30 Semi-Synthetic Data Sets for X-ray Diffraction Error Analysis

Datasets as described several years ago in https://journals.iucr.org/d/issues/2016/04/00/gm5043/index.html#SEC10 30 low-dose data sets (i.e. a `large' number) were recorded sequentially on beamline I04 at Diamond Light Source with identical data-collection parameters (1027 images per data set from a Pilatus 6M detector with an image width of 0.15° and 1% beam transmission at a wavelength of 1.2 Å). Scaling all of the data together indicated that radiation damage across the 30 sweeps was small; however, some systematic differences between them remained owing to factors such as beam-intensity variation. The photon counts from these 30 `original' sweeps were then `reshuffled' to create a population of 30 equivalent `new' data sets (for convenience, to allow reuse of the image headers) by considering every active pixel in the data set (i.e. each of around six billion) independently using the following procedure. Firstly, create a summed data set, which we call `total': for image in range(1027): for pixel in image: total[image][pixel] = sum(images[j][image][pixel] for j in range(30)) Then create 30 `new' data sets with every pixel set initially to 0, and randomly redistribute each photon count from every pixel of the `total' set to the same pixel position in one of the 30 `new' sets: for image in range(1027): for pixel in image: for k in range(total[image, pixel]): rnd = random(30) rebin[rnd, image, pixel] += 1