Sidetracked for a little bit because I was curious what would happen with reaction-diffusion if I manipulate the probability of a reaction happening. It could represent whether reaction conditions are favourable or the presence of a catalyst. The catalyst could diffuse as well.
I guess if the probability of a reaction is lower the kill rate will need to be lower.
Maybe it won’t be too different. But I’m also curious about introducing some randomness. What if the probability of a reaction occurring changes pixel to pixel and moment to moment?
What other parameters can I tweak in that equation?
newalpha = constrain(alpha + (case.diffuse_a * laplace2d(alpha) - p_reaction * alpha * beta * beta + case.feed_rate * (1 - alpha)) * case.time_step, 0, 1)
I wonder how using probabilistic values for feed and kill rate would affect the simulation. Or what if diffusion was uneven? It could go faster in one direction or change with currents.
Lol, as if this doesn’t already run slow enough on my computer. Now I want to run a reaction-diffusion simulation on top of a fluid simulation?
Probably a lot of stuff that’s already been done, too.
Anyway, here’s what it looks like varying the feed rate across and varying the probability of a reaction down.
All the interesting stuff happens when the probability of a reaction is high, and it’s most interesting when it’s at 1.0 like before.
I should’ve varied the kill rate instead of the feed rate. Recalculating…
Oh weird, I ended up with a bunch of bubbles. Ohhh, the kill rate is too low to clear out Chemical B when the feed rate and reaction probability are taken into account. I guess I would’ve seen that without the reaction probability.
Another idea: what if the feed/kill rates depend on how much of the chemical is already there.
Looking at the equations again:
newalpha = constrain(alpha + (case.diffuse_a * laplace2d(alpha) - p_reaction * alpha * beta * beta + case.feed_rate * (1 - alpha)) * case.time_step, 0, 1)
newbeta = constrain(beta + (case.diffuse_b * laplace2d(beta) + p_reaction * alpha * beta * beta - (case.kill_rate + case.feed_rate) * beta) * case.time_step, 0, 1)
The feed rate already does scale with the amount that’s already present.
The kill rate… oh, the feed rate is added to ensure it’s always higher than the kill rate. But my probability parameter messed that up.