---
title: Unit Interval Parking Functions Overview
url: https://www.emergentmind.com/topics/unit-interval-parking-functions
type: topic
---

# Unit Interval Parking Functions Overview

Unit interval parking functions are parking functions in which every car parks either in its preferred spot or in the immediately following spot. In current notation they are the \(1\)-interval parking functions \(\UPF_n=\IPF_n(1)\), and they occupy a distinctive position between classical parking functions and ordered set-partition models: they admit rigid local displacement constraints, a block decomposition that controls all valid rearrangements, bijections with Fubini rankings, ordered set partitions, Cayley permutations, certain path families, and faces of the permutohedron, and they are enumerated by the Fubini numbers rather than by \((n+1)^{n-1}\) [2507.07243] [2401.06937] [2305.15554].

## 1. Definition and basic characterizations

A parking function of length \(n\) is a word \(\alpha=(a_1,\dots,a_n)\in[n]^n\) for which the standard parking procedure parks all \(n\) cars. If car \(i\) parks in spot \(s_i\), its displacement is \(s_i-a_i\). A unit interval parking function is a parking function for which every displacement is at most \(1\); equivalently, each car parks in spot \(a_i\) or \(a_i+1\) [2507.07243]. In the notation of \(\ell\)-interval parking functions, this is exactly the case \(\ell=1\), so recent work treats “unit interval parking function” and “\(1\)-interval parking function” as synonymous [2507.07243].

Two equivalent formulations are especially useful. First, for a parking function \(p\) with outcome \(O(p)=(o_1,\dots,o_n)\), the condition \(p\in \UPF_n\) is equivalent to \(o_i\in\{p_i,p_i+1\}\) for all \(i\) [2403.17438]. Second, if \(\alpha^\uparrow=(a_1',\dots,a_n')\) is the weakly increasing rearrangement of \(\alpha\), then \(\alpha\in \UPF_n\) if and only if \(a_i'\in\{i-1,i\}\) for all \(i\); this is the block-word characterization used throughout the recent enumerative theory [2507.07243].

The nondecreasing case is particularly transparent. For a nondecreasing parking function \(p\in \PF_n^{\mathrm{inc}}\), the outcome is \(O(p)=(1,2,\dots,n)\), so \(d_i=i-p_i\). Hence a nondecreasing parking function is unit-interval if and only if
\[
p_1=1,\qquad p_i\in\{i-1,i\}\quad\text{for all }i\ge 2.
\]
This yields an immediate count \(2^{n-1}\) for the nondecreasing subclass \(\UPF_n^{\mathrm{inc}}\) [2403.17438].

A useful structural caveat is that the class is not closed under arbitrary rearrangement. For unit interval parking functions, only very particular permutations of the entries preserve unit-intervality, and even path-theoretic images that characterize the nondecreasing subclass do not characterize the full labeled class [2403.17438].

## 2. Block structure, prime decomposition, and internal organization

The central structural invariant is the block structure of the sorted word. If \(\alpha\in \UPF_n\) and \(\alpha^\uparrow=(a_1',\dots,a_n')\), one inserts a separator before each position \(i>1\) with \(a_i'=i\). This decomposes \(\alpha^\uparrow\) into blocks
\[
\alpha^\uparrow=\pi_1\mid \pi_2\mid \cdots\mid \pi_m.
\]
Each block has the form
\[
\pi_j=(r,\ r,\ r+1,\ r+2,\dots,r+\ell_j-1),
\]
where \(\ell_j=|\pi_j|\). Within each block, entries appear in increasing order in \(\alpha\); conversely, every shuffle of the blocks that preserves the internal order within each block is again a unit interval parking function [2507.07243].

This implies a complete rearrangement criterion. If \(\alpha^\uparrow=\pi_1\mid\cdots\mid\pi_m\), then the number of rearrangements of \(\alpha\) that remain in \(\UPF_n\) is exactly
\[
\binom{n}{|\pi_1|,\dots,|\pi_m|},
\]
and a rearrangement is valid if and only if it preserves the relative order of the entries within each block [2401.06937]. This result explains why the block structure is the correct analogue of a canonical factorization for unit interval parking functions.

A complementary description uses prime parking functions. A parking function is prime when its only breakpoint is \(n\). In the unit-interval setting, the prime object of length \(n\) is unique:
\[
(1,1,2,3,\dots,n-1).
\]
A parking-ordered unit interval parking function decomposes uniquely as a pipe of such prime unit-interval parking functions, shifted appropriately. This makes the class unusually rigid: the only freedom lies in how prime blocks of specified lengths are interleaved by labels [2305.15554].

The block count has a direct displacement meaning. If the block lengths are \(\ell_1,\dots,\ell_m\), then the total displacement is
\[
D(\alpha)=\sum_{j=1}^m(\ell_j-1)=n-m.
\]
Thus the number of blocks, the number of lucky cars, and the total displacement are equivalent statistics in this class [2607.01273].

## 3. Bijections with Fubini rankings, ordered set partitions, and Cayley permutations

A Fubini ranking of length \(n\) is a ranking with ties in which, if \(k>0\) entries are equal to \(x\), then the next largest value is \(x+k\). These objects are counted by the Fubini numbers. There is a bijection
\[
\phi: FR_n \longrightarrow UPF_n
\]
and an explicit inverse
\[
\psi: UPF_n \longrightarrow FR_n,
\]
giving \(|UPF_n|=|FR_n|=Fub_n\) [2401.06937].

The inverse map \(\psi\) is expressed directly in terms of the block structure. If \(\alpha\in \UPF_n\) has block structure \(\pi_1\mid\cdots\mid\pi_k\), then \(\psi(\alpha)_i\) is the minimum entry of the block containing \(a_i\). This collapses each unit-interval block to a single rank value, and the block minima satisfy exactly the Fubini ranking condition [2401.06937].

The same block structure yields a bijection with ordered set partitions. If \(\alpha\in \UPF_n\) has blocks \(\pi_1,\dots,\pi_m\), then one may record, for each position \(i\), the block index of \(a_i\). This gives a surjection \([n]\twoheadrightarrow[m]\), equivalently an ordered set partition of \([n]\) into \(m\) blocks. Summing over \(m\) recovers the Fubini numbers, so unit interval parking functions are canonically equivalent to ordered set partitions [2507.07243].

Another equivalent model is given by Cayley permutations, also called packed words, surjective words, or Fubini words. The bijection \(\psi:\UPF_n\to\C_n\) sends each entry to its block index; it preserves inversions:
\[
\Inv(\psi(\alpha))=\Inv(\alpha).
\]
This identification is the basis of the symmetric-function and \(q\)-enumerative theory of unit interval parking functions [2508.11587].

These correspondences are not merely equinumerative. They transfer statistics and restrictions faithfully. Block sizes become composition parts, ordered set-partition block sizes, and content of packed words; the number of blocks becomes the number of distinct ranks; and total displacement becomes \(n-\ell(c)\) for a content composition \(c\) [2508.11587].

## 4. Enumerative geometry and path models

The displacement-refined enumeration of unit interval parking functions is governed by Stirling numbers and the face structure of the permutohedron. For \(0\le k<n\), the unit interval parking functions of length \(n\) with total displacement \(k\) are in bijection with the \(k\)-dimensional faces of the permutohedron of order \(n\). Consequently,
\[
|UPF_{n,k}^{\mathrm{disp}}|=(n-k)!\,\left\{\begin{matrix} n \\ n-k \end{matrix}\right\},
\]
which is equivalent to the statement that the number with total displacement \(n-k\) is \(k!\,S(n,k)\) [2305.15554].

This geometric correspondence is compatible with the prime decomposition. If a unit interval parking function has prime block lengths \(n_1,\dots,n_t\), then the corresponding face has combinatorial type
\[
P(n_1)\times\cdots\times P(n_t),
\]
and the stabilizer size under the natural \(S_n\)-action is \(\prod_j n_j!\) [2305.15554].

For nondecreasing unit interval parking functions, several path models coexist. Under the bijection between nondecreasing parking functions and Łukasiewicz paths, the maximal displacement corresponds to path height and the total displacement to path area. Restricting to unit interval parking functions yields a bijection
\[
\UPF_n^{\mathrm{inc}} \longleftrightarrow \Motz_n^{\le 1},
\]
where \(\Motz_n^{\le1}\) denotes Motzkin paths of length \(n\) and height at most one, called \(1\)-Motzkin paths. This again gives
\[
|\UPF_n^{\mathrm{inc}}|=|\Motz_n^{\le1}|=2^{n-1}
\]
[2403.17438].

A separate path encoding identifies nondecreasing \(\ell\)-interval parking functions with Dyck paths of bounded height. Specializing to \(\ell=1\), nondecreasing unit interval parking functions correspond to Dyck paths of semilength \(n\) and height at most \(2\) [2311.14055].

| Family | Corresponding objects | Enumeration |
|---|---|---|
| \(\UPF_n\) | Fubini rankings / ordered set partitions | \(Fub_n\) |
| \(\UPF_n^{\mathrm{inc}}\) | \(1\)-Motzkin paths | \(2^{n-1}\) |
| displacement \(k\) in \(\UPF_n\) | \(k\)-faces of \(P(n)\) | \((n-k)!\,S(n,n-k)\) |
| \(\DUPF_n\) | deranged ordered set partitions | \(\tilde F_n\) |

The path correspondences also clarify an important limitation: the \(1\)-Motzkin characterization applies to the nondecreasing subclass, not to the full labeled class. The paper exhibiting the Łukasiewicz bijection gives an explicit counterexample showing that a \(1\)-Motzkin image need not come from a general unit interval parking function unless the nondecreasing restriction is imposed [2403.17438].

## 5. Statistics, symmetric functions, and algebraic refinements

The inversion enumerator for unit interval parking functions is especially tractable because of the bijection with Cayley permutations. Writing
\[
\UPF_n(q)=\sum_{\alpha\in \UPF_n} q^{\inv(\alpha)},
\]
one has the \(q\)-exponential generating function
\[
\sum_{n\ge0}\UPF_n(q)\,\frac{z^n}{[n]_q!}=\frac{1}{2-\exp_q(z)}.
\]
Equivalently,
\[
\UPF_n(q)=\sum_{c\in \Comp(n)} \binom{n}{c_1,\dots,c_{\ell(c)}}_q,
\]
with the sum over compositions of \(n\) [2508.11587].

The Frobenius characteristic is unusually simple:
\[
\UPF_n(\mathbf{x})=\sum_{c\in \Comp(n)} h_c,
\]
and the corresponding generating series is
\[
\sum_{n\ge0}\UPF_n(\mathbf{x})\,z^n=\frac{1}{2-H(z)}.
\]
The graded version with respect to the area statistic satisfies
\[
\UPF_n(\mathbf{x},t)=\sum_{c\in\Comp(n)} t^{n-\ell(c)}h_c,
\qquad
\sum_{n\ge0}\UPF_n(\mathbf{x},t)\,z^n=\frac{t}{[2]_t-H(tz)}.
\]
In this formulation, the content composition is exactly the block-size composition [2508.11587].

The displacement–inversion bivariate enumerator of unit interval parking functions has the form
\[
\sum_{\alpha\in\UPF_n} q^{\disp(\alpha)} t^{\inv(\alpha)}
=
\sum_{\sigma\in S_n}(1+q)^{\asc(\sigma)}t^{\inv(\sigma)}.
\]
Setting \(t=1\) yields
\[
\sum_{\alpha\in\UPF_n} q^{\disp(\alpha)}
=
\sum_{k=1}^n k!\,S(n,k)\,q^{\,n-k},
\]
so the number of unit interval parking functions with total displacement \(n-k\) is \(k!\,S(n,k)\), in agreement with the permutohedral model [2507.07243].

Several further statistics admit closed formulas. The total numbers of descents and inversions across all unit interval parking functions are
\[
\sum_{\alpha\in\UPF_n}\des(\alpha)=\frac{n-1}{2}\,(\Fub_n-\Fub_{n-1}),
\qquad
\sum_{\alpha\in\UPF_n}\inv(\alpha)=\frac{n(n-1)}{4}\,(\Fub_n-\Fub_{n-1}),
\]
and these formulas hold equally for Cayley permutations under the inversion-preserving bijection [2508.11587].

For the major index, Foata’s transform preserves \(\IPF_n(\ell)\) exactly when \(\ell\in\{0,1,2,n-2,n-1\}\). In particular it preserves \(\UPF_n\), so inversions and major index are equidistributed on unit interval parking functions [2507.07243]. The same paper proves a cyclic sieving phenomenon: for fixed total displacement \(k\), the triple \((\UPF_{n,k}^{\disp},\, f_{n,k}^{\disp}(t),\, C_n)\) exhibits the cyclic sieving phenomenon [2507.07243].

## 6. Generalizations, restricted families, and terminological boundaries

Several recent works study refined or extended versions of unit interval parking functions. One direction fixes additional constraints on the first cars. An \(r\)-Fubini ranking of length \(n+r\) is a Fubini ranking whose first \(r\) entries are distinct, and it is in bijection with unit interval parking functions of length \(n+r\) whose first \(r\) preferences are distinct. These objects are counted by the \(r\)-Fubini numbers
\[
Fb_n^{(r)}=\sum_{k=0}^{n}(k+r)!\left\{\begin{matrix} n+r \\ k+r \end{matrix}\right\}_r
\]
[2401.06937].

Another direction imposes restrictions on block data. “Restricted unit interval parking functions” have been developed in three forms: restrictions on the number of lucky cars, on allowed block sizes, and on block sizes position by position. For type 1 restrictions, if \(S\subseteq\mathbb{Z}^+\), then the exponential generating function is
\[
\sum_{n\ge0}|UPF_{n,S}|\,\frac{x^n}{n!}=\sum_{k\in S}(e^x-1)^k.
\]
These restrictions are transported through explicit bijections to restricted Fubini rankings and restricted ordered set partitions [2511.01997].

Deranged unit-interval parking functions form a further distinguished subclass. Under the ordered-set-partition bijection, a unit interval parking function is deranged if no block occupies the position indexed by its minimum element. The resulting set \(\DUPF_n\) satisfies
\[
|\DUPF_n|=\tilde F_n,
\qquad
\sum_{n\ge0}\tilde F_n\frac{x^n}{n!}=\frac{e^{1-e^x}}{2-e^x},
\]
and the refinement by number of blocks is
\[
|DUPF_{n,m}|=d_m\,S(n,m).
\]
Intrinsically, the deranged condition can be stated in terms of lucky cars, equivalently block leaders [2607.01273].

The \(\ell\)-interval theory extends beyond \(\ell=1\). For general \(\ell\), \(\IPF_n(\ell)\) denotes the class of parking functions with maximum displacement at most \(\ell\), and \(\IPF_n(1)=\UPF_n\). Rational analogues with \(m\ge n\) parking spots are also studied; in particular, the \(1\)-interval rational parking functions with \(n\) cars and \(m\) spots are in bijection with barred preferential arrangements of \([n]\) with \(m-n\) bars [2311.14055].

A terminological caution is necessary. There is also a distinct notion of interval parking functions in which each car may park only within a prescribed interval \([a(i),b(i)]\), encoded as a pair \((a,b)\). This “interval parking function” theory generalizes ordinary parking functions in a different direction and leads to the pseudoreachability order on permutations [2006.09321]. In current usage, however, “unit interval parking function” usually refers to the displacement-bounded class \(\IPF_n(1)\), not to the interval-constrained pair model.

Taken together, these results place unit interval parking functions at a well-defined intersection of Catalan combinatorics, ordered set partitions, permutohedral geometry, symmetric functions, and refined word statistics. Their rigidity is strong enough to permit explicit bijections and closed formulas, but weak enough to retain a large and varied combinatorial ecosystem.

Source: https://www.emergentmind.com/topics/unit-interval-parking-functions