Data for publication: 3D local measurement of the magnetic field strength inside Martian meteorite NWA 7034 by polarimetric neutron imaging

The following files contain the relevant data and analysis notebooks used for the work pertaining the article: "3D local measurement of the magnetic field strength inside Martian meteorite NWA 7034 by polarimetric neutron imaging"Analysis was done by Cédric Qvistgaard: https://orcid.org/0009-0007-3341-4508Authors: Cédric Holme Qvistgaard, Estrid Buhl Naver, Alexander Wolfertz, Vahid Karimi, Takenao Shinohara, Joseph Don Parker, Tetsuya Kai, Hirotoshi Hayashida, Søren Schmidt, and Luise Theil Kuhn.Neutron polarimetric imaging performed of a sample of meteorite North West Africa (NWA) 7034 at the RADEN beamline at JPARC in Japan.<br>It should contain 2 notebooks,1 tiff file and 4 npy files.<br>Rock1_Stack.npy is the polarization images for each tomography angle and component. It should be a 33 x 9 x 128 x 128 voxel grid, with 33 angles spanning 0 to 360 degrees evenly, and the 9 matrix components [Pxx,Pxy,Pxz,Pyx,Pyy,Pyz,Pzx,Pzy,Pzz]. CCBY-4.0 (https://creativecommons.org/licenses/by/4.0/)<br>Rock1_Stack_Cut.npy is a cut 80 x 80 ROI version used for the reconstruction. CCBY-4.0<br>Structure_Recon_rot.tiff contains the structure reconstructed from the attenuation images. CCBY-4.0<br>B_MSE.tiff is the final reconstruction that we have used for further plotting etc.CCBY-4.0<br>Raden_Beam_Profile.npy shows the wavelength profile of the beam, in case a more accurate ToF weighted reconstruction would be needed. This does drastically increase computing power required though. CCBY-4.0<br>Iterative Solver Mars Rock.ipybn is the document wherein most of the analysis was done. In broad strokes it:<br>- Initiates the GPU, and finds the mean wavelength of the beam.<br>- Defines the forward simulation via PNI_experiment, which simply loops through each voxel and updates the spin orientation of each pixel accordingly.<br>- Defines an optimizer which utilizes gradient descent backpropagation and the ADAM optimizer from pytorch to "guess" new structures until the forward agrees with our measurements.<br>- Defines some plotting functions to compare simulated components with measurements, and produce a 3D plot.<br>Copyright @ Technical University of Denmark, BSD-3-Clause (https://opensource.org/license/bsd-3-clause)<br>The notebook "Comparison and visualization.ipybn" contains further plotting functions utilized, such as the slice quiver plot. However, for most purposes the "iterative solver" document can be run alone.<br>Copyright @ Technical University of Denmark, BSD-3-Clause<br>