Random unfriendly seating arrangement in a dining table (Q2343189)

In this article, the so called ``unfriendly seating problem'' is analysed for \(2\times n\) dimensions. The general problem \((m\times n)\) has been stated in the sixties and solved for \(1\times n\) (expected number of occupied seats, error relation). Here, it is shown that the \(2\times n\) problem has some surprising properties compared with the \(1\times n\) problem:{\parindent=0.7cm\begin{itemize}\item[--] There exists a configuration with more empty seats available and the expected value of occupied seats is lower than the configuration with less seats.\item[--] Also a configuration is shown with more empty seats available and the probability of having more people seated is lower than the probability for a configuration with less seats. \end{itemize}} For the \(2\times n\) problem, a recurrence relation is developed and, using this recurrence, a differential equation (of Riccati type) for the corresponding generating functions is stated and solved. The main result is an expected value (of occupied seats) and an error bound for all \(n\). For large \(n\) (the limit of \(n\) going to infinity, respectively) the distribution is shown to be asymptotically normal. At the end, the dynamic approach of seating people is compared with the combinatorial approach (counting all possible seatings) and the main result is that the ``space utilization is better in the sequential model than in the combinatorial model''.