The known linear-time kernelizations for d-Hitting Set guarantee linear worst-case running times using a quadratic-size data structure (that is not fully initialized). Getting rid of this data structure, we show that problem kernels of asymptotically optimal size O(kd) for d-Hitting Set are computable in linear time and space. Additionally, we experimentally compare the linear-time kernelizations for d-Hitting Set to each other and to a classical data reduction algorithm due to Weihe.
- Combinatorial problems
- Data reduction
- NP-hard problem
- Parameterized complexity
- 1.02 COMPUTER AND INFORMATION SCIENCES