Top Cycle: Social Choice, Markets & Geometry

Updated 5 July 2026

Top Cycle is a concept defined by cyclic reachability that identifies the smallest dominant set or canonical cycle across social choice, market design, and arithmetic geometry.
It employs cyclic and balanced-trade procedures to resolve majority ambiguities and indifferences, ensuring strategy-proofness and efficiency in decision-making and allocation.
The approach underpins key applications, from robust social choice correspondences and fractional trading mechanisms to generating canonical top homology in Voronoi complexes.

to=arxiv_search _人人碰彩娱乐彩票{"query":"Top cycle social choice correspondence strategyproofness arXiv (Brandt et al., 2021) Top trading cycle fractional arXiv (Yu et al., 2020)", "max_results": 5} Top cycle denotes several distinct but structurally related notions centered on cyclic dominance or cycle-based construction. In social choice, the Top Cycle is a set-valued social choice correspondence that returns the maximal elements of the transitive closure of the weak majority relation; equivalently, it is the smallest dominant set in the majority graph (Brandt et al., 2021). In allocation and market design, “top trading cycle” refers to the deterministic Top Trading Cycle (TTC) mechanism and its fractional generalization, Fractional Top Trading Cycle (FTTC), which implements balanced trades of fractional endowments and extends to weak preferences through a labeling procedure (Yu et al., 2020). In a distinct geometric setting, a “top cycle” is an explicit canonical cycle in the top-dimensional homology of the Voronoi complex associated with arithmetic groups such as $SL_n(\mathbb{Z})$ , yielding a canonical generator of cohomology at the virtual cohomological dimension (Durán, 4 Apr 2026). The shared terminology reflects a common motif—closure under cyclic reachability or cyclic cancellation—rather than a single unified object.

In the social-choice sense, let $A$ be a finite set of $m \ge 3$ alternatives, $N$ a finite nonempty electorate, and each voter $i \in N$ have a strict total order $\succ_i$ over $A$ . For a profile $R$ , the majority margin of $x$ over $y$ is

$A$ 0

The weak majority relation is defined by $A$ 1 iff $A$ 2, with ties treated as bidirectional edges. Its transitive closure $A$ 3 captures reachability in the weighted majority graph (Brandt et al., 2021).

The Top Cycle, denoted $A$ 4 in the source, is defined by

$A$ 5

Thus, an alternative belongs to the Top Cycle exactly when it can reach every other alternative by a path in the weak majority relation. The same object is characterized as the smallest dominant set, where a nonempty set $A$ 6 is dominant if every element of $A$ 7 strictly dominates every element outside $A$ 8 (Brandt et al., 2021).

This definition is especially significant when pairwise majority comparisons are cyclic. If a Condorcet winner exists, the Top Cycle collapses to the singleton containing that winner. When there is no Condorcet winner, it identifies the “dominant core” of the majority graph rather than forcing a single outcome (Brandt et al., 2021). This suggests that the Top Cycle functions as a canonical Condorcet extension under cyclical majority structures.

2. Structural properties and axiomatic characterization

The Top Cycle is studied as a social choice correspondence (SCC), that is, a rule mapping each profile to a nonempty set of alternatives. The relevant informational restriction is pairwiseness: an SCC is pairwise if it depends only on the matrix of majority margins $A$ 9. The paper distinguishes pairwise SCCs from majoritarian SCCs, but proves that under the strategyproofness conditions considered, pairwise SCCs in the characterized class are necessarily majoritarian in effect (Brandt et al., 2021).

A central result is that, under non-imposition, homogeneity, neutrality, and pairwiseness, Fishburn-strategyproofness characterizes the robust dominant set rules. In the paper’s theorem:

Let $m \ge 3$ 0 be a pairwise SCC that satisfies non-imposition, homogeneity, and neutrality. Then, $m \ge 3$ 1 is strategyproof iff it is a robust dominant set rule. (Brandt et al., 2021)

Here, a dominant set rule always returns a dominant set with respect to the weak majority relation, and robustness requires that if the previously chosen set remains dominant under another profile, the choice set cannot expand. Because the Top Cycle is the smallest dominant set, it is the finest rule in this class (Brandt et al., 2021).

A stronger uniqueness statement is obtained through set non-imposition, meaning that every nonempty subset of $m \ge 3$ 2 must arise as an outcome at some profile. The source proves that the Top Cycle is the only robust dominant set rule satisfying set non-imposition, and therefore:

The top cycle is the only pairwise SCC that satisfies strategyproofness, set non-imposition, and homogeneity. (Brandt et al., 2021)

This characterization is presented as a set-valued analogue of the Gibbard–Satterthwaite theorem. Whereas single-valued strategyproof and non-imposing rules are dictatorial, the move to set-valued correspondences and Fishburn’s extension of preferences over sets isolates the Top Cycle rather than dictatorship (Brandt et al., 2021).

The relevant manipulation concept is Fishburn-strategyproofness. For nonempty sets $m \ge 3$ 3, Fishburn’s extension defines $m \ge 3$ 4 iff $m \ge 3$ 5 and $m \ge 3$ 6. Intuitively, every newly gained alternative must be preferred to every member of $m \ge 3$ 7, and every member of $m \ge 3$ 8 must be preferred to every lost alternative (Brandt et al., 2021). Under this extension, the Top Cycle is strategyproof, and robust dominant set rules are group-strategyproof as well.

The paper derives several structural implications of Fishburn-strategyproofness for pairwise SCCs: weak monotonicity, weak set-monotonicity, independence of unchosen alternatives, and weak localizedness. It also proves strong Condorcet-consistency under the mild axioms of non-imposition, homogeneity, neutrality, and pairwiseness (Brandt et al., 2021). These lemmas form the route from local incentive constraints to the dominant-set characterization.

The Top Cycle is neutral and homogeneous. It is Condorcet-consistent because $m \ge 3$ 9 whenever a Condorcet winner exists. It is also pairwise, although the characterization shows that the outcome ultimately depends only on the direction of majority edges rather than their magnitudes (Brandt et al., 2021).

The rule has well-known refinement and coarsening relations. The source states that Copeland’s rule, the uncovered set, the essential set, Kemeny’s rule, ranked pairs, and Schulze’s rule refine the Top Cycle. By contrast, the Condorcet rule that returns the Condorcet winner if one exists and otherwise all alternatives is a coarsening of the Top Cycle (Brandt et al., 2021). The paper also notes a drawback: the Top Cycle may include Pareto-dominated alternatives. To enforce Pareto-optimality, it proposes first removing Pareto-dominated alternatives and then computing the Top Cycle, producing a strategyproof but non-pairwise variant denoted $N$ 0 (Brandt et al., 2021).

A common misconception is that the Top Cycle requires strict majority edges only. The source explicitly rejects this: majority ties, i.e. $N$ 1, are treated as bidirectional edges, and this tie-handling is critical for the reachability-based definition and the equivalence with the smallest dominant set (Brandt et al., 2021).

4. Top Trading Cycle and fractional extensions in market design

A different use of the term arises in matching and allocation. The classical Top Trading Cycle mechanism is defined for housing markets with endowments and strict preferences. At each iteration, every agent points to her top remaining object, every object points to its current owner, cycles are identified, trades are executed along each cycle, and the corresponding agents and objects are removed. In this setting, TTC is Pareto efficient, individually rational, and strategy-proof (Yu et al., 2020).

The deterministic nature of TTC limits its ability to accommodate ex-ante fairness, which often requires randomization. The paper “Fractional Top Trading Cycle on the Full Preference Domain” develops the Fractional Top Trading Cycle mechanism, building on Yu and Zhang’s 2020 FTTC and extending it from strict preferences to weak preferences while preserving efficiency and fairness properties (Yu et al., 2020).

The formal environment allows agents $N$ 2, indivisible objects $N$ 3, capacities $N$ 4, and possibly non-strict preferences $N$ 5. A fractional endowment exchange model is used, with $N$ 6 the share of object $N$ 7 owned by agent $N$ 8, and total supply $N$ 9. A feasible fractional allocation is a matrix $i \in N$ 0 satisfying

$i \in N$ 1

Efficiency is expressed through Pareto efficiency and stochastic-dominance efficiency; fairness through equal treatment of equals, SD-envy-freeness, equal-endowment no-envy, bounded envy, and full envy-freeness in the special cases discussed in the source (Yu et al., 2020).

Under strict preferences, FTTC replaces discrete cycle execution with a balanced trade system. At step $i \in N$ 2, each remaining agent points to a unique top object $i \in N$ 3. A ratio matrix $i \in N$ 4 and quota matrix $i \in N$ 5 determine how object owners participate in trades. The core equations are

$i \in N$ 6

subject to

$i \in N$ 7

The coefficient matrix is column-stochastic, guaranteeing existence of the maximum solution $i \in N$ 8 (Yu et al., 2020). The resulting mechanism is individually rational and SD-efficient, and suitable choices of $i \in N$ 9 can ensure fairness properties such as ETE, EENE, and BE.

The paper also gives an equivalent cycle-based fractional view in house allocation. If a cycle $\succ_i$ 0 is found in the agent-object graph, one may allocate a common step size

$\succ_i$ 1

and update allocations and residual demands and capacities along the cycle (Yu et al., 2020). This view parallels deterministic TTC while embedding it into a fractional system.

5. FTTC on the full preference domain and relation to probabilistic serial

The full-domain extension addresses indifferences. The paper argues that naive tie-breaking can cause inefficiency and introduces a labeling stage that converts some previously consumed amounts into tradable endowments when an agent is indifferent between an exhausted object and an available object (Yu et al., 2020).

For each agent $\succ_i$ 2, the current top set is

$\succ_i$ 3

If $\succ_i$ 4, the agent may point to multiple objects. The extended FTTC proceeds in three stages at each step $\succ_i$ 5:

Labeling: exhausted but previously consumed objects tied with currently available objects are labeled as tradable; this may generate chains of labels.
Pointing: active agents point to all favorite objects among the remaining and labeled objects.
Trading: a balanced trade system is solved using a ratio matrix $\succ_i$ 6, a quota matrix $\succ_i$ 7, and a division matrix $\succ_i$ 8 that determines how agents split demand across pointed objects (Yu et al., 2020).

The full-domain balanced trade equations are

$\succ_i$ 9

subject to

$A$ 0

Existence of the maximum solution again follows from the stochastic structure (Yu et al., 2020).

The main theorem of the paper establishes that FTTC on the full preference domain is individually rational and SD-efficient. A shrinking-availability lemma states that once an object becomes unavailable for trading, it stays unavailable. Fairness is preserved by stepwise conditions on $A$ 1: stepwise ETE implies ETE; stepwise EEET implies equal-endowment no-envy and, in house allocation with equal endowments, ex-ante envy-freeness; bounded advantage implies bounded envy (Yu et al., 2020).

A corollary is an extension of the probabilistic serial mechanism to weak preferences. In the strict-preference PS mechanism, each agent eats her top available object at unit speed; the final allocation is the integral of eating rates over time. In the FTTC-based extension, agents eat from their top sets, division across ties is governed by $A$ 2, and labeling augments the instantaneous rate: $A$ 3 In house allocation, any FTTC satisfying stepwise EEET coincides with this PS extension and is SD-efficient and ex-ante envy-free (Yu et al., 2020). On the dichotomous domain, the same construction yields the egalitarian solution, matching the outcome of the Katta–Sethuraman extension, but via an eating-algorithm interpretation and elementary computation (Yu et al., 2020).

6. Canonical top cycle in the Voronoi complex

A third notion of top cycle appears in the geometry and cohomology of arithmetic groups. For $A$ 4, let $A$ 5 be the cone of positive definite quadratic forms and $A$ 6 the cone of nonnegative definite forms whose kernel is defined over $A$ 7. For a perfect form $A$ 8, its Voronoi domain is

$A$ 9

where $R$ 0 is the set of minimal vectors of $R$ 1 (Durán, 4 Apr 2026). The family of such cones gives Voronoi’s tessellation of $R$ 2, and $R$ 3 acts linearly on $R$ 4 by $R$ 5.

For a finite-index subgroup $R$ 6, the associated Voronoi complex $R$ 7 is formed from orbit representatives of faces of Voronoi domains. The top degree is

$R$ 8

Its cellular boundary uses incidence coefficients

$R$ 9

with orientation signs determined by induced bases of face spans (Durán, 4 Apr 2026).

The paper constructs the canonical top cycle

$x$ 0

where $x$ 1 is the stabilizer of the top cell $x$ 2 (Durán, 4 Apr 2026). The main theorem states that this chain is a nontrivial cycle generating the top homology: $x$ 3

The proof rests on a rigidity property of the Voronoi tessellation. Every codimension-one face is incident either to two neighboring top cells or to a self-intersection of one top cell under the group action. In the former case, the two top cells induce opposite orientations on the shared face, and the weighted contributions cancel because

$x$ 4

In the self-intersection case, the $x$ 5-orbit splits into two orbits with opposite signs and equal cardinalities, producing internal cancellation (Durán, 4 Apr 2026).

The geometric significance is amplified by Borel–Serre duality. Since

$x$ 6

the paper identifies

$x$ 7

so the canonical top cycle yields a canonical generator of cohomology in virtual cohomological degree (Durán, 4 Apr 2026). For $x$ 8, the paper notes that the relevant codimension-one representative set is empty, so the cycle property is immediate. For $x$ 9, cancellations across ordinary and self-intersecting facets become nontrivial and illustrate the full mechanism (Durán, 4 Apr 2026).

7. Comparison of the three notions

The three uses of “Top Cycle” differ in domain, mathematical object, and normative role.

Context	Mathematical object	Core characterization
Social choice	Set-valued SCC on alternatives	Smallest dominant set; maximal elements of $y$ 0 (Brandt et al., 2021)
Market design	Allocation mechanism and its fractional extension	Cycle-based or balanced-trade assignment rule with IR and SD-efficiency (Yu et al., 2020)
Arithmetic geometry	Homology class in Voronoi complex	Canonical weighted sum of top cells generating top homology (Durán, 4 Apr 2026)

In social choice, the Top Cycle is an outcome set extracted from pairwise majority structure. In allocation, top trading cycles are procedural devices for executing mutually beneficial exchanges, and FTTC generalizes them to fractional and weak-preference environments. In the Voronoi-complex setting, the top cycle is a chain-level object whose significance comes from orientation cancellation and cohomological duality.

A plausible implication is that the recurrence of the term reflects a deeper mathematical pattern: cyclic structure often identifies a maximal stable object when direct acyclic dominance is unavailable. In majority graphs, this is reachability under weak majority. In exchange economies, it is the execution of trades along cycles or fractional balanced-trade analogues. In Voronoi complexes, it is the cancellation of codimension-one boundaries across adjacent top-dimensional cells. The meanings are therefore non-interchangeable, but each version of top cycle formalizes a highest-order object determined by local cyclic relations.