arXiv subjects

Philip Bille

Publications and source records attributed to Philip Bille.

At least 37 records · Page 2Linked to original sources

arXiv2023

Sliding Window String Indexing in Streams

Given a string $S$ over an alphabet $Σ$, the 'string indexing problem' is to preprocess $S$ to subsequently support efficient pattern matching queries, i.e., given a pattern string $P$ report all the occurrences of $P$ in $S$. In this paper we study the 'streaming sliding window string indexing problem'. Here the string $S$ arrives as a stream, one character at a time, and the goal is to maintain an index of the last $w$ characters, called the 'window', for a specified parameter $w$. At any point in time a pattern matching query for a pattern $P$ may arrive, also streamed one character at a time, and all occurrences of $P$ within the current window must be returned. The streaming sliding window string indexing problem naturally captures scenarios where we want to index the most recent data (i.e. the window) of a stream while supporting efficient pattern matching. Our main result is a simple $O(w)$ space data structure that uses $O(\log w)$ time with high probability to process each character from both the input string $S$ and the pattern string $P$. Reporting each occurrence from $P$ uses additional constant time per reported occurrence. Compared to previous work in similar scenarios this result is the first to achieve an efficient worst-case time per character from the input stream. We also consider a delayed variant of the problem, where a query may be answered at any point within the next $δ$ characters that arrive from either stream. We present an $O(w + δ)$ space data structure for this problem that improves the above time bounds to $O(\log(w/δ))$. In particular, for a delay of $δ= εw$ we obtain an $O(w)$ space data structure with constant time processing per character. The key idea to achieve our result is a novel and simple hierarchical structure of suffix trees of independent interest, inspired by the classic log-structured merge trees.

cs.DS↗

arXiv2022

The Complexity of the Co-Occurrence Problem

Let $S$ be a string of length $n$ over an alphabet $Σ$ and let $Q$ be a subset of $Σ$ of size $q \geq 2$. The 'co-occurrence problem' is to construct a compact data structure that supports the following query: given an integer $w$ return the number of length-$w$ substrings of $S$ that contain each character of $Q$ at least once. This is a natural string problem with applications to, e.g., data mining, natural language processing, and DNA analysis. The state of the art is an $O(\sqrt{nq})$ space data structure that -- with some minor additions -- supports queries in $O(\log\log n)$ time [CPM 2021]. Our contributions are as follows. Firstly, we analyze the problem in terms of a new, natural parameter $d$, giving a simple data structure that uses $O(d)$ space and supports queries in $O(\log\log n)$ time. The preprocessing algorithm does a single pass over $S$, runs in expected $O(n)$ time, and uses $O(d)$ space in addition to the input. Furthermore, we show that $O(d)$ space is optimal and that $O(\log\log n)$-time queries are optimal given optimal space. Secondly, we bound $d = O(\sqrt{nq})$, giving clean bounds in terms of $n$ and $q$ that match the state of the art. Furthermore, we prove that $Ω(\sqrt{nq})$ bits of space is necessary in the worst case, meaning that the $O(\sqrt{nq})$ upper bound is tight to within polylogarithmic factors. All of our results are based on simple and intuitive combinatorial ideas that simplify the state of the art.

cs.DS↗

arXiv2022

Hierarchical Relative Lempel-Ziv Compression

Relative Lempel-Ziv (RLZ) parsing is a dictionary compression method in which a string $S$ is compressed relative to a second string $R$ (called the reference) by parsing $S$ into a sequence of substrings that occur in $R$. RLZ is particularly effective at compressing sets of strings that have a high degree of similarity to the reference string, such as a set of genomes of individuals from the same species. With the now cheap cost of DNA sequencing, such data sets have become extremely abundant and are rapidly growing. In this paper, instead of using a single reference string for the entire collection, we investigate the use of different reference strings for subsets of the collection, with the aim of improving compression. In particular, we form a rooted tree (or hierarchy) on the strings and then compressed each string using RLZ with parent as reference, storing only the root of the tree in plain text. To decompress, we traverse the tree in BFS order starting at the root, decompressing children with respect to their parent. We show that this approach leads to a twofold improvement in compression on bacterial genome data sets, with negligible effect on decompression time compared to the standard single reference approach. We show that an effective hierarchy for a given set of strings can be constructed by computing the optimal arborescence of a completed weighted digraph of the strings, with weights as the number of phrases in the RLZ parsing of the source and destination vertices. We further show that instead of computing the complete graph, a sparse graph derived using locality sensitive hashing can significantly reduce the cost of computing a good hierarchy, without adversely effecting compression performance.

cs.DS↗

arXiv2022

Predecessor on the Ultra-Wide Word RAM

We consider the predecessor problem on the ultra-wide word RAM model of computation, which extends the word RAM model with 'ultrawords' consisting of $w^2$ bits [TAMC, 2015]. The model supports arithmetic and boolean operations on ultrawords, in addition to 'scattered' memory operations that access or modify $w$ (potentially non-contiguous) memory addresses simultaneously. The ultra-wide word RAM model captures (and idealizes) modern vector processor architectures. Our main result is a simple, linear space data structure that supports predecessor in constant time and updates in amortized, expected constant time. This improves the space of the previous constant time solution that uses space in the order of the size of the universe. Our result holds even in a weaker model where ultrawords consist of $w^{1+ε}$ bits for any $ε> 0 $. It is based on a new implementation of the classic $x$-fast trie data structure of Willard [Inform. Process. Lett. 17(2), 1983] combined with a new dictionary data structure that supports fast parallel lookups.

cs.DS↗

arXiv2021

Random Access in Persistent Strings and Segment Selection

We consider compact representations of collections of similar strings that support random access queries. The collection of strings is given by a rooted tree where edges are labeled by an edit operation (inserting, deleting, or replacing a character) and a node represents the string obtained by applying the sequence of edit operations on the path from the root to the node. The goal is to compactly represent the entire collection while supporting fast random access to any part of a string in the collection. This problem captures natural scenarios such as representing the past history of an edited document or representing highly-repetitive collections. Given a tree with $n$ nodes, we show how to represent the corresponding collection in $O(n)$ space and $O(\log n/ \log \log n)$ query time. This improves the previous time-space trade-offs for the problem. Additionally, we show a lower bound proving that the query time is optimal for any solution using near-linear space. To achieve our bounds for random access in persistent strings we show how to reduce the problem to the following natural geometric selection problem on line segments. Consider a set of horizontal line segments in the plane. Given parameters $i$ and $j$, a segment selection query returns the $j$th smallest segment (the segment with the $j$th smallest $y$-coordinate) among the segments crossing the vertical line through $x$-coordinate $i$. The segment selection problem is to preprocess a set of horizontal line segments into a compact data structure that supports fast segment selection queries. We present a solution that uses $O(n)$ space and support segment selection queries in $O(\log n/ \log \log n)$ time, where $n$ is the number of segments. Furthermore, we prove that that this query time is also optimal for any solution using near-linear space.

cs.DS↗

arXiv2021

Gapped Indexing for Consecutive Occurrences

The classic string indexing problem is to preprocess a string S into a compact data structure that supports efficient pattern matching queries. Typical queries include existential queries (decide if the pattern occurs in S), reporting queries (return all positions where the pattern occurs), and counting queries (return the number of occurrences of the pattern). In this paper we consider a variant of string indexing, where the goal is to compactly represent the string such that given two patterns P1 and P2 and a gap range [α,β] we can quickly find the consecutive occurrences of P1 and P2 with distance in [α,β], i.e., pairs of occurrences immediately following each other and with distance within the range. We present data structures that use Õ(n) space and query time Õ(|P1|+|P2|+n^(2/3)) for existence and counting and Õ(|P1|+|P2|+n^(2/3)*occ^(1/3)) for reporting. We complement this with a conditional lower bound based on the set intersection problem showing that any solution using Õ(n) space must use \tildeΩ}(|P1|+|P2|+\sqrt{n}) query time. To obtain our results we develop new techniques and ideas of independent interest including a new suffix tree decomposition and hardness of a variant of the set intersection problem.

cs.DS↗

arXiv2020

Partial Sums on the Ultra-Wide Word RAM

We consider the classic partial sums problem on the ultra-wide word RAM model of computation. This model extends the classic $w$-bit word RAM model with special ultrawords of length $w^2$ bits that support standard arithmetic and boolean operation and scattered memory access operations that can access $w$ (non-contiguous) locations in memory. The ultra-wide word RAM model captures (and idealizes) modern vector processor architectures. Our main result is a new in-place data structure for the partial sum problem that only stores a constant number of ultraword in addition to the input and supports operations in doubly logarithmic time. This matches the best known time bounds for the problem (among polynomial space data structures) while improving the space from superlinear to a constant number of ultrawords. Our results are based on a simple and elegant in-place word RAM data structure, known as the Fenwick tree. Our main technical contribution is a new efficient parallel ultra-wide word RAM implementation of the Fenwick tree, which is likely of independent interest.

cs.DS↗

arXiv2019

Space Efficient Construction of Lyndon Arrays in Linear Time

We present the first linear time algorithm to construct the $2n$-bit version of the Lyndon array for a string of length $n$ using only $o(n)$ bits of working space. A simpler variant of this algorithm computes the plain ($n\lg n$-bit) version of the Lyndon array using only $\mathcal{O}(1)$ words of additional working space. All previous algorithms are either not linear, or use at least $n\lg n$ bits of additional working space. Also in practice, our new algorithms outperform the previous best ones by an order of magnitude, both in terms of time and space.

cs.DS↗

arXiv2019

Decompressing Lempel-Ziv Compressed Text

We consider the problem of decompressing the Lempel--Ziv 77 representation of a string $S$ of length $n$ using a working space as close as possible to the size $z$ of the input. The folklore solution for the problem runs in $O(n)$ time but requires random access to the whole decompressed text. Another folklore solution is to convert LZ77 into a grammar of size $O(z\log(n/z))$ and then stream $S$ in linear time. In this paper, we show that $O(n)$ time and $O(z)$ working space can be achieved for constant-size alphabets. On general alphabets of size $σ$, we describe (i) a trade-off achieving $O(n\log^δσ)$ time and $O(z\log^{1-δ}σ)$ space for any $0\leq δ\leq 1$, and (ii) a solution achieving $O(n)$ time and $O(z\log\log (n/z))$ space. The latter solution, in particular, dominates both folklore algorithms for the problem. Our solutions can, more generally, extract any specified subsequence of $S$ with little overheads on top of the linear running time and working space. As an immediate corollary, we show that our techniques yield improved results for pattern matching problems on LZ77-compressed text.

cs.DS↗

arXiv2019

Top Tree Compression of Tries

We present a compressed representation of tries based on top tree compression [ICALP 2013] that works on a standard, comparison-based, pointer machine model of computation and supports efficient prefix search queries. Namely, we show how to preprocess a set of strings of total length $n$ over an alphabet of size $σ$ into a compressed data structure of worst-case optimal size $O(n/\log_σn)$ that given a pattern string $P$ of length $m$ determines if $P$ is a prefix of one of the strings in time $O(\min(m\log σ,m + \log n))$. We show that this query time is in fact optimal regardless of the size of the data structure. Existing solutions either use $Ω(n)$ space or rely on word RAM techniques, such as tabulation, hashing, address arithmetic, or word-level parallelism, and hence do not work on a pointer machine. Our result is the first solution on a pointer machine that achieves worst-case $o(n)$ space. Along the way, we develop several interesting data structures that work on a pointer machine and are of independent interest. These include an optimal data structures for random access to a grammar-compressed string and an optimal data structure for a variant of the level ancestor problem.

cs.DS↗

arXiv2019

From Regular Expression Matching to Parsing

Given a regular expression $R$ and a string $Q$, the regular expression parsing problem is to determine if $Q$ matches $R$ and if so, determine how it matches, e.g., by a mapping of the characters of $Q$ to the characters in $R$. Regular expression parsing makes finding matches of a regular expression even more useful by allowing us to directly extract subpatterns of the match, e.g., for extracting IP-addresses from internet traffic analysis or extracting subparts of genomes from genetic data bases. We present a new general techniques for efficiently converting a large class of algorithms that determine if a string $Q$ matches regular expression $R$ into algorithms that can construct a corresponding mapping. As a consequence, we obtain the first efficient linear space solutions for regular expression parsing.

cs.DS↗

arXiv2019

Mergeable Dictionaries With Shifts

We revisit the mergeable dictionaries with shift problem, where the goal is to maintain a family of sets subject to search, split, merge, make-set, and shift operations. The search, split, and make-set operations are the usual well-known textbook operations. The merge operation merges two sets and the shift operation adds or subtracts an integer from all elements in a set. Note that unlike the join operation on standard balanced search tree structures, such as AVL trees or 2-4 trees, the merge operation has no restriction on the key space of the input sets and supports merging arbitrarily interleaved sets. This problem is a key component in searching Lempel-Ziv compressed texts, in the mergeable trees problem, and in the union-split-find problem. We present the first solution achieving O(log U) amortized time for all operations, where {1, 2, ..., U} is the universe of the sets. This bound is optimal when the size of the universe is polynomially bounded by the sum of the sizes of the sets. Our solution is simple and based on a novel extension of biased search trees.

cs.DS↗

arXiv2018

Compressed Communication Complexity of Longest Common Prefixes

We consider the communication complexity of fundamental longest common prefix (Lcp) problems. In the simplest version, two parties, Alice and Bob, each hold a string, $A$ and $B$, and we want to determine the length of their longest common prefix $l=\text{Lcp}(A,B)$ using as few rounds and bits of communication as possible. We show that if the longest common prefix of $A$ and $B$ is compressible, then we can significantly reduce the number of rounds compared to the optimal uncompressed protocol, while achieving the same (or fewer) bits of communication. Namely, if the longest common prefix has an LZ77 parse of $z$ phrases, only $O(\lg z)$ rounds and $O(\lg \ell)$ total communication is necessary. We extend the result to the natural case when Bob holds a set of strings $B_1, \ldots, B_k$, and the goal is to find the length of the maximal longest prefix shared by $A$ and any of $B_1, \ldots, B_k$. Here, we give a protocol with $O(\log z)$ rounds and $O(\lg z \lg k + \lg \ell)$ total communication. We present our result in the public-coin model of computation but by a standard technique our results generalize to the private-coin model. Furthermore, if we view the input strings as integers the problems are the greater-than problem and the predecessor problem.

cs.DS↗

arXiv2018

Time-Space Trade-Offs for Lempel-Ziv Compressed Indexing

Given a string $S$, the \emph{compressed indexing problem} is to preprocess $S$ into a compressed representation that supports fast \emph{substring queries}. The goal is to use little space relative to the compressed size of $S$ while supporting fast queries. We present a compressed index based on the Lempel--Ziv 1977 compression scheme. We obtain the following time-space trade-offs: For constant-sized alphabets; (i) $O(m + occ \lg\lg n)$ time using $O(z\lg(n/z)\lg\lg z)$ space, or (ii) $O(m(1 + \frac{\lg^εz}{\lg(n/z)}) + occ(\lg\lg n + \lg^εz))$ time using $O(z\lg(n/z))$ space. For integer alphabets polynomially bounded by $n$; (iii) $O(m(1 + \frac{\lg^εz}{\lg(n/z)}) + occ(\lg\lg n + \lg^εz))$ time using $O(z(\lg(n/z) + \lg\lg z))$ space, or (iv) $O(m + occ(\lg\lg n + \lg^ε z))$ time using $O(z(\lg(n/z) + \lg^ε z))$ space, where $n$ and $m$ are the length of the input string and query string respectively, $z$ is the number of phrases in the LZ77 parse of the input string, $occ$ is the number of occurrences of the query in the input and $ε> 0$ is an arbitrarily small constant. In particular, (i) improves the leading term in the query time of the previous best solution from $O(m\lg m)$ to $O(m)$ at the cost of increasing the space by a factor $\lg \lg z$. Alternatively, (ii) matches the previous best space bound, but has a leading term in the query time of $O(m(1+\frac{\lg^ε z}{\lg (n/z)}))$. However, for any polynomial compression ratio, i.e., $z = O(n^{1-δ})$, for constant $δ> 0$, this becomes $O(m)$. Our index also supports extraction of any substring of length $\ell$ in $O(\ell + \lg(n/z))$ time. Technically, our results are obtained by novel extensions and combinations of existing data structures of independent interest, including a new batched variant of weak prefix search.

cs.DS↗

arXiv2017

A Separation Between Run-Length SLPs and LZ77

In this paper we give an infinite family of strings for which the length of the Lempel-Ziv'77 parse is a factor $Ω(\log n/\log\log n)$ smaller than the smallest run-length grammar.

cs.DS↗

arXiv2017

Fast Dynamic Arrays

We present a highly optimized implementation of tiered vectors, a data structure for maintaining a sequence of $n$ elements supporting access in time $O(1)$ and insertion and deletion in time $O(n^ε)$ for $ε> 0$ while using $o(n)$ extra space. We consider several different implementation optimizations in C++ and compare their performance to that of vector and multiset from the standard library on sequences with up to $10^8$ elements. Our fastest implementation uses much less space than multiset while providing speedups of $40\times$ for access operations compared to multiset and speedups of $10.000\times$ compared to vector for insertion and deletion operations while being competitive with both data structures for all other operations.

cs.DS↗

arXiv2017

Succinct Partial Sums and Fenwick Trees

We consider the well-studied partial sums problem in succint space where one is to maintain an array of n k-bit integers subject to updates such that partial sums queries can be efficiently answered. We present two succint versions of the Fenwick Tree - which is known for its simplicity and practicality. Our results hold in the encoding model where one is allowed to reuse the space from the input data. Our main result is the first that only requires nk + o(n) bits of space while still supporting sum/update in O(log_b n) / O(b log_b n) time where 2 <= b <= log^O(1) n. The second result shows how optimal time for sum/update can be achieved while only slightly increasing the space usage to nk + o(nk) bits. Beyond Fenwick Trees, the results are primarily based on bit-packing and sampling - making them very practical - and they also allow for simple optimal parallelization.

cs.DS↗

arXiv2017

Practical and Effective Re-Pair Compression

Re-Pair is an efficient grammar compressor that operates by recursively replacing high-frequency character pairs with new grammar symbols. The most space-efficient linear-time algorithm computing Re-Pair uses $(1+ε)n+\sqrt n$ words on top of the re-writable text (of length $n$ and stored in $n$ words), for any constant $ε>0$; in practice however, this solution uses complex sub-procedures preventing it from being practical. In this paper, we present an implementation of the above-mentioned result making use of more practical solutions; our tool further improves the working space to $(1.5+ε)n$ words (text included), for some small constant $ε$. As a second contribution, we focus on compact representations of the output grammar. The lower bound for storing a grammar with $d$ rules is $\log(d!)+2d\approx d\log d+0.557 d$ bits, and the most efficient encoding algorithm in the literature uses at most $d\log d + 2d$ bits and runs in $\mathcal O(d^{1.5})$ time. We describe a linear-time heuristic maximizing the compressibility of the output Re-Pair grammar. On real datasets, our grammar encoding uses---on average---only $2.8\%$ more bits than the information-theoretic minimum. In half of the tested cases, our compressor improves the output size of 7-Zip with maximum compression rate turned on.

cs.DS↗