Oscillating Tableaux: Concepts & Connections

• Oscillating tableaux are combinatorial objects defined as sequences of partitions where each step adds or removes a cell or rim hook, providing a unified framework for various combinatorial models.
• They establish bijections with matchings, lattice paths, and growth diagrams, yielding elegant enumerative results such as connections to Catalan numbers and precise crossing/nesting statistics.
• They play a pivotal role in representation theory by modeling invariant tensor decompositions for classical groups and enabling dynamic operations like promotion and evacuation in crystal structures.

Oscillating tableaux are fundamental combinatorial objects that generalize standard Young tableaux by allowing the stepwise addition and removal of cells or more complex shapes such as rim hooks. They serve as a unifying framework connecting matchings, lattice path models, symmetric function theory, and the representation theory of classical groups, notably the symplectic group. Their paper underlies several deep bijective correspondences, elegant enumerative results, and generalizations to various tableau and crystal families.

1. Formal Definition and Variants

An oscillating tableau is a finite sequence of partitions (Young diagrams)

$$\emptyset = \lambda^{0}, \lambda^{1}, \dotsc, \lambda^{l}$$

where each partition $\lambda^{i}$ is obtained by adding or removing a specified combinatorial object from $\lambda^{i-1}$:

• In the classical setting, the object is a single cell (box). Each step adds or deletes a single box.
• In the $m$-rim hook generalization ("oscillating $m$-rim hook tableau"), each step adds or deletes a rim hook of length $m$, i.e., a connected border strip of $m$ contiguous cells tracing the rim of the Young diagram (possibly turning at corners) (Chen et al., 2011).

Additional restrictions can be imposed, such as requiring that all diagrams in the sequence have at most $k$ columns or end at a specified shape. The endpoint conditions (starting/ending with $\emptyset$, a one-column shape, or another fixed partition) are crucial for bijective correspondences.

Special cases include:

• Oscillating domino tableaux: $m=2$; rim hooks are dominoes (rectangles of 2 adjacent cells).
• Generalized oscillating tableaux: Steps may consist of adding/removing "horizontal (or vertical) strips"—collections of cells aligned in a row (or column) with at most one per column (or row).
• Semistandard oscillating tableaux (SSOT) and crystal-theoretic oscillating tableaux: These generalize the construction to track more subtle representation-theoretic data (see Section 6) (Lee, 2019, Kobayashi et al., 7 Jun 2025).

2. Bijections: Matchings, Lattice Paths, and Growth Diagrams

Oscillating tableaux admit combinatorial correspondences to several classical structures:

1. Perfect matchings and colored matchings: There is a bijection between oscillating $m$-rim hook tableaux of length $2n$ and $m$-colored matchings on $[2n]$, where each arc is assigned a color corresponding to the rim hook's type (Chen et al., 2011). For $m=2$ and additional restrictions (two columns), this yields noncrossing 2-colored matchings, counted by $C_n C_{n+1}$ (product of consecutive Catalan numbers).
2. Dyck path packings: Two-column oscillating domino tableaux of length $2n$ correspond bijectively to pairs $(D,E)$, where $D$ is a Dyck path of length $2n$ and $E$ is a dispersed Dyck path weakly covered by $D$ (Chen et al., 2011). This mapping is realized by encoding the conjugate of a Young diagram as $(u_i,v_i)$ and defining $a_i = (u_i+v_i)/2$, $b_i = (u_i-v_i)/2$, where the points $(i,a_i)$ and $(i,b_i)$ trace $D$ and $E$ respectively.
3. Growth diagram bijections: Using Fomin's growth diagrams and local rules, oscillating tableaux with at most $k$ columns and ending at a one-column shape correspond to standard Young tableaux of size $n$ with $m$ columns of odd length and all columns at most $2k$ (Krattenthaler, 2014). The approach generalizes to semistandard tableaux and their "Knuth-type" analogues, with generalizations to symmetric matrices and the combinatorics underlying the Robinson–Schensted–Knuth correspondence.
4. Chains in lattice and other tableaux families: Oscillating tableaux relate to families such as vacillating tableaux, stammering tableaux, and more, through bijective or limiting processes (Josuat-Vergès, 2016, Berikkyzy et al., 2022, Berikkyzy et al., 11 May 2024).

3. Enumerative and Structural Results

Oscillating tableaux lead to notable enumerative results:

• The number of length-$2n$, two-column oscillating domino tableaux (or equivalently, noncrossing 2-colored matchings on $[2n]$ or Dyck path packings) equals $C_n C_{n+1}$, where $C_n$ is the $n$-th Catalan number:

$C_n = \frac{1}{n+1}\binom{2n}{n}$

• The crossing number $cr(M)$ and nesting number $ne(M)$ of the associated $m$-colored matching $M$ are obtained from the maximum number of columns and rows in the tableaux, respectively:

$$ne(M) = \left\lfloor \frac{r(X)}{m} \right\rfloor, \qquad cr(M) = \left\lfloor \frac{c(X)}{m} \right\rfloor$$

for an oscillating $m$-rim hook tableau $X$ (Chen et al., 2011).

• Among oscillating tableaux of given endpoint and length, the average of additive statistics such as the total area (sum of sizes of all visited partitions) is polynomial in the shape size and excess length; for standard weight $wt(T) := \sum | \lambda^i |$, the average is

$$\frac{1}{| \mathcal{OT}(\lambda, k+2n) | } \sum_{T} wt(T) = \frac{1}{6}(4n^2 + 3k^2 + 8kn + 2n + 3k)$$

with $k = |\lambda|$ (Hopkins et al., 2014, Han et al., 2017).

4. Connections with Representation Theory and Symplectic/Orthogonal Groups

Oscillating tableaux model decomposition multiplicities in the tensor powers of irreducible representations for classical groups:

• Symplectic group $Sp(2m)$: Oscillating tableaux (or their slightly generalized forms) index a basis for invariant tensors under the symplectic group. Semistandard oscillating tableaux (SSOT) serve as the "Q-tableau" in the symplectic Robinson–Schensted–Knuth correspondence, recording the sequence of shape changes under Berele insertion. These models connect to King tableaux, crystal bases for $Sp(2m)$, and combinatorial rules for tensor product multiplicities (e.g., Littlewood–Richardson coefficients) (Lee, 2019, Kobayashi et al., 7 Jun 2025).
• Orthogonal group $SO(2k+1)$: The analogous role is played by vacillating tableaux, which, in the $k=1$ case (SO(3)), are essentially Riordan paths (lattice paths with up, down, and horizontal steps). These objects correspond bijectively to pairs consisting of standard Young tableaux and orthogonal Littlewood–Richardson tableaux, and preserve additional statistics such as descent sets that facilitate the quasi-symmetric expansion of Frobenius characters (Braunsteiner, 2018, Jagenteufel, 2019).
• Classical Pieri rules extension: The number of generalized oscillating tableaux of a fixed type equates to the multiplicity $K^{\lambda}_{\mu}$ of the irreducible representation indexed by $\lambda$ in the tensor product of GL, Sp, or O group representations, as described via horizontal/vertical strip additions, supporting the bridge between tableau combinatorics and tensor product decomposition (Okada, 2016).

5. Probabilistic and Random Walk Models

The evolution of the size (area) of the diagram along an oscillating tableau is described by a Markov process, with upward and downward transitions at each step governed by position-dependent probabilities. For large tableau length $N$, the (properly rescaled) area process converges to a Gaussian process with deterministic mean path $h(x) = x(1-x)$ and explicitly computable covariance kernel (Keating, 2020).

The evolution is characterized by:

• Upward step probability: $P[H(X+1) = Y+1 | H(X) = Y] = (N-X-Y)/(N-X)$
• Downward step probability: $P[H(X+1) = Y-1 | H(X) = Y] = Y/(N-X)$

Recursive formulas for the moments and joint moments of the area are derived, and the scaling limit (central limit theorem) quantifies fluctuations about the mean behavior.

The analysis enables efficient random sampling of oscillating tableaux (e.g., via "hook-walk" or related algorithms), with practical and theoretical impact on random partition models.

6. Generalizations and Crystal Structures

Oscillating tableaux reside within a hierarchy of tableau classes parametrized by how and what is added or removed at each step:

• Fluctuating tableaux (Editor's term): Sequences in which at each step a skew column of arbitrary height (possibly negative) is added or removed, encompassing oscillating, vacillating, and alternating tableaux, as well as semistandard and rational analogues (Gaetz et al., 2023).
• Semistandard oscillating tableaux (SSOT): Sequences where at each (sub)step, horizontal strips are added or deleted, generalizing the insertion moves to record the rich shape dynamics in Berele's symplectic correspondence. The symmetry of their generating function is shown via a novel Bender–Knuth involution constructed through the interplay of type A and type C RSK correspondences (Kobayashi et al., 7 Jun 2025).
• Crystal structures: On SSOT and related tableaux, Kashiwara crystal operators $e_i$, $f_i$ act by "moving" entries among components, satisfying the axioms for crystals of $sp_{2m}$. These local rules enable categorification of the tableaux class—each SSOT forms part of the crystal graph for a corresponding irreducible representation (Lee, 2019).

This organizational approach not only unifies many existing combinatorial models but also clarifies the algebraic and representation-theoretic principles governing their behavior.

7. Promotion, Evacuation, and Dynamical Symmetries

Promotion and evacuation—classically Schützenberger's operations on standard tableaux—extend to oscillating tableaux and their generalizations via cactus groups in the crystal context:

• Promotion: Defined via a sequence of piecewise operations (alternating reversals and involutions via cactus generators), it permutes the sequence of tableaux, often cycling combinatorial statistics such as descents, content, and matching structures (Pfannerer et al., 2018).
• Evacuation: Implemented as a global reversal with crystal involution, yields an operation dual to promotion, often corresponding to global symmetries such as chord reversal in matchings.
• The Sundaram bijection intertwines promotion on oscillating tableaux with rotation on perfect matchings (as embedded in a circular chord diagram), and evacuation with diagram reversal (Pfannerer et al., 2018).
• In the fluctuating tableau framework, promotion matrices and promotion permutations encode the evolution and symmetry properties under these operations, establishing relations such as $\mathrm{prom}_i(T) = \mathrm{prom}_{r-i}(T)^{-1}$ and conjugation behaviors with respect to standard cycles and involutions (Gaetz et al., 2023).

These dynamic symmetries manifest as resonance phenomena: for instance, under promotion, certain natural statistics exhibit periodic cycling with frequencies much smaller than the combinatorial orbit size (as in K-promotion on increasing tableaux and its resonance property) (Dilks et al., 2015).

---

Oscillating tableaux thus form a central structure in algebraic combinatorics and representation theory, integrating bijective combinatorics, enumerative identities, probabilistic models, crystal and symmetric function theory, and diagrammatic/algebraic dynamical systems. The continued generalization and categorification of oscillating tableau frameworks—particularly via fluctuating, semistandard, and crystal-structured forms—are active areas of research with connections spanning symmetric functions, invariant theory, and statistical mechanics.