DLAs were my "breakthrough" in college. Where I could first apply my self-learned c++ chops to code some fast DLA number crunching including a real-time OpenGL visualization. Professors were happy to pay me money for code from there on, which in hindsight kickstarted my career.
DLAs have some interesting properties but aren't really important in the grand scheme of things. They also can't be parallelized (or it's really hard). And the quality of the results highly depends on the quality of the RNG. It's also an interesting problem to quickly appreciate memory-boundness as a performance bottleneck. In general it's a good problem for a beginner to think about performance and how to approach optimization as the general logic is almost trivial.
Back then I also wrote a web-based 3D DLA simulation with a threejs visualization. But as so often is the fate of web-apps: This no longer runs today. Something about a used dependency (highcharts) being blocked.
Fun memories, together with the equally cherished Ising model. Seeing these patterns still puts a smile on my face. Btw they can be made to look nicer if you simulate a larger lattice and sample it down to a smaller resolution to emulate anti-aliasing.
DLAs have some interesting properties but aren't really important in the grand scheme of things. They also can't be parallelized (or it's really hard). And the quality of the results highly depends on the quality of the RNG. It's also an interesting problem to quickly appreciate memory-boundness as a performance bottleneck. In general it's a good problem for a beginner to think about performance and how to approach optimization as the general logic is almost trivial.
Back then I also wrote a web-based 3D DLA simulation with a threejs visualization. But as so often is the fate of web-apps: This no longer runs today. Something about a used dependency (highcharts) being blocked.
Fun memories, together with the equally cherished Ising model. Seeing these patterns still puts a smile on my face. Btw they can be made to look nicer if you simulate a larger lattice and sample it down to a smaller resolution to emulate anti-aliasing.