Kaiserslautern - Fachbereich Informatik
Refine
Document Type
- Bachelor Thesis (1)
- Doctoral Thesis (1)
- Master's Thesis (1)
Language
- English (3)
Has Fulltext
- yes (3)
Keywords
Faculty / Organisational entity
Dual-Pivot Quicksort and Beyond: Analysis of Multiway Partitioning and Its Practical Potential
(2016)
Multiway Quicksort, i.e., partitioning the input in one step around several pivots, has received much attention since Java 7’s runtime library uses a new dual-pivot method that outperforms by far the old Quicksort implementation. The success of dual-pivot Quicksort is most likely due to more efficient usage of the memory hierarchy, which gives reason to believe that further improvements are possible with multiway Quicksort.
In this dissertation, I conduct a mathematical average-case analysis of multiway Quicksort including the important optimization to choose pivots from a sample of the input. I propose a parametric template algorithm that covers all practically relevant partitioning methods as special cases, and analyze this method in full generality. This allows me to analytically investigate in depth what effect the parameters of the generic Quicksort have on its performance. To model the memory-hierarchy costs, I also analyze the expected number of scanned elements, a measure for the amount of data transferred from memory that is known to also approximate the number of cache misses very well. The analysis unifies previous analyses of particular Quicksort variants under particular cost measures in one generic framework.
A main result is that multiway partitioning can reduce the number of scanned elements significantly, while it does not save many key comparisons; this explains why the earlier studies of multiway Quicksort did not find it promising. A highlight of this dissertation is the extension of the analysis to inputs with equal keys. I give the first analysis of Quicksort with pivot sampling and multiway partitioning on an input model with equal keys.
Recently, a new Quicksort variant due to Yaroslavskiy was chosen as standard sorting
method for Oracle's Java 7 runtime library. The decision for the change was based on
empirical studies showing that on average, the new algorithm is faster than the formerly
used classic Quicksort. Surprisingly, the improvement was achieved by using a dual pivot
approach — an idea that was considered not promising by several theoretical studies in the
past. In this thesis, I try to find the reason for this unexpected success.
My focus is on the precise and detailed average case analysis, aiming at the flavor of
Knuth's series “The Art of Computer Programming”. In particular, I go beyond abstract
measures like counting key comparisons, and try to understand the efficiency of the
algorithms at different levels of abstraction. Whenever possible, precise expected values are
preferred to asymptotic approximations. This rigor ensures that (a) the sorting methods
discussed here are actually usable in practice and (b) that the analysis results contribute to
a sound comparison of the Quicksort variants.
It has been observed that for understanding the biological function of certain RNA molecules, one has to study joint secondary structures of interacting pairs of RNA. In this thesis, a new approach for predicting the joint structure is proposed and implemented. For this, we introduce the class of m-dimensional context-free grammars --- an extension of stochastic context-free grammars to multiple dimensions --- and present an Earley-style semiring parser for this class. Additionally, we develop and thoroughly discuss an implementation variant of Earley parsers tailored to efficiently handle dense grammars, which embraces the grammars used for structure prediction. A currently proposed partitioning scheme for joint secondary structures is transferred into a two-dimensional context-free grammar, which in turn is used as a stochastic model for RNA-RNA interaction. This model is trained on actual data and then used for predicting most likely joint structures for given RNA molecules. While this technique has been widely used for secondary structure prediction of single molecules, RNA-RNA interaction was hardly approached this way in the past. Although our parser has O(n^3 m^3) time complexity and O(n^2 m^2) space complexity for two RNA molecules of sizes n and m, it remains practically applicable for typical sizes if enough memory is available. Experiments show that our parser is much more efficient for this application than classical Earley parsers. Moreover the predictions of joint structures are comparable in quality to current energy minimization approaches.