Parameterized Complexity

It asks whether problems that are intractable in total input size can be solved efficiently when a feature such as solution size, dimension, or graph treewidth is small. Fixed-parameter algorithms target runtime of the form f(k)n^O(1), using tools such as tree-decomposition dynamic programming, kernelization, and backdoor reductions; approximation and streaming methods also exploit small parameters. W-hardness results and ETH-based lower bounds distinguish fixed-parameter tractability from XP algorithms whose polynomial exponent grows with the parameter, and reveal limits in areas including neural-network verification, optimization, and reasoning. A central challenge is choosing a parameter that captures useful structure: treewidth or combinations of parameters can unlock tractability, while dimension or numeric ranges may remain hard on their own.