arXiv subjects

Susanne Albers

Publications and source records attributed to Susanne Albers.

23 records · Page 2Linked to original sources

arXiv2017

Tight Bounds for Online Coloring of Basic Graph Classes

We resolve a number of long-standing open problems in online graph coloring. More specifically, we develop tight lower bounds on the performance of online algorithms for fundamental graph classes. An important contribution is that our bounds also hold for randomized online algorithms, for which hardly any results were known. Technically, we construct lower bounds for chordal graphs. The constructions then allow us to derive results on the performance of randomized online algorithms for the following further graph classes: trees, planar, bipartite, inductive, bounded-treewidth and disk graphs. It shows that the best competitive ratio of both deterministic and randomized online algorithms is $Θ(\log n)$, where $n$ is the number of vertices of a graph. Furthermore, we prove that this guarantee cannot be improved if an online algorithm has a lookahead of size $O(n/\log n)$ or access to a reordering buffer of size $n^{1-ε}$, for any $0<ε\leq 1$. A consequence of our results is that, for all of the above mentioned graph classes except bipartite graphs, the natural $\textit{First Fit}$ coloring algorithm achieves an optimal performance, up to constant factors, among deterministic and randomized online algorithms.

cs.DS↗

arXiv2017

On the Value of Penalties in Time-Inconsistent Planning

People tend to behave inconsistently over time due to an inherent present bias. As this may impair performance, social and economic settings need to be adapted accordingly. Common tools to reduce the impact of time-inconsistent behavior are penalties and prohibition. Such tools are called commitment devices. In recent work Kleinberg and Oren connect the design of prohibition-based commitment devices to a combinatorial problem in which edges are removed from a task graph $G$ with $n$ nodes. However, this problem is NP-hard to approximate within a ratio less than $\sqrt{n}/3$. To address this issue, we propose a penalty-based commitment device that does not delete edges but raises their cost. The benefits of our approach are twofold. On the conceptual side, we show that penalties are up to $1/β$ times more efficient than prohibition, where $β\in (0,1]$ parameterizes the present bias. On the computational side, we significantly improve approximability by presenting a $2$-approximation algorithm for allocating the penalties. To complement this result, we prove that optimal penalties are NP-hard to approximate within a ratio of $1.08192$.

cs.DS↗

arXiv2016

Motivating Time-Inconsistent Agents: A Computational Approach

In this paper we investigate the computational complexity of motivating time-inconsistent agents to complete long term projects. We resort to an elegant graph-theoretic model, introduced by Kleinberg and Oren, which consists of a task graph $G$ with $n$ vertices, including a source $s$ and target $t$, and an agent that incrementally constructs a path from $s$ to $t$ in order to collect rewards. The twist is that the agent is present-biased and discounts future costs and rewards by a factor $β\in [0,1]$. Our design objective is to ensure that the agent reaches $t$ i.e.\ completes the project, for as little reward as possible. Such graphs are called motivating. We consider two strategies. First, we place a single reward $r$ at $t$ and try to guide the agent by removing edges from $G$. We prove that deciding the existence of such motivating subgraphs is NP-complete if $r$ is fixed. More importantly, we generalize our reduction to a hardness of approximation result for computing the minimum $r$ that admits a motivating subgraph. In particular, we show that no polynomial-time approximation to within a ratio of $\sqrt{n}/4$ or less is possible, unless ${\rm P}={\rm NP}$. Furthermore, we develop a $(1+\sqrt{n})$-approximation algorithm and thus settle the approximability of computing motivating subgraphs. Secondly, we study motivating reward configurations, where non-negative rewards $r(v)$ may be placed on arbitrary vertices $v$ of $G$. The agent only receives the rewards of visited vertices. Again we give an NP-completeness result for deciding the existence of a motivating reward configuration within a fixed budget $b$. This result even holds if $b=0$, which in turn implies that no efficient approximation of a minimum $b$ within a ration grater or equal to $1$ is possible, unless ${\rm P}={\rm NP}$.

cs.CC↗

arXiv2013

Online Makespan Minimization with Parallel Schedules

In online makespan minimization a sequence of jobs $σ= J_1,..., J_n$ has to be scheduled on $m$ identical parallel machines so as to minimize the maximum completion time of any job. We investigate the problem with an essentially new model of resource augmentation. Here, an online algorithm is allowed to build several schedules in parallel while processing $σ$. At the end of the scheduling process the best schedule is selected. This model can be viewed as providing an online algorithm with extra space, which is invested to maintain multiple solutions. The setting is of particular interest in parallel processing environments where each processor can maintain a single or a small set of solutions. We develop a $(4/3+\eps)$-competitive algorithm, for any $0<\eps\leq 1$, that uses a number of $1/\eps^{O(\log (1/\eps))}$ schedules. We also give a $(1+\eps)$-competitive algorithm, for any $0<\eps\leq 1$, that builds a polynomial number of $(m/\eps)^{O(\log (1/\eps) / \eps)}$ schedules. This value depends on $m$ but is independent of the input $σ$. The performance guarantees are nearly best possible. We show that any algorithm that achieves a competitiveness smaller than 4/3 must construct $Ω(m)$ schedules. Our algorithms make use of novel guessing schemes that (1) predict the optimum makespan of a job sequence $σ$ to within a factor of $1+\eps$ and (2) guess the job processing times and their frequencies in $σ$. In (2) we have to sparsify the universe of all guesses so as to reduce the number of schedules to a constant. The competitive ratios achieved using parallel schedules are considerably smaller than those in the standard problem without resource augmentation.

cs.DS↗

arXiv2012

On the Value of Job Migration in Online Makespan Minimization

Makespan minimization on identical parallel machines is a classical scheduling problem. We consider the online scenario where a sequence of $n$ jobs has to be scheduled non-preemptively on $m$ machines so as to minimize the maximum completion time of any job. The best competitive ratio that can be achieved by deterministic online algorithms is in the range $[1.88,1.9201]$. Currently no randomized online algorithm with a smaller competitiveness is known, for general $m$. In this paper we explore the power of job migration, i.e.\ an online scheduler is allowed to perform a limited number of job reassignments. Migration is a common technique used in theory and practice to balance load in parallel processing environments. As our main result we settle the performance that can be achieved by deterministic online algorithms. We develop an algorithm that is $α_m$-competitive, for any $m\geq 2$, where $α_m$ is the solution of a certain equation. For $m=2$, $α_2 = 4/3$ and $\lim_{m\rightarrow \infty} α_m = W_{-1}(-1/e^2)/(1+ W_{-1}(-1/e^2)) \approx 1.4659$. Here $W_{-1}$ is the lower branch of the Lambert $W$ function. For $m\geq 11$, the algorithm uses at most $7m$ migration operations. For smaller $m$, $8m$ to $10m$ operations may be performed. We complement this result by a matching lower bound: No online algorithm that uses $o(n)$ job migrations can achieve a competitive ratio smaller than $α_m$. We finally trade performance for migrations. We give a family of algorithms that is $c$-competitive, for any $5/3\leq c \leq 2$. For $c= 5/3$, the strategy uses at most $4m$ job migrations. For $c=1.75$, at most $2.5m$ migrations are used.

cs.DS↗