ebsdtorch 0.0.40

No project description provided

EBSDTorch

PyTorch-only library for electron backscatter diffraction (EBSD) patterns.

Installation

To install EBSDTorch, first install PyTorch, then run this command in your terminal:

pip install ebsdtorch

Features (and TODOs)

• :white_check_mark: wide GPU support via PyTorch device abstraction & backends

• :white_check_mark: Uniform sampling on sphere / SO(3)

• :white_check_mark: Laue symmetry operations on sphere / SO(3)

• :white_check_mark: Modified square Lambert projection and inverse

• :white_check_mark: dictionary indexing (conventional pixel space)

• :white_check_mark: dictionary indexing (covariance matrix PCA)

• :white_large_square: dictionary indexing (Halko randomized PCA)

• :white_check_mark: 8-bit Quantization on CPU for fast indexing

• :white_large_square: 8-bit Quantization on GPU for (very) fast indexing

• :white_large_square: Further reduced bit depth quantization (CPU or GPU)

• :white_check_mark: EBSD master pattern direct space convolution with detector annulus

• :white_check_mark: Spherical covariance matrix calculation

• :white_large_square: Spherical covariance matrix interpolation onto detector

• :white_check_mark: pattern projection with average projection center

• :white_check_mark: pattern projection with individual projection centers

• :white_large_square: pattern projection with single camera matrix

• :white_large_square: pattern center fitting (conventional)

• :white_large_square: geometry fitting (single camera matrix)

• :white_check_mark: Wigner D matrices

• :white_large_square: spherical harmonics

• :white_large_square: SO3 FFT for cross correlation / convolution

• :white_large_square: EBSD master pattern blur via SO3 FFT (for BSE image simulation)

• :white_large_square: Support for generic crystal unit cells

• :white_large_square: Monte Carlo backscatter electron simulation

• :white_large_square: Dynamical scattering simulation