search

marmust / internet-scanner

see the internet as a physical network of web pages
279 stars 32 forks source link

Nodes get zoomies and orbit base node group (+some clusters unaffected?) #35

Open Inertia-Squared opened 9 months ago

Inertia-Squared commented 9 months ago

More of a funny bug here, not sure how this could be fixed without a zero-sum physics solve or some unorthodox changes to the physics, but its very fun to watch it go.

Planning to spend a couple hours today looking over all the new changes and help make improvements where I can!

https://github.com/marmust/internet-scanner/assets/26962602/4c7cc25f-a6b4-4200-879b-4229cc980e8a

marmust commented 9 months ago

tysm for trying to fix it :D

some clusters might be unaffected because theyre out of physics range (click e until it says colormode in_range)

i already tried fixing it here is all i know: 1: already encountered this on the porevious version of the physics (idk if it worked there) 2: the thing is spinning only because the root node is anchored, if it wasnt, then they would just fly in one direction possible fixes: 3: maybe can be fixed by adding a counter-force to the node currently pushing or pulling (kinda makes it unstable) (and then you need to half all the force magnitudes) 4: bug goes away when repulsion is removed (or interaction radius is made much smaller) 5: possibly make only nodes who have children be able to repel (only other nodes who have children) anyways good luck ur super helpful

Inertia-Squared commented 9 months ago

Thanks for the tips! That helps a lot actually.

I'm fairly certain that the current issue with the simulation is that it doesn't follow newton's laws of motion. If we added up all the forces being applied in each physics frame it would likely not be 0.

If we can get it to be a zero-sum simulation, that would be pretty awesome because it would also mean that we shouldn't need to anchor the first node either. I'll have a look into it!

PlasmaPig13 commented 9 months ago

I made a graph rearranger recently and it doesn't (as far as I'm aware) have some of these issues. If anyone would like to have a look and consider the system's potential applications to this project this is a link to a Desmos graph that showcases my system: https://www.desmos.com/calculator/v61ydfzdzm