Physics, Biochemistry, Cell, Biology, Research Neat AI does Smooth Life – the next step in Conway's Game of Life

In particular, i wanted to use the same settings described in the original paper and see if i could get the smooth live glider to make an appearance turns out. I could its all, based on this 2011 paper, a generalization of conways game of life to a continuous domain which describes the theoretical model and also the best way of implementing it. And in this setup we talk about the state of the effective cell at a point as the average of the field over a neighborhood around that point, and you move from the blocky and familiar conways setup to the continuous output you get from smoothlife. To give things a sense of scale and an organic setting, ive set the output against a backdrop of a highly magnified water environment, featuring amoebas and other lifeforms a while back. I did a video on conways game of life and used a neural network to try and find new and novel shapes and patterns. I started by just randomly populating a large matrix with cells which predictably just generated random unstructured patterns in the game. A complexity factor was taken from this paper here, which attempted to measure the degree to which an object is meaningful, with a focus on the shapes which emerge in the game of life, and i used this complexity factor as a fitness function and combined it with a Neural network, using the neat algorithm to steer a population towards more and more complex shapes which did result in some pleasing outputs, Music, but theres, something a bit special about smooth life, its very organic in nature.

The structures and creatures that emerge have a much more lifelike quality than any ive managed to produce in other simulations. In this example, here you can see a long ribbon like structure, slowly growing in size, which bears an uncanny resemblance to the yellow, real life ribbon structure. Just to its right, gliders, of course, are everywhere in smooth life. They pop up all the time. If you get the cell birth and survival rules set correctly here, you can see one making its way over to a doughnut shaped kernel Music after colliding with it theres a spate of what looks like cell division before two gliders are launched: Music one heading left, while The other heads back down again Music, another example, here of again what resembles cell division followed by the launch of a glider. It heads to the right and ducts under another structure in the environment and eventually collides with a donut shape and gets absorbed Music. It might not have been obvious with all thats going on in that clip, but if i remove the background you can see just the raw smooth life output, you can see the glider doing something a little odd for no particular reason it veers off to its left. There are no other structures around it: theres no cohesion or avoidance rules built in its just following basic smooth life rules and its this type of behavior, which makes it feel quite organic Music. My own implementation gives the standard output using the initialization parameters defined in the original paper Music theres a lot going on.

You can get these tentacle type structures which can grow and shrink over time and waft about, like you, might see similar structures doing on coral reefs. Trying to catch prey in the water Music heres, another one that just gets bigger and bigger until eventually it moves off screen and gets reabsorbed, Music Applause, Music and heres another one of those gliders that likes to give the occasional course, correction. Music. If you want to play around with it yourself, theres a nice python implementation on github ducky, the scientist has coded it up and done a little intro on the various components around continuous space and the smoothest function, and how a cell will combine with its neighbors to Form the fuzzy donut, he also mentions the transition functions and the parameters are used for cell survival Music. These are also detailed under the rules class and its worth playing around with them just to see how dependent a stable setup is on these parameters, and you can also see the logistic functions used. But the easiest way to get going with smooth life is to jump over to shader toy ill leave a link in the description, theres, no download or install required. All you have to do is press go under the buffer. A tag on the right youll find the code along with the survival parameters. Funnily enough, they are exactly the same as those in the original paper. Music.

What do you think?

Written by freotech


Leave a Reply

Your email address will not be published. Required fields are marked *



Physics, Biochemistry, Cell, Biology, Research Scientist Stories | Meet PhD student Tombi Makuyana

Fortnite, Battle pass, Carnage, Epic Games, Venom hapter 2 Season 8 Battle Pass