arXiv subjects

Eva Rotenberg

Publications and source records attributed to Eva Rotenberg.

At least 37 records · Page 2Linked to original sources

arXiv2025

From Theory to Practice: Engineering Approximation Algorithms for Dynamic Orientation

Dynamic graph algorithms have seen significant theoretical advancements, but practical evaluations often lag behind. This work bridges the gap between theory and practice by engineering and empirically evaluating recently developed approximation algorithms for dynamically maintaining graph orientations. We comprehensively describe the underlying data structures, including efficient bucketing techniques and round-robin updates. Our implementation has a natural parameter $\lambda$, which allows for a trade-off between algorithmic efficiency and the quality of the solution. In the extensive experimental evaluation, we demonstrate that our implementation offers a considerable speedup. Using different quality metrics, we show that our implementations are very competitive and can outperform previous methods. Overall, our approach solves more instances than other methods while being up to 112 times faster on instances that are solvable by all methods compared.

cs.DS↗

arXiv2025

Instance-Optimal Imprecise Convex Hull

Imprecise measurements of a point set P = (p1, ..., pn) can be modelled by a family of regions F = (R1, ..., Rn), where each imprecise region Ri contains a unique point pi. A retrieval models an accurate measurement by replacing an imprecise region Ri with its corresponding point pi. We construct the convex hull of an imprecise point set in the plane, where regions in F may be retrieved at unit cost. The goal is to determine the cyclic ordering of the convex hull vertices of P as efficiently as possible. Here, efficiency is interpreted in two ways: (i) minimising the number of retrievals, and (ii) computing each retrieval location quickly. Prior works focused on only one of these two aspects: either minimising retrievals or optimising algorithmic runtime. Our contribution is the first to simultaneously achieve both. Let r(F, P) denote the minimal number of retrievals required by any algorithm to determine the convex hull of P for a given instance (F, P). For a family F of n constant-complexity polygons, our main result is a reconstruction algorithm that performs O(r(F, P)) retrievals in O(r(F, P) log^3 n) time. Compared to previous approaches that achieve optimal retrieval counts, we improve the runtime per retrieval by a exponential factor, from polynomial to polylogarithmic. Compared to near-linear time algorithms, we significantly reduce the number of retrievals used, and broaden the input families to include overlapping regions. We further extend our results to simple k-gons and to pairwise disjoint disks with radii in [1,k], where our runtime scales linearly with k.

cs.CG↗

arXiv2025

Fully dynamic biconnectivity in $\tilde{\mathcal{O}}(\log^2 n)$ time

We present a deterministic fully-dynamic data structure for maintaining information about the cut-vertices in a graph; i.e. the vertices whose removal would disconnect the graph. Our data structure supports insertion and deletion of edges, as well as queries to whether a pair of connected vertices are either biconnected, or can be separated by a cutvertex, and in the latter case we support access to separating cutvertices. All update operations are supported in amortized $O(\log^2 n \log^2 \log n)$ time, and queries take worst-case $O(\log n \log^2 \log n)$ time. Note that these time bounds match the current best for deterministic dynamic connectivity up to $\log \log n$ factors. We obtain our improved running time by a series of reductions from the original problem into well-defined data structure problems. While we do apply the well-known techniques for improving running time of two-edge connectivity [STOC'00, SODA'18], these techniques alone do not lead to an update time of $\tilde{O}(\log^3 n)$, let alone the $\tilde{O}(\log^2 n)$ we give as a final result. Our contributions include a formally defined transient expose operation, which can be thought of as a cheaper read-only expose operation on a top tree. For each vertex in the graph, we maintain a data structure over its neighbors, and in this data structure we apply biasing (twice) to save two $\tilde{O}(\log n)$ factors. One of these biasing techniques is a new biased disjoint sets data structure, which may be of independent interest. Moreover, in this neighborhood data structure, we facilitate that the vertex can select two VIP neighbors that get special treatment, corresponding to its potentially two neighbors on an exposed path, improving a $\log n$-time operation down to constant time. It is this combination of VIP neighbors with the transient expose that saves an $\tilde{O}(\log n)$-factor from another bottleneck.

cs.DS↗

arXiv2025

Efficient Greedy Discrete Subtrajectory Clustering

We cluster a set of trajectories T using subtrajectories of T. Clustering quality may be measured by the number of clusters, the number of vertices of T that are absent from the clustering, and by the Fr\'{e}chet distance between subtrajectories in a cluster. A $\Delta$-cluster of T is a cluster ${\mathcal{P}}$ of subtrajectories of T with a centre $P \in {\mathcal{P}}$ with complexity $\ell$, where all subtrajectories in ${\mathcal{P}}$ have Fr\'{e}chet distance at most $\Delta$ to $P$. Buchin, Buchin, Gudmundsson, L\"{o}ffler and Luo present two $O(n^2 + n m \ell)$-time algorithms: SC($\max$, $\ell$, $\Delta$, T) computes a single $\Delta$-cluster where $P$ has at least $\ell$ vertices and maximises the cardinality $m$ of ${\mathcal{P}}$. SC($m$, $\max$, $\Delta$, T) computes a single $\Delta$-cluster where ${\mathcal{P}}$ has cardinality $m$ and maximises the complexity $\ell$ of $P$. We use such maximum-cardinality clusters in a greedy clustering algorithm. We provide an efficient implementation of SC($\max$, $\ell$, $\Delta$, T) and SC($m$, $\max$, $\Delta$, T) that significantly outperforms previous implementations. We use these functions as a subroutine in a greedy clustering algorithm, which performs well when compared to existing subtrajectory clustering algorithms on real-world data. Finally, we observe that, for fixed $\Delta$ and T, these two functions always output a point on the Pareto front of some bivariate function $\theta(\ell, m)$. We design a new algorithm PSC($\Delta$, T) that in $O( n^2 \log^4 n)$ time computes a $2$-approximation of this Pareto front. This yields a broader set of candidate clusters, with comparable quality. We show that using PSC($\Delta$, T) as a subroutine improves the clustering quality and performance even further.

cs.CG↗

arXiv2025

Dynamic Indexing Through Learned Indices with Worst-case Guarantees

Indexing data is a fundamental problem in computer science. Recently, various papers apply machine learning to this problem. For a fixed integer $\varepsilon$, a \emph{learned index} is a function $h : \mathcal{U} \rightarrow [0, n]$ where $\forall q \in \mathcal{U}$, $h(q) \in [\text{rank}(q) - \varepsilon, \text{rank}(q) + \varepsilon]$. These works use machine learning to compute $h$. Then, they store $S$ in a sorted array $A$ and access $A[\lfloor h(q) \rfloor]$ to answer queries in $O(k + \varepsilon + \log |h|)$ time. Here, $k$ denotes the output size and $|h|$ the complexity of $h$. Ferragina and Vinciguerra (VLDB 2020) observe that creating a learned index is a geometric problem. They define the PGM index by restricting $h$ to a piecewise linear function and show a linear-time algorithm to compute a PGM index of approximate minimum complexity. Since indexing queries are decomposable, the PGM index may be made dynamic through the logarithmic method. When allowing deletions, range query times deteriorate to worst-case $O(N + \sum\limits_i^{\lceil \log n \rceil } (\varepsilon + \log |h_i|))$ time (where $N$ is the largest size of $S$ seen so far). This paper offers a combination of theoretical insights and experiments as we apply techniques from computational geometry to dynamically maintain an approximately minimum-complexity learned index $h : \mathcal{U} \rightarrow [0, n]$ with $O(\log^2 n)$ update time. We also prove that if we restrict $h$ to lie in a specific subclass of piecewise-linear functions, then we can combine $h$ and hash maps to support queries in $O(k + \varepsilon + \log |h|)$ time (at the cost of increasing $|h|$). We implement our algorithm and compare it to the existing implementation. Our empirical analysis shows that our solution supports more efficient range queries in the special case where the update sequence contains many deletions.

cs.CG↗

arXiv2025

Augmenting Plane Straight-Line Graphs to Meet Parity Constraints

Given a plane geometric graph $G$ on $n$ vertices, we want to augment it so that given parity constraints of the vertex degrees are met. In other words, given a subset $R$ of the vertices, we are interested in a plane geometric supergraph $G'$ such that exactly the vertices of $R$ have odd degree in $G'\setminus G$. We show that the question whether such a supergraph exists can be decided in polynomial time for two interesting cases. First, when the vertices are in convex position, we present a linear-time algorithm. Building on this insight, we solve the case when $G$ is a plane geometric path in $O(n \log n)$ time. This solves an open problem posed by Catana, Olaverri, Tejel, and Urrutia (Appl. Math. Comput. 2020).

cs.CG↗

arXiv2025

Sparsity-Parameterised Dynamic Edge Colouring

We study the edge-colouring problem, and give efficient algorithms where the number of colours is parameterised by the graph's arboricity, $α$. In a dynamic graph, subject to insertions and deletions, we give a deterministic algorithm that updates a proper $Δ+ O(α)$ edge~colouring in $\operatorname{poly}(\log n)$ amortized time. Our algorithm is fully adaptive to the current value of the maximum degree and arboricity. In this fully-dynamic setting, the state-of-the-art edge-colouring algorithms are either a randomised algorithm using $(1 + \varepsilon)Δ$ colours in $\operatorname{poly}(\log n, ε^{-1})$ time per update, or the naive greedy algorithm which is a deterministic $2Δ-1$ edge colouring with $\log(Δ)$ update time. Compared to the $(1+\varepsilon)Δ$ algorithm, our algorithm is deterministic and asymptotically faster, and when $α$ is sufficiently small compared to $Δ$, it even uses fewer colours. In particular, ours is the first $Δ+O(1)$ edge-colouring algorithm for dynamic forests, and dynamic planar graphs, with polylogarithmic update time. Additionally, in the static setting, we show that we can find a proper edge colouring with $Δ+ 2α$ colours in $O(m\log n)$ time. Moreover, the colouring returned by our algorithm has the following local property: every edge $uv$ is coloured with a colour in $\{1, \max\{deg(u), deg(v)\} + 2α\}$. The time bound matches that of the greedy algorithm that computes a $2Δ-1$ colouring of the graph's edges, and improves the number of colours when $α$ is sufficiently small compared to $Δ$.

cs.DS↗

arXiv2024

Local Density and its Distributed Approximation

The densest subgraph problem is a classic problem in combinatorial optimisation. Danisch, Chan, and Sozio propose a definition for \emph{local density} that assigns to each vertex $v$ a value $ρ^*(v)$. This local density is a generalisation of the maximum subgraph density of a graph. I.e., if $ρ(G)$ is the subgraph density of a finite graph $G$, then $ρ(G)$ equals the maximum local density $ρ^*(v)$ over vertices $v$ in $G$. They approximate the local density of each vertex with no theoretical (asymptotic) guarantees. We provide an extensive study of this local density measure. Just as with (global) maximum subgraph density, we show that there is a dual relation between the local out-degrees and the minimum out-degree orientations of the graph. We introduce the definition of the local out-degree $g^*(v)$ of a vertex $v$, and show it to be equal to the local density $ρ^*(v)$. We consider the local out-degree to be conceptually simpler, shorter to define, and easier to compute. Using the local out-degree we show a previously unknown fact: that existing algorithms already dynamically approximate the local density. Next, we provide the first distributed algorithms that compute the local density with provable guarantees: given any $\varepsilon$ such that $\varepsilon^{-1} \in O(poly \, n)$, we show a deterministic distributed algorithm in the LOCAL model where, after $O(\varepsilon^{-2} \log^2 n)$ rounds, every vertex $v$ outputs a $(1 + \varepsilon)$-approximation of their local density $ρ^*(v)$. In CONGEST, we show a deterministic distributed algorithm that requires $\text{poly}(\log n,\varepsilon^{-1}) \cdot 2^{O(\sqrt{\log n})}$ rounds, which is sublinear in $n$. As a corollary, we obtain the first deterministic algorithm running in a sublinear number of rounds for $(1+\varepsilon)$-approximate densest subgraph detection in the CONGEST model.

cs.DS↗

arXiv2024

Approximating Klee's Measure Problem and a Lower Bound for Union Volume Estimation

Union volume estimation is a classical algorithmic problem. Given a family of objects $O_1,\ldots,O_n \subseteq \mathbb{R}^d$, we want to approximate the volume of their union. In the special case where all objects are boxes (also known as hyperrectangles) this is known as Klee's measure problem. The state-of-the-art algorithm [Karp, Luby, Madras '89] for union volume estimation and Klee's measure problem in constant dimension $d$ computes a $(1+\varepsilon)$-approximation with constant success probability by using a total of $O(n/\varepsilon^2)$ queries of the form (i) ask for the volume of $O_i$, (ii) sample a point uniformly at random from $O_i$, and (iii) query whether a given point is contained in $O_i$. We show that if one can only interact with the objects via the aforementioned three queries, the query complexity of [Karp, Luby, Madras '89] is indeed optimal, i.e., $\Omega(n/\varepsilon^2)$ queries are necessary. Our lower bound already holds for estimating the union of equiponderous axis-aligned polygons in $\mathbb{R}^2$, and even if the algorithm is allowed to inspect the coordinates of the points sampled from the polygons, and still holds when a containment query can ask containment of an arbitrary (not sampled) point. Guided by the insights of the lower bound, we provide a more efficient approximation algorithm for Klee's measure problem improving the $O(n/\varepsilon^2)$ time to $O((n+\frac{1}{\varepsilon^2}) \cdot \log^{O(d)}n)$. We achieve this improvement by exploiting the geometry of Klee's measure problem in various ways: (1) Since we have access to the boxes' coordinates, we can split the boxes into classes of boxes of similar shape. (2) Within each class, we show how to sample from the union of all boxes, by using orthogonal range searching. And (3) we exploit that boxes of different classes have small intersection, for most pairs of classes.

cs.CG↗

arXiv2024

Data Structures for Approximate Discrete Fréchet Distance

The Fréchet distance is a popular distance measure between curves $P$ and $Q$. Conditional lower bounds prohibit $(1 + \varepsilon)$-approximate Fréchet distance computations in strongly subquadratic time, even when preprocessing $P$ using any polynomial amount of time and space. As a consequence, the Fréchet distance has been studied under realistic input assumptions, for example, assuming both curves are $c$-packed. In this paper, we study $c$-packed curves in Euclidean space $\mathbb R^d$ and in general geodesic metrics $\mathcal X$. In $\mathbb R^d$, we provide a nearly-linear time static algorithm for computing the $(1+\varepsilon)$-approximate continuous Fréchet distance between $c$-packed curves. Our algorithm has a linear dependence on the dimension $d$, as opposed to previous algorithms which have an exponential dependence on $d$. In general geodesic metric spaces $\mathcal X$, little was previously known. We provide the first data structure, and thereby the first algorithm, under this model. Given a $c$-packed input curve $P$ with $n$ vertices, we preprocess it in $O(n \log n)$ time, so that given a query containing a constant $\varepsilon$ and a curve $Q$ with $m$ vertices, we can return a $(1+\varepsilon)$-approximation of the discrete Fréchet distance between $P$ and $Q$ in time polylogarithmic in $n$ and linear in $m$, $1/\varepsilon$, and the realism parameter $c$. Finally, we show several extensions to our data structure; to support dynamic extend/truncate updates on $P$, to answer map matching queries, and to answer Hausdorff distance queries.

cs.CG↗

arXiv2024

Simpler Optimal Sorting from a Directed Acyclic Graph

Fredman proposed in 1976 the following algorithmic problem: Given are a ground set $X$, some partial order $P$ over $X$, and some comparison oracle $O_L$ that specifies a linear order $L$ over $X$ that extends $P$. A query to $O_L$ has as input distinct $x, x' \in X$ and outputs whether $x <_L x'$ or vice versa. If we denote by $e(P)$ the number of linear extensions of $P$, then $\log e(P)$ is a worst-case lower bound on the number of queries needed to output the sorted order of $X$. Fredman did not specify in what form the partial order is given. Haeupler, Hlad\'ik, Iacono, Rozhon, Tarjan, and T\v{e}tek ('24) propose to assume as input a directed acyclic graph, $G$, with $m$ edges and $n=|X|$ vertices. Denote by $P_G$ the partial order induced by $G$. Algorithmic performance is measured in running time and the number of queries used, where they use $\Theta(m + n + \log e(P_G))$ time and $\Theta(\log e(P_G))$ queries to output $X$ in its sorted order. Their algorithm is worst-case optimal in terms of running time and queries, both. Their algorithm combines topological sorting with heapsort. Their analysis relies upon sophisticated counting arguments using entropy, recursively defined sets defined over the run of their algorithm, and vertices in the graph that they identify as bottlenecks for sorting. In this paper, we do away with sophistication. We show that when the input is a directed acyclic graph then the problem admits a simple solution using $\Theta(m + n + \log e(P_G))$ time and $\Theta(\log e(P_G))$ queries. Especially our proofs are much simpler as we avoid the usage of advanced charging arguments and data structures, and instead rely upon two brief observations.

cs.DS↗

arXiv2024

Fully-Adaptive Dynamic Connectivity of Square Intersection Graphs

A classical problem in computational geometry and graph algorithms is: given a dynamic set S of geometric shapes in the plane, efficiently maintain the connectivity of the intersection graph of S. Previous papers studied the setting where, before the updates, the data structure receives some parameter P. Then, updates could insert and delete disks as long as at all times the disks have a diameter that lies in a fixed range [1/P, 1]. The state-of-the-art for storing disks in a dynamic connectivity data structure is a data structure that uses O(Pn) space and that has amortized O(P log^4 n) expected amortized update time. Connectivity queries between disks are supported in O( log n / loglog n) time. The state-of-the-art for Euclidean disks immediately implies a data structure for connectivity between axis-aligned squares that have their diameter in the fixed range [1/P, 1], with an improved update time of O(P log^4 n) amortized time. We restrict our attention to axis-aligned squares, and study fully-dynamic square intersection graph connectivity. Our result is fully-adaptive to the aspect ratio, spending time proportional to the current aspect ratio ψ, as opposed to some previously given maximum P. Our focus on squares allows us to simplify and streamline the connectivity pipeline from previous work. When $n$ is the number of squares and ψ is the aspect ratio after insertion (or before deletion), our data structure answers connectivity queries in O(log n / loglog n) time. We can update connectivity information in O(ψ log^4 n + log^6 n) amortized time. We also improve space usage from O(P n log n) to O(n log^3 n log ψ) -- while generalizing to a fully-adaptive aspect ratio -- which yields a space usage that is near-linear in n for any polynomially bounded ψ.

cs.CG↗

arXiv2024

Distributed Computation with Local Advice

In this work we study local computation with advice: the goal is to solve a graph problem $\Pi$ with a distributed algorithm in $T(\Delta)$ communication rounds, for some function $T$ that only depends on the maximum degree $\Delta$ of the graph, and the key question is how many bits of advice per node are needed. Some of our results regard Locally Checkable Labeling problems (LCLs), which are constraint-satisfaction graph problems that can be defined with a finite set of valid input/output-labeled neighborhoods. Our main results are: - Any LCL can be solved with only $1$ bit of advice per node in graphs with sub-exponential growth. Moreover, we can make the set of nodes that carry advice bits arbitrarily sparse. As a corollary, any LCL admits a locally checkable proof with $1$ bit per node in graphs with sub-exponential growth. - The assumption of sub-exponential growth is complemented by a conditional lower bound: assuming the Exponential-Time Hypothesis, there are locally checkable labeling problems that cannot be solved in general with any constant number of bits per node. - In any graph we can find an almost-balanced orientation with $1$ bit of advice per node, and again we can make the advice arbitrarily sparse. As a corollary, we can also compress an arbitrary subset of edges so that a node of degree $d$ stores only $d/2 + 2$ bits, and we can decompress it locally, in $T(\Delta)$ rounds. - In any graph of maximum degree $\Delta$, we can find a $\Delta$-coloring (if it exists) with $1$ bit of advice per node, and again, we can make the advice arbitrarily sparse. - In any $3$-colorable graph, we can find a $3$-coloring with $1$ bit of advice per node. As a corollary, in bounded-degree graphs there is a locally checkable proof that certifies $3$-colorability with $1$ bit of advice per node.

cs.DC↗

arXiv2024

Private graph colouring with limited defectiveness

Differential privacy is the gold standard in the problem of privacy preserving data analysis, which is crucial in a wide range of disciplines. Vertex colouring is one of the most fundamental questions about a graph. In this paper, we study the vertex colouring problem in the differentially private setting. To be edge-differentially private, a colouring algorithm needs to be defective: a colouring is d-defective if a vertex can share a colour with at most d of its neighbours. Without defectiveness, the only differentially private colouring algorithm needs to assign n different colours to the n different vertices. We show the following lower bound for the defectiveness: a differentially private c-edge colouring algorithm of a graph of maximum degree {\Delta} > 0 has defectiveness at least d = {\Omega} (log n / (log c+log {\Delta})). We also present an {\epsilon}-differentially private algorithm to {\Theta} ( {\Delta} / log n + 1 / {\epsilon})-colour a graph with defectiveness at most {\Theta}(log n).

cs.DS↗

arXiv2024

Gapped String Indexing in Subquadratic Space and Sublinear Query Time

In Gapped String Indexing, the goal is to compactly represent a string $S$ of length $n$ such that for any query consisting of two strings $P_1$ and $P_2$, called patterns, and an integer interval $[α, β]$, called gap range, we can quickly find occurrences of $P_1$ and $P_2$ in $S$ with distance in $[α, β]$. Gapped String Indexing is a central problem in computational biology and text mining and has thus received significant research interest, including parameterized and heuristic approaches. Despite this interest, the best-known time-space trade-offs for Gapped String Indexing are the straightforward $O(n)$ space and $O(n+occ)$ query time or $Ω(n^2)$ space and $\tilde{O}(|P_1| + |P_2| + occ)$ query time. We break through this barrier obtaining the first interesting trade-offs with polynomially subquadratic space and polynomially sublinear query time. In particular, we show that, for every $0\leq δ\leq 1$, there is a data structure for Gapped String Indexing with either $\tilde{O}(n^{2-δ/3})$ or $\tilde{O}(n^{3-2δ})$ space and $\tilde{O}(|P_1| + |P_2| + n^δ\cdot (occ+1))$ query time, where $occ$ is the number of reported occurrences. As a new tool towards obtaining our main result, we introduce the Shifted Set Intersection problem. We show that this problem is equivalent to the indexing variant of 3SUM (3SUM Indexing). Via a series of reductions, we obtain a solution to the Gapped String Indexing problem. Furthermore, we enhance our data structure for deciding Shifted Set Intersection, so that we can support the reporting variant of the problem. Via the obtained equivalence to 3SUM Indexing, we thus give new improved data structures for the reporting variant of 3SUM Indexing, and we show how this improves upon the state-of-the-art solution for Jumbled Indexing for any alphabet of constant size $σ>5$.

cs.DS↗

arXiv2024

String Indexing for Top-$k$ Close Consecutive Occurrences

The classic string indexing problem is to preprocess a string $S$ into a compact data structure that supports efficient subsequent pattern matching queries, that is, given a pattern string $P$, report all occurrences of $P$ within $S$. In this paper, we study a basic and natural extension of string indexing called the string indexing for top-$k$ close consecutive occurrences problem (SITCCO). Here, a consecutive occurrence is a pair $(i,j)$, $i < j$, such that $P$ occurs at positions $i$ and $j$ in $S$ and there is no occurrence of $P$ between $i$ and $j$, and their distance is defined as $j-i$. Given a pattern $P$ and a parameter $k$, the goal is to report the top-$k$ consecutive occurrences of $P$ in $S$ of minimal distance. The challenge is to compactly represent $S$ while supporting queries in time close to the length of $P$ and $k$. We give three time-space trade-offs for the problem. Let $n$ be the length of $S$, $m$ the length of $P$, and $ε\in(0,1]$. Our first result achieves $O(n\log n)$ space and optimal query time of $O(m+k)$. Our second and third results achieve linear space and query times either $O(m+k^{1+ε})$ or $O(m + k \log^{1+ε} n)$. Along the way, we develop several techniques of independent interest, including a new translation of the problem into a line segment intersection problem and a new recursive clustering technique for trees.

cs.DS↗

arXiv2024

Surface Reconstruction Using Rotation Systems

Inspired by the seminal result that a graph and an associated rotation system uniquely determine the topology of a closed manifold, we propose a combinatorial method for reconstruction of surfaces from points. Our method constructs a spanning tree and a rotation system. Since the tree is trivially a planar graph, its rotation system determines a genus zero surface with a single face which we proceed to incrementally refine by inserting edges to split faces and thus merging them. In order to raise the genus, special handles are added by inserting edges between different faces and thus merging them. We apply our method to a wide range of input point clouds in order to investigate its effectiveness, and we compare our method to several other surface reconstruction methods. We find that our method offers better control over outlier classification, i.e. which points to include in the reconstructed surface, and also more control over the topology of the reconstructed surface.

cs.CG↗

arXiv2023

Dynamic Dynamic Time Warping

The Dynamic Time Warping (DTW) distance is a popular similarity measure for polygonal curves (i.e., sequences of points). It finds many theoretical and practical applications, especially for temporal data, and is known to be a robust, outlier-insensitive alternative to the \frechet distance. For static curves of at most $n$ points, the DTW distance can be computed in $O(n^2)$ time in constant dimension. This tightly matches a SETH-based lower bound, even for curves in $\mathbb{R}^1$. In this work, we study \emph{dynamic} algorithms for the DTW distance. Here, the goal is to design a data structure that can be efficiently updated to accommodate local changes to one or both curves, such as inserting or deleting vertices and, after each operation, reports the updated DTW distance. We give such a data structure with update and query time $O(n^{1.5} \log n)$, where $n$ is the maximum length of the curves. As our main result, we prove that our data structure is conditionally \emph{optimal}, up to subpolynomial factors. More precisely, we prove that, already for curves in $\mathbb{R}^1$, there is no dynamic algorithm to maintain the DTW distance with update and query time~\makebox{$O(n^{1.5 - δ})$} for any constant $δ> 0$, unless the Negative-$k$-Clique Hypothesis fails. In fact, we give matching upper and lower bounds for various trade-offs between update and query time, even in cases where the lengths of the curves differ.

cs.CG↗