Kolmogorov complexity and combinatorial methods in communication complexity

We introduce a method based on Kolmogorov complexity to prove lower bounds on communication complexity. The intuition behind our technique is close to information theoretic methods.We use Kolmogorov complexity for three different things: first, to give a general lower bound in terms of Kolmogorov mutual information; second, to prove an alternative to Yao’s minmax principle based on Kolmogorov complexity; and finally, to identify hard inputs.We show that our method implies the rectangle and corruption bounds, known to be closely related to the subdistribution bound. We apply our method to the hidden matching problem, a relation introduced to prove an exponential gap between quantum and classical communication. We then show that our method generalizes the VC dimension and shatter coefficient lower bounds. Finally, we compare one-way communication and simultaneous communication in the case of distributional communication complexity and improve the previous known result.