This project is the work in progress for the practical part of my masters thesis.
The engine code is based on https://github.com/piellardj/water-webgpu.
The data format is CSV. Use the file picker to select a csv file from your computer. After selecting the parsing begins. You can press the save button to store a direct copy of the csv file inside the browser file system. This is handy if you want to use a dataset multiple times.
The GPU simulation uses euler integration. The basic principle is that for each frame, a number of physic iterations are computed. A single iteration uses the time delta specified in the time settings. During one frame the specified number of substeps will happen.
The radius scaling is a normalized scaling factor between 0 and 1 that determines the actual point radius used in the GPU computation. This is handy when the points are too close together to see a correlation.
- Click an axis
- Select spaghetti as layout
- Select a column
- The selected axis will be the primary axis where the unique line attributes will be shown and the other axis is the time axis
- Either click an axis or the UMAP symbol on the top (this determines if UMAP will be 1 dimensional or 2 dimensional)
- Select UMAP as layout
- Select a set of features you want to project
- Click an axis
- Select group by as layout
- Select a column
- The points are grouped along the axis
