RealTensorValue in RotationMatrix

[jessecarterMOOSE]

For a RealTensorValue type, do you fill in values moving left to right across rows (how arrays are usually done), or do you fill down each column?

I ask because I'm looking at the rotation matrix from euler angles computed here. It looks like it is following the Z1X2Z3 convention as expected, but for instance the 2nd entry (row 1 column 2) isn't going in the 2nd entry for the variable, it's going in the 4th (index 3), which is where would expect column 1 row 2, unless it really does need to be filled in top to bottom. Or perhaps the function is just doing the transpose?

[aeslaughter]

@dschwen Do you know the answer to this?

[dschwen]

@SudiptaBiswas worked with Euler angle rotations in her sintering work.

[jessecarterMOOSE]

@SudiptaBiswas Just curious if you know the answer to this. It looks like the rotation matrix being built in https://github.com/idaholab/moose/blob/next/modules/tensor_mechanics/src/utils/RotationTensor.C is the transpose of what you would expect based on the vector indices being stored, assuming you enter in the values row-wise (index 0 is R11, index 1 is R12, etc). Depending on the application, the transpose may be what you want, but I just want to be clear what's going on. The Euler angles for a rotation matrix and its transpose will be different.

[jessecarterMOOSE]

In case anybody finds this, I think the answer is clear and I just wasn't seeing it. As the documentation clearly stated, the Bunge convention is used for the Euler angles, meaning the angles are used to build the passive (sample-to-crystal) rotation matrix, so yes, the transpose of the matrix is computed and applied when using Euler angles.

The documentation has been updated a bit. See also #16341. I added some functionality such that one can apply a rotation matrix directly to the elasticity tensor.