This publication contains a set of 76980 samples of crystallographic textures (as lists of orientations) and corresponding properties (Youngs modulus E and an anisotropy measure R*, similar to the Lankford coefficients, in three room directions). The data originates from a simulated multi-step metal forming process, which was published in [1] and described in detail in [2]. In contrast to [2], the simulation was constrained to perform seven successive process steps of 10% strain at the material point in different directions. In each step, the orientation of the tension operation is chosen randomly from a set of 25 uniformly distributed orientations in the orientation space SO(3). 

The crystallographic texture-properties data is contained in the following files:
- properties.npy, which contains a 2d-array (number of samples, six properties). The properties are the Youngs moduli E11, E22, E33 and the anisotropy measures R*11, R*22, R*33
- orentations.npy, which contains a 3d-array (number of samples, 512 orientations, three Euler angles). The Euler angles are defined by the Bunge convention and are given in degree.

For learning the relation between crystallographic texture and properties, we suggest using randomly defined train/test sets. For a split of 80% train and 20% test data, the indices that are listed in the following files can be used:
- ids_train.txt
- ids_test.txt 

References
[1] L. Morand, J. Dornheim, J. Pagenkopf, D. Helm. Simulation of texture evolution for a multi-step metal forming process, https://fordatis.fraunhofer.de/handle/fordatis/201.2
[2] J. Dornheim, L. Morand, S. Zeitvogel, T. Iraki, N. Link, D. Helm. Deep reinforcement learningmethods for structure-guided processing path optimization, Journal of Intelligent Manufacturing 33:333–352, 2022