This is a collection of sets of artificially generated crystallographic textures of steel sheets, histogram-based representations of them and their corresponding properties. The properties are calculated using a Taylor-type crystal plasticity model.

The collection contains the following data sets which are all save in numpy's .npy format:

* training set - a set of 50000 samples consisting of
** lists of orientations forming the originally generated data. Array size [50000 samples, 8000 orientations per sample, Bunge euler angles in degree]
** the orientation lists transformed into an angular histogram-based representation (512 uniformly distributed bins in the cubic-orthorhombic fundamental zone, soft-assignment parameter = 3).  Array size [50000 samples, weight vector of the 512 bins]
** corresponding properties. Array size [50000 samples, Young's modulus and R-values at 0, 45 and 90 degree to rolling direction]

* test set - a set of 10000 samples consisting of
** lists of orientations forming the originally generated data. Array size [10000 samples, 8000 orientations per sample, Bunge euler angles in degree]
** the orientation lists transformed into an angular histogram-based representation (512 uniformly distributed bins in the cubic-orthorhombic fundamental zone, soft-assignment parameter = 3). Array size [10000 samples, weight vector of the 512 bins]
** corresponding properties. Array size [10000 samples, Young's modulus and R-values at 0, 45 and 90 degree to rolling direction]

* anomalie set - a set of 10000 samples consisting of
** lists of orientations forming the originally generated data. Array size [10000 nsamples, 8000 orientations per sample, Bunge euler angles in degree]
** the orientation lists transformed into an angular histogram-based representation (512 uniformly distributed bins in the cubic-orthorhombic fundamental zone, soft-assignment parameter = 3). Array size [10000 samples, weight vector of the 512 bins]