ is unbounded in δ belongs to C. A club satisfies both requirements. Closedness here concerns limits below κ; it does not require κ itself to belong to a subset of κ. Tails are clubs, and the nonzero limit ordinals below ω1 form a club.
A subset S of κ is stationary if it meets every club, and nonstationary if some club misses it. Every stationary set is unbounded because it must meet each tail, but an unbounded set need not be stationary. The successor ordinals below ω1 are unbounded and disjoint from the club of nonzero limits. The club filter consists of the subsets of κ containing a club, rather than only the clubs themselves. It is proper because a club cannot be empty under the unboundedness requirement.
Regularity keeps intersections below the boundary
The intersection of fewer than κ clubs is a club. Closedness is preserved directly. To obtain an element above a chosen starting point, repeatedly choose larger points meeting the required clubs, then take a limit still below κ. Regularity ensures that all suprema used in a construction of length smaller than κ remain below κ; uncountable cofinality lets the final countable approach stay below the boundary. This makes the club filter closed under intersections of fewer than κ members, which is often called κ-completeness.
An intersection of κ many clubs can be empty, as shown by taking increasingly late tails. The replacement is their diagonal intersection: δ belongs when it lies in every club whose index is below δ. This diagonal intersection is again club for a κ-long sequence of clubs. It asks each point to satisfy only its earlier tests, while ordinary intersection asks it to satisfy all tests at once. That distinction is the mechanism behind the pressing-down argument, not a harmless change in notation.
A regressive function repeats on a stationary set
Fodor's lemma says that if S is stationary in κ and f maps S into κ with f(α)<α for every nonzero α in S, then f is constant on a stationary subset of S. Removing zero from S does not affect stationarity. Suppose every fiber were nonstationary and choose a club Cξ missing the ξ-fiber. A member δ of S in the diagonal intersection must lie in C at index f(δ), since that index is below δ. But δ also belongs to that very fiber, a contradiction.
The strict inequality in regressiveness and the stationarity of the domain both matter. An injective regressive function on the unbounded successor ordinals supplies no stationary constant fiber, because that domain is nonstationary. Likewise the identity function on a stationary set is not regressive. These examples distinguish the theorem from a simple claim that a large domain forces repetition. The proof uses the club filter's normal diagonal behavior, which depends on the regular uncountable setting and cannot be inferred from recurrence in a finite cycle.
Work it through
Countably many colors cannot all avoid stationarity
Let S be the nonzero countable limit ordinals below ω1, and assign each α∈S a natural-number color f(α). Prove that one color occurs on a stationary subset.
S is closed and unbounded in ω1: β+ω is a countable limit above any countable β, and a nonzero limit of its limit points is still a countable limit. Therefore S is stationary.
Every α∈S is at least ω, so f(α)<ω≤α. The coloring is a regressive map into ω1 and satisfies Fodor’s hypotheses.
Fodor yields a stationary fiber. Equivalently, if every natural-number fiber missed a club C_n, the countable intersection of these clubs would still be club in ω1 and could not meet S, a contradiction.
Some single natural-number color is assumed on a stationary set, rather than merely at infinitely many points.
The living notebook
Explore the model
Interactive lab · modular
Test why cyclic recurrence is not a club condition
Pause and Reset, then set modulus m = 12 and additive step a = 3. List the orbit {0,3,6,9}. Reset and set both m = 12 and a = 4, obtaining {0,4,8}. Predict their intersection is only {0}. Both paths recur forever in playback, but their intersection supplies no upward progress in an ordinal order. Use this calculation to reject the proposed proof that recurrent finite orbits behave like intersecting clubs; the instrument has no uncountable order or stationarity test.
- Residue r0
- 0
- Orbit period · m/gcd(m,a)
- 12
- Distinct residues visited
- 1 of 12
Addition modulo m starts at zero: rₙ = na mod m. Playback advances two additions per second. The orbit has period m/gcd(m,a), including period one for a = 0. This is not a cryptographic security test or a primality algorithm. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forRecurrence on a circle is insufficient for the closed-unbounded hypotheses used by the club filter.
Pause & reason
Check your understanding
Your turn
Try it yourself
Let T={α+1:α<ω1} and define g(α+1)=α. Is T stationary? Is g regressive? Explain why its singleton fibers do not contradict Fodor.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Find a club disjoint from T before checking the inequality at each successor.
Compare with the solution
T is unbounded but nonstationary, because the club of nonzero limit ordinals is disjoint from it. At every point α+1, g(α+1)=α<α+1, so g is regressive. Each fiber is a singleton and therefore nonstationary, since a tail above that singleton is a disjoint club. Fodor does not apply because its stationary-domain assumption fails. This example also shows why a proof that substitutes unboundedness for stationarity would make an incorrect conclusion despite having the correct strict inequality for the function.
Keep the idea
Take it with you
- Clubs are closed and unbounded; stationary sets meet every club and need not themselves be clubs.
- The club filter is closed under fewer than κ intersections and under diagonal intersections at regular uncountable κ.
- Fodor’s stationary constant fiber requires both stationarity and strict regressiveness.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.