Carla Groenland, Tom Johnston, Jamie Radcliffe and Alex Scott, Short reachability networks.
How many transpositions pi_1,...,pi_k do you need to be able to send the first two positions to any other tuple of distinct positions, i.e. such that for all x<y in {1,...,n}, there is a subsequence 1<=i_1<...<i_j<=k for which the composition pi_{i_1} \circ \dots \circ pi_{i_j} sends 1 to x and 2 to y? We solve this question and study related problems.
Carla Groenland, Tom Johnston, Jamie Radcliffe and Alex Scott, Perfect shuffling with fewer lazy transpositions.
Related to the problem above, what now if each transposition pi_i cannot be freely chosen to be present or not, but rather equals pi_i with a probability p_i and the identity with probability 1-p_i? Each transposition can be assigned a different probability. The resulting composition of random transpositions is now a random permutation. How many are needed to obtain the uniformly random permutation? We provide a construction disproving the natural conjecture that (n choose 2) random transpositions are needed.
Carla Groenland and Tom Johnston, The lengths for which bicrucial square-free permutations exist. Journal publication: PP2021 special issue of Enumerative Combinatorics and Applications.
We characterize for which lengths bicrucial square-free permutations exist via a combination of reductions and computer search.
Carla Groenland, Tom Johnston, Dániel Korándi, Alexander Roberts, Alex Scott and Jane Tan, Decomposing random permutations into order-isomorphic subpermutations, SIAM Journal of Discrete Mathematics.
Given two permutations, what is the largest pattern they have in common? What if the permutations are chosen uniformly at random? And once you have a single pattern in common, can you find more? This motivates the following problem: given two uniformly random permutations, how many parts k do you need, so that the first can be decomposed into s_1,...,s_k and the second into p_1,...,p_k, with p_i order-isomorphic to (''the same as'')p_i for all i? We answer this question (up to log-factors).