Periodic Shuffling Operator (PS)

• Periodic Shuffling Operator is a deterministic permutation acting on sequences, defined via stack-and-sort and matrix transformations.
• It employs base‑k expansion and modular arithmetic to characterize cycle structures, invariant stacks, and period divisors.
• Its applications in algebraic combinatorics and symbolic dynamics highlight parallels to stochastic shuffling processes in discrete growth models.

The periodic shuffling operator (PS) is an explicit permutation acting on sequences, chiefly studied in the context of ordered card decks but with applications in symbolic dynamics and algebraic combinatorics. PS rearranges elements through a deterministic stack-and-sort or matrix transformation, resulting in rich periodic and invariant structures that can be precisely characterized when the deck size $N = k^t q$ with $\gcd(q, k)=1$. The cycle structure, existence of invariant and periodic stacks, and possible periods are governed by arithmetic derived from the base-$k$ expansion and the algebraic properties of the permutation induced by the operator (Butler et al., 2010).

1. Formal Definition and Mechanisms

Let $N = k^t q$ with integer $q \ge 1$ and $t \ge 1$ maximal (so $k^t \parallel N$), and fix a total order on the set of $N$ labels. The PS operator is defined via two equivalent processes:

• Stack-and-Sort Description:

The deck is split into $k$ stacks of equal size $n = N/k = k^{t-1} q$. At each shuffling step, the top card from each stack is removed, sorted in decreasing label order, and placed at the end of the new stack. The process repeats until all cards are processed.

• Matrix Description:

The deck, represented as $\gcd(q, k)=1$0, is arranged into a $\gcd(q, k)=1$1 array, filled row-wise. Within each column, entries are sorted by label, and the columns are read top to bottom, left to right, to generate the new sequence.

Both define a permutation $\gcd(q, k)=1$2 on $\gcd(q, k)=1$3.

2. Algebraic Structure and Cycle Decomposition

When $\gcd(q, k)=1$4, the algebraic form of PS is described using the base-$\gcd(q, k)=1$5 expansion $\gcd(q, k)=1$6: $\gcd(q, k)=1$7 The action reveals a highly structured permutation matrix $\gcd(q, k)=1$8 with $\gcd(q, k)=1$9. The cycle decomposition of $k$0 is best analyzed via the above map, offering insight into the lengths and structure of cycles.

After $k$1 shuffles, one has: $k$2 Iterating $k$3 such blocks yields: $k$4 The characterization of periods and cycles then reduces to properties of $k$5 modulo $k$6.

3. Invariants, Fixed and Periodic Stacks

A stack is fixed (period 1) if its labeling is invariant under PS. This imposes two necessary and sufficient conditions:

1. Cycle Constancy: All labels in a cycle of $k$7 must be equal.
2. Sorting Poset Consistency: If a label $k$8 would, via sorting, be repositioned below $k$9 in some column, invariance fails unless labels are weakly increasing along every poset comparability chain.

A stack has exact period $N = k^t q$0 if it returns to itself after $N = k^t q$1 shuffles but not before. Labels must be constant on the orbits of $N = k^t q$2, and for cycles of length $N = k^t q$3, the labeling must be $N = k^t q$4-periodic under rotation.

4. Main Theorem on Possible Periods

Let $N = k^t q$5 denote the minimal positive integer $N = k^t q$6 such that $N = k^t q$7. The key result is:

Given $N = k^t q$8 and $N = k^t q$9, every cycle in $q \ge 1$0 has length dividing $q \ge 1$1, and there exists a cycle of full length $q \ge 1$2. Thus, all possible periods of PS and of periodic stacks divide $q \ge 1$3 (Butler et al., 2010).

This is proved via the algebraic recurrence for the position $q \ge 1$4 after repeated application of PS and the structure of the permutation's action modulo $q \ge 1$5.

5. Worked Examples

$q \ge 1$6	$q \ge 1$7	$q \ge 1$8	$q \ge 1$9	$t \ge 1$0	$t \ge 1$1	Cycle Structure / Possible Periods
2	3	1	$t \ge 1$2	7	3	One 3-cycle, one 1-cycle; periods 1, 3
3	2	2	$t \ge 1$3	16	4	4-cycle; periods 1, 2, 4
3	1	4	12	8	2	All stacks are fixed or 2-periodic

For $t \ge 1$4, the permutation $t \ge 1$5 (with $t \ge 1$6 as the second digit in binary of $t \ge 1$7) has cycles of length dividing $t \ge 1$8, as predicted. In the ternary case ($t \ge 1$9), the maximum cycle length is $k^t \parallel N$0. For $k^t \parallel N$1, all cycles are either fixed or have period 2.

6. Connection to Related Shuffling and Growth Models

While PS is deterministic and concerned with ordered labels, it shares conceptual parallels with stochastic shuffling operations such as the domino-shuffling operator described in discrete growth and tiling models (Chhita et al., 2019). Both involve periodic actions and induced permutations, though the domino-shuffling process operates on matchings in $k^t \parallel N$2 and relates to interface growth in the anisotropic KPZ universality class. The deterministically periodic structure of PS finds analogs in the periodicity and limit-shape phenomena of domino tilings, with rigorous combinatorial underpinnings rooted in permutation cycles and group actions.

7. Summary and Implications

The periodic shuffling operator, as defined for $k^t \parallel N$3 elements and $k^t \parallel N$4, provides a framework for analyzing periodic rearrangements and invariant structures in ordered sequences. Its algebraic action via base-$k^t \parallel N$5 digit manipulations allows precise determination of cycle structure, period divisors, and characterization of invariant and periodic labelings. Applications of the operator extend to problems in algebraic combinatorics and symbolic dynamics, and its structural features inform the analysis of more general shuffling processes (Butler et al., 2010).