Quanquan C. Liu

Quanquan C. Liu, Vaidehi Srinivas

We formulate the predicted-updates dynamic model, one of the first beyond-worst-case models for dynamic algorithms, which generalizes a large set of well-studied dynamic models including the offline dynamic, incremental, and decremental models to the fully dynamic setting when given predictions about the update times of the elements. In the most basic form of our model, we receive a set of predicted update times for all of the updates that occur over the event horizon. We give a novel framework that "lifts" offline divide-and-conquer algorithms into the fully dynamic setting with little overhead. Using this, we are able to interpolate between the offline and fully dynamic settings; when the $\ell_1$ error of the prediction is linear in the number of updates, we achieve the offline runtime of the algorithm (up to $\mathrm{poly} \log n$ factors). Provided a fully dynamic backstop algorithm, our algorithm will never do worse than the backstop algorithm regardless of the prediction error. Furthermore, our framework achieves a smooth linear trade-off between $\ell_1$ error in the predictions and runtime. These correspond to the desiderata of consistency, robustness, and graceful degradation of the algorithms-with-predictions literature. We further extend our techniques to incremental and decremental settings, transforming algorithms in these settings when given predictions of only the deletion and insertion times, respectively. Our framework is general, and we apply it to obtain improved efficiency bounds over the state-of-the-art dynamic algorithms for a variety of problems including triconnectivity, planar digraph all pairs shortest paths, $k$-edge connectivity, and others, for prediction error of reasonable magnitude.

Thaddeus Dryja, Quanquan C. Liu, Sunoo Park

Pebble games were originally formulated to study time-space tradeoffs in computation, modeled by games played on directed acyclic graphs (DAGs). Close connections between pebbling and cryptography have been known for decades. A series of recent research starting with (Alwen and Serbinenko, STOC 2015) has deepened our understanding of the notion of memory-hardness in cryptography --- a useful property of hash functions for deterring large-scale password-cracking attacks --- and has shown memory-hardness to have intricate connections with the theory of graph pebbling. In this work, we improve upon two main limitations of existing models of memory-hardness. First, existing measures of memory-hardness only account for dynamic (i.e., runtime) memory usage, and do not consider static memory usage. We propose a new definition of static-memory-hard function (SHF) which takes into account static memory usage and allows the formalization of larger memory requirements for efficient functions, than in the dynamic setting (where memory usage is inherently bounded by runtime). We then give two SHF constructions based on pebbling; to prove static-memory-hardness, we define a new pebble game ("black-magic pebble game"), and new graph constructions with optimal complexity under our proposed measure. Secondly, existing memory-hardness models implicitly consider linear tradeoffs between the costs of time and space. We propose a new model to capture nonlinear time-space trade-offs and prove that nonlinear tradeoffs can in fact cause adversaries to employ different strategies from linear tradeoffs. Finally, as an additional contribution of independent interest, we present the first asymptotically tight graph construction that achieves the best possible space complexity up to $\log{\log{n}}$-factors for an existing memory-hardness measure called cumulative complexity in the sequential pebbling model.