Improved Online Sorting
By: Jubayer Nirjhor, Nicole Wein
Potential Business Impact:
Sorts numbers faster, saving space and time.
We study the online sorting problem, where $n$ real numbers arrive in an online fashion, and the algorithm must immediately place each number into an array of size $(1+\varepsilon) n$ before seeing the next number. After all $n$ numbers are placed into the array, the cost is defined as the sum over the absolute differences of all $n-1$ pairs of adjacent numbers in the array, ignoring empty array cells. Aamand, Abrahamsen, Beretta, and Kleist introduced the problem and obtained a deterministic algorithm with cost $2^{O\left(\sqrt{\log n \cdot\log\log n +\log \varepsilon^{-1}}\right)}$, and a lower bound of $\Omega(\log n / \log\log n)$ for deterministic algorithms. We obtain a deterministic algorithm with quasi-polylogarithmic cost $\left(\varepsilon^{-1}\log n\right)^{O\left(\log \log n\right)}$. Concurrent and independent work by Azar, Panigrahi, and Vardi achieves polylogarithmic cost $O(\varepsilon^{-1}\log^2 n)$.
Similar Papers
Nearly Tight Bounds for the Online Sorting Problem
Data Structures and Algorithms
Sorts numbers efficiently using less computer memory.
Nearly Optimal Bounds for Stochastic Online Sorting
Data Structures and Algorithms
Makes sorting numbers faster when they arrive randomly.
A Polylogarithmic Algorithm for Stochastic Online Sorting
Data Structures and Algorithms
Sorts numbers better when they arrive randomly.