Photo

Konstantin Tikhomirov

Email: ktikhomi at andrew dot cmu dot edu

Some publications and preprints

Paper Statement of AI Use
D. J. Altschuler, K. Tikhomirov The threshold for online balancing of i.i.d. binary vectors. arXiv:2609.14975 The discrepancy lower bound as well as the upper bound in the very sparse regime were proved without AI assistance. Subsequently, also without AI, the upper bound was extended to the full range of sparsity considered in the current paper under the simplifying assumption that the non-zero entries of incoming vectors are independent and uniform on {-1,1}, rather than deterministically 1. The random signing condition was removed with help of ChatGPT. Further, ChatGPT suggested and implemented potential function approach in place of an earlier multiscale majority argument.
D. J. Altschuler, Q. Dubroff, K. Tikhomirov Online Permutation Embedding: Optimal Stopping and Scaling Laws. arXiv:2608.19050 Original proof of the recursive formula for the expectation of the optimal online embedding time was developed without AI assistance and was based on a discretization argument and a reduction to 231-avoiding permutations. The much more concise martingale-based argument was obtained with help of ChatGPT.
K. Tikhomirov Level-set entropy and sparse randomized embeddings. arXiv:2607.23017 [To be revised] Architecture of the proof (including the use of level set entropy estimates; decomposition into light and heavy columns; canonical and reduced flat contributions; two level partitions) is developed without AI assistance. ChatGPT contribution to the proofs is at the level of a co-author. The proof of the (key) entropy lemma for the level sets is generated by ChatGPT.
D. J. Altschuler, K. Tikhomirov Online Beck--Fiala Down to Logarithmic Sparsity. arXiv:2607.14238 All proofs are ChatGPT-generated. Authors wrote high-level prompts and revised ChatGPT output.
H. Huang, M. Rudelson, K. Tikhomirov Well-invertible column subsets of sparse matrices are rare. arXiv:2607.05384 In the setting of random matrices, the main result of the work was derived by the authors without AI assistance. Subsequent generalization to deterministic matrices with the column overlap condition was obtained with substantial use of ChatGPT.
H. Huang, K. Tikhomirov Cotype of random polytopes. arXiv:2603.04749 [To be revised] No AI use

Before 2026