Final density fields of the low-resolution box as chain 10 evolves. This chain has F_eff = 411, so every F_eff sample (going vertically down each column) can be considered as an independent sample. The circles highlight a few haloes that do not shift by much between consecutive samples, illustrating the substantial autocorrelation between samples separated by significantly less than F_eff. We note that sample 0 (which occurs after one iteration of the HMC) is almost completely uniform because we start our chains from a uniform field. Visualisation here was done using pySPHViewer (Benitez-Llambay 2015).