Code for the VIS2025 paper 'Locally Adapted Reference Frame Fields using Moving Least Squares'
https://doi.org/10.1109/TVCG.2025.3634845

Based on the code in https://github.com/fau-vc/optimal-frames


============================================
How to build
============================================
Use CMake to setup your build environment. 
Two build paths controlled with CMake option USE_CUDA: CPU or CUDA (recommended)
During configuration, the path to VTK (build folder), Eigen (header-only library) and CUDA (if enabled) must be specified.

============================================
How to use
============================================
Once compiled, a custom experiment can be run from the command line using:
'LocalAdaptiveOptimalFrames2D.exe dataset_name lambda_weights nbU FTLE_window lambda_Tikhonov'
1 : dataset_name (string, without extension), .vti dataset name 
2 : lambda_weights (double), parameter for the weighting function e^(-lambda_w*distance*distance)
3 : nbU (int), 'radius' of neighborhood. Usually it is 15, yielding a 31*31 neighborhood
4 : FTLE_window (int), integration time in integer number of timesteps. Usually 80
5 : lambda_Tikhonov (double), Tikhonov regularization parameter. Usually 10, 0 sets uniform weights (Baseline)

Datasets:
The datasets should be in the .vti format, and include two float arrays called 'u' and 'v' for each component of the 2D unsteady field.
Alongside the executable, create a folder called 'datasets' and place the datasets inside
An example dataset is included in ./datasets. Copy the folder alongside the executable and run


============================================
Requirements
============================================
Dependencies:
- C++ 17                 (tested with Visual Studio 2022)
- CMake                  (tested with CMake 4.0.3)
- VTK                    (tested with VTK 9.3)
- Eigen                  (tested with Eigen 3.X)
- CUDA					 (tested with CUDA 13.1)

Tested on:
- 
Windows 10, Intel Xeon CPU E5-2687W (3GHz) and an Nvidia GeForce RTX3090.