/sci/ - Science & Math

how to play?

Hint: Start with a high mutation rate, once you hit a high score (say 250) decrease your mutation rate and leave it running for a 48hours.

>>2405554
the only thing you can change, is the mutation rate, thats the beauty of evolution.

after 20 children, the simulation will start the next generation

This is awesome. Thanks, OP.

wtf, i still don't get the point of this.

/v/ has a bigger thread running right now, but most of their models experience retardation

all this does is spawn some random shit and reset 1 second after

what the fuck are you supposed to do?

what does mutation rate control, randomness in new spawns?

>>2405599
i don't know exactly how it works, but it looks for the models with the highest scores and takes their traits (shapes of the parts) to the next generation.

after 5 or 6 generations, you should see a vast improvement in their score

>>2405603
This is /sci/, you guys should understand basic evolution

This was originally from reddit. Hopefully the guy tinkers with it like he claims he will.

http://www.reddit.com/r/programming/comments/f6g98/genetic_algorithm_car_physics/c1dmm1h

<span class="math"> \Sigma [\math][/spoiler]

>>2405619
>This is /sci/, you guys should understand basic selection
minor fix'd

heres another similar game:

http://www.kongregate.com/games/BryceSummer/walkinator

Rate of 20 %, they keep dying around th 200 mark for 10 generations now (best case, that is). Hmpf.

>>>/b/304446026

Autism Powered Cars

NOPE

I just had one that went to 150 and kept going, but it reset anyways.

What, did it get sick or something?

>>2405652
this is awesome
don't like the >inteligent design
part though

>>2405656

I've let this thing run for up to 40 generations and you'll find that there are sections of the map that are simply too difficult for any car. Maybe they have to go hundreds of generations before a few cars get lucky enough to get past a really difficult part of the map so that their traits can be passed down, and that section won't be so difficult anymore, but it's unlikely.

>>2405757
An automotive 'bottleneck'!

>>2405740

The cars reset when they've gone twice the distance of the previous generation.

Small front wheel and big back wheel works great with me, with a triangle shaped block in the middle.
I've only been up to 209 yet because it bounces too much at that hard part...

I had a unicycle (other wheel was being useless) run for over 100

I keep getting lots of great designs that fail because the wheels are on the top instead of the bottom.

WTF evolution?

All my good cars now have a vestigial sail thing because of one in the 3rd gen that did well. I guess what I'm trying to say is that Eugenics is a good idea.

pic related, its my best shape after 19 generations

Aww yeah, getting the good designs now.

20 %, generation 32, still crashing on the 200 spike. :c

behold

I call this one:

The unicorn

>>2405869
>aplicaciones

The fuck are you, spanish?

>cars slowly improving gen 12, half of them getting to 200, one or two getting over every gen
>me: fuck this is going too slow
>bump up mutations to 20%
>suddenly, Autism, Autism everywhere

>>2405876
No, i'm learning it. I'm french.

>>2405810

It's really interesting to see what comes along generation to generation. Many times you'll see things appearing in Generation 1 that carry on to generation 20, and simply don't have much of an effect on the effectiveness of the car, like an appendix or tail bone in a human.

this simulation is stupid. it throws out a perfectly viable design because it goes so fast as to catch air...

>>2405885
>it is an important attachment for various muscles, tendons and ligaments — which makes it necessary for physicians and patients to pay special attention to these attachments when considering surgical removal of the coccyx.[2] Additionally, it is also part of the weight-bearing tripod structure which act as a support for a sitting person. When a person sits leaning forward, the ischial tuberosities and inferior rami of the ischium take most of the weight, but as the sitting person leans backward, more weight is transferred to the coccyx.[2]

Tailbone isn't vestigial. I hate this myth.

>>2405891
nature's a bitch

>>2405891
It throws it out because it doesn't function well in its environment.

Natural selection doesn't look at something and say "This design is good, let's keep it." It destroys every design that isn't capable of surviving. If the design fucks up, it dies. Catching air is fucking up, no matter how "good" it might be otherwise.

>>2405891
>Choose to run simulation that selects for models that don't leave the ground.
>This simulation is stupid because it selects for models that don't leave the ground.

>>2405902
Bitch needs to get laid. Can't we like get brother Moon to give her a good fucking?

all my designs head towards "superbike with wheelie bar" but theres a bump at 203 that nothing seems to be able to get past

>>2405894
Same with the appendix, it functions as a reservoir for symbiotic and commensal microorganisms so that the intestines can be recolonized after the lining is flushed during diarrhea.

>>2405911
father time you mean

What does the number in the paranthesis mean?

It's tough to watch when one that you know would be a lean mean rape machine is born on its head. Such is life.

>>2405934

youd think that would be the first lesson that the program learned.. no wheels on the ground = no go

Protip: Run multiple instances of the simulation in different tabs.

So what does tinkering with mutation rate do?

>>2405943
The program isn't "learning". Does everything that reproduces have two wheels on the ground? Looks like selection pressures already favor two wheels on the ground, derpy.

>>2405949
It changes the rate of mutation. Are you dim?

>>2405891
There could be selection against species that leave the ground. Nature operates in weird ways. An animal that runs so fast that it's able to get airborne could subsequently get predated upon by birds.

woo! 230.2

>>2405934

I legitimately lol'd

+1 internets to you

Why does a child keep resetting?

This one has been unusually successful for me as of late

Yeah, it's also frustrating when a good design that you think might be able to get 210+ just happens to get flipped while one that goes slow and steady but will never pass 196 gets to breed.

You guys have to keep mutation rate low or else you'll get retards. Look at what it did in only 7 generations.

>>2406020

>You guys have to keep mutation rate low or else you'll get retards.

FUKKEN LOLD

>>2406020
You tell me how to play when you get 200+

I get the feeling like the "game" lacks of any selection algorithm killing bad cars.

>>2406047
>355.1

>>2406043
brb. Have some sister spanking to do.

>>2406051
Successful cars as judged by their score get to breed and pass along their traits including wheel size, suspension angle and geometry. Then the game randomizes some of the variables depending on mutation rate and produces the next gen. So your feeling is wrong.

>>2406051

its just like nature, it will let the retards live a short, painful life.

>>2406047
It got stuck on one hard place, I increased it to 10%. But I've seen some improvement in few cars in last generation. I think they will solve it in 3 generations.

>>2406052
That's not max score, that's the score where the game will declare the car "successful" and stop it running

Thanks for the gift, op. I shall give you a gift in return:
http://www.youtube.com/watch?v=mcAq9bmCeR0
He pretty much gives a summarization of his program that mutates watch parts to form a functional watch and includes timelines of the watch species.

Oh god, all my cars are complete derps.

The part that pisses me off the most is when you have a badass car and it gets to a fukken steep hill and dies.

>>2406081
Kinda makes you scream DAMN YOU, EVOLUTION! doesn't it?

wow
mfw alsost 400 on gen 8
havent been past 200 since

I went eating after I was around ~290.

>>2406115
That's actually the one.
I've seen her mama do 229 just before.

ALL HAIL THE DERPMOBILE

actually went 13 lol

>>2406125
Damn that one did 225.
I'll never see the 400 one.

It's very hard for it to get very far past 219.
The best that I have had so far made it to 250 or so.

Just so you know

>To keep the fitness scores for each round fairly close, there is a target score in parenthesis that is 2 times the previous rounds max score. Once any car reaches that point it wins that round and we move on.

>A car is considered stalled when its linear velocity is below a certain threshold in both the x and y direction (after a grace period at the beginning).

Guys, starfox is in this game!

All of my cars turn either into derpcars with a spike pointing in front of them, getting stuck when falling forward, or with a spike in the back, getting stuck when falling backwards.

Fucking evolution, how does it work?

Tails are an evolutionary advantage.

They keep you from doing backflips

>>2406057
Nah, Im observing this since about an hour now but it doesn't seem to clone the good models with slight alterations. It rather looks like it tries out 20 different species and let every single mutate.

That one is good until it finished on the back...
Needs a tail?

It sort of looks like a bird on wheels.
And the hind wheel is retardedly small and constantly glitches through the scenery. I hope that gets evolution'd out soon.

fffuuu I smell skynet

>>2406160
you lucky faggot
all my sons have big heads and tiny bodies.
Bitches die in every hill.

I don't get this game. All my cars do is land on their backs and get 0.

>>2406188
they got your genes
what do you expect

Many of mine are developing tails to keep them from tipping over backward. It's pretty awesome.

My tails seem to evolved away... they're mostly vestigial now, though useful for balance. Big wheels become more important around 300-400.

just like real mutation and adaptation, keep it low, and you get good results.

i've been running one at 1% mutation, and have already hit 400 by generation 17

1% is good because the good stuff stays and the garbage is decreased.

i bread 3 designs that are pretty constant

>>2406264
forgot pic

I seem to be getting alot of long nosed cars.

What's this game called?

>>2406282
World of Warcraft

>>2406282
even more so now than this pic, they are also mostly teal coloured

>>2406282
>>2406295
they're called cow catchers, and provide a very important evolutionary role for cars.

I ran it for the first time for 20 generations, and it gave me an awesome buggy like vehicle that ALWAYS went further than 100, and it seemed to be optimal for coming generations.

Then I closed it, now running another it only evolves into shitty matchbox cars that wouldn't pass the first hump

i took this after the car had already been on its back 4 times
most of them have some sort of self righting mechanism

You know those annoying ass bumps in the track that catch the bottom of your car?

Me Neither!

>>2406354
give it time, and a slow mutation rate.

i'm telling you, from what i can tell, 1% is perfect. thinking about going to 0 to see what happens.

>>2406390
self righting is good, they just dont die

This is the first thing I'll open when getting on a PC, for ever.

>>2406472
forever?
forever ever? ever ever?

>mfw a perfectly good car fails because it starts out upside down

if you want to keep the good cars from a generation, keep a low rate.
If they all fail in some part of the way, increase the rate so they can evolve and develop some shit like a tail or big wheels.
That's pretty much it

#10 : 27.7

>>2406543
which generation?

Jesus, this is retarded. I mean, I don't see what's going on. The scoring is not based on distance; it's based on time. The cars don't interact; there's no sexual reproduction, just randomizing each individual. That comes out to the same as running one individual at a time, 20 times. Boring, and stupid, and expensive, and not powerful enough to get a good car fast. I'm going to figure out the basic idea and then I'm going to make a better one.

almost all of my get into 100s. But almost all of them have low body types and get stuck on spikes. cmooooon evolve faster loooool.

>>2406579
car porn on /sci/ ?

>>2406596
>evolve faster
>mutation rate 1%

>>2406579
>scoring is not based on distance; it's based on time.
you wrong

but the rest, yeah, this shit is pretty boring. I hope you can improve it for good

if you find a good design set the mutation rate lower, if you keep getting shit designs put it high, then repeat.

>>2406606

lol i'm scared, I brought my mutation up to 4 % at gen 15 or something and it all fucked up, and I am now getting back to where I was before.

Well this was interesting. After about 2 generations everything was fish like. Several distinct styles showed up and would repeat many times:

"Longjaw"s had a longer lower lip that would weigh the front of the vehicle down, preventing it from going to far airborne. They were good initially and had a decent success up until a hill around 50-70, where the jaw would dig into the incline and stop them immediately. They still exist but are slowly getting phased out by gen 8.

"Big belly": These showed up around gen 2, and have a belly just protruding enough to scrape the ground. It keeps the vehicle stable and slow-going (far from reckless). They were initially successful for the same reason as the longjaw, but would get phased out by gen 5 because they would slow down too quickly to compete.

"Balanced": The original was actually in gen 1, and would serve as the father of most of the other "species" described above. They were mildly successful first, but would run off some bumps and get airborne, and lose. But after some additional balancing they were the best from gen 4-6.

"Bigjaw": These showed up around gen 4, but wouldn't become apparent until gen 6. One big jaw would get the highest score in gen 6, 7, and 8. The jaw serves a similar purpose as the longjaw, but the Bigjaw does not dig into inclines. This additional balance allows it to pass over a specific hill balanced-fish would fly off of.


Well, this was fun.

>>2406579
i agree to a point
i think it can't decide whether it wants to be a vis of an algorithm or a game
a better twist to the game would be that you can control the cars motion
a better algorithm would take variants on the top scoring of all of the previous cars

Why does it spontaneously make a new car even if the previous one was doing perfectly good?

>>2406659

cause your thing has % mutation set to a big number???
If you have a particular car that's doing great than just turn down the % mutation to ummm below 5%....to avoid any major change.

>>2406659

SEE THE SCORE IN PARENTHESIS? THAT'S 2X THE PREVIOUS SCORE'S HIGHEST ROUND, ANY CAR THAT BEATS THAT SCORE AUTOMATICALLY "WINS".

>>2406659
it's only 3%

>>2406685

ya it doesn't take alot to fuck up the car. I just had one that went to 220, and now gets stuck at 160. and I am at 1%

I like to think that cars which spawn upside down are just failed pregnancies or aborted fetuses. Makes me warm inside.

i have the extremely old version of this, like theres no colour...

and whenever a reasonably good functioning car
staggers and stalls for a second it resets it :(

>mfw after 50 generations, 40 is the furthest I get because whenever a promising candidate shows up, the next iteration gets a sharp keel which prevents it from moving

I think this is my computer's way of flipping me off.

1.Download swf
2.Open in flash player
3.Use memory hax to speed up program
4.???
5.profit!

it converged on a design, since gen ten its been the same- see graph

>>2406659
Because it got more than twice the distance than the previous generations best car (The number in the brackets)

Team 1 (1% M.R.): Gen 12, (193.1)
Team 2 (10% M.R.): Gen 7, (390.2)
Team 3 (20%): Gen 12, (192.5)
Team 4 (35%) Gen 14 (246)

C.: 10% M.rate is the perfect rate.
But i have to say, it also depends on the enviroment. Some paths have really tough starts.

>>2406817
>mfw nature doesn't care how far your car goes

one of my paths has the mountanous steep hill that i don't think any car can get past
what does evolution do now?

>>2406885
T1 (1%) : Gen 15 (404.1)
T2 (10%) Gen 11 (372.1)
T3 (20%) Gen 17 (365.1)
T4 (35%) Gen 25 (241.5)
So far, Team 1 went up really fast, but it's 3 gen ahead of T2, hence why it's not the best yet .
Team 2 is still the best choice, though it's slower than T1.

Something to keep in mind is that generation number does not correlate to time. Theoretically, I could be on gen 2 for 24 hours or could get to gen 50 in 24 hours, since "generations" are successful round completions. Also keep in mind that each trial is a mutation on the car that won the previous stage, not on the car from the previous trial within the current stage.

High rate of mutation = greater variance in each design = greater difference in each trial. Favors punctuated equilibrium, where success or failure are somewhat "random", but should theoretically get past stages faster.

Low rate of mutation = less variance in design = less difference in each trial. Favors smooth gradual evolution. Each trial should score close to the same, but it should take longer to complete each stage, since a low rate of mutation means it will take lopnger to adapt to new terrain (selection pressures).

>>2406931
Set the Rate to a higher number (40-60) until a good car make it. Then decrease it to 1-10 % and wait until that car spread its superiority among its childrens

>>2406863

GOT A LINK FOR THE LAZY AMONG US?

>>2406963
You're right. Look
>>2406945
All of those started at the same time. The highest rates may not be as succesfull, but they are faster.

shouldn't the number of wheels be a random variable?

>>2406982
http://www.mediafire.com/?mumpeqckcf9yt1c

get a flash player if you can't reproduce it

>>2407004

WORKS WONDERFULLY. DELICIOUS.

>>2406994
That would fuck everything up.

I've seen several designs (with the mutation on 100%) that managed to do a full backflip and continue. Also one that had tiny-ass wheels on stilt-like legs that got pretty far.

Does anyone have the link to the one where you design the car and try to make it go over a similar course?

Kinda reminds me of this.

>>2406945
T1 (1%) G 18 (282)
T2 (10%) G 16 (390.8)
T3 (20%) G 25 (220.1)
T4 (35%) G 42 (62.2)
Without administration, shit get fucked up.
Team 4 is the proof that a high rate requires more monitoring.
Team 2 is still the best.

>>2407041

http://www.kongregate.com/games/BryceSummer/walkinator

>mfw when this thread is still alive, and /v/'s died hours ago
>mfw i have no face

>>2407054
Only if you're defining "best" as "traveled farthest distance" rather than "most generations" or "most generations in a unit of time"

>>2407056
No, it's one with a car, not a stick-thing.

>>2407093
>he actually thinks most generations per unit time is something good

so as you can see for the first 16 or so generations i was stuck with some piece of crap variation of this. granted i am running with only 5% mutation. even if the design "tips over" to the front, the green/pink spike at the front usually stalls it out.

>>2407065
threads survive on /sci/ for a long time.

thinking throws most people off this board.

>>2407103
>He thinks this is a "game" with predefined goals and "winners".

>>2407104
then all of a sudden, this godsend comes along and blows everyone away. you can see the huge spike in the red graph. the green/pink part on the front of the design is still troublesome on certain individuals, but huge improvements have been made.

>>2407103
So, by your logic, an elephant, with a single generation per unit of time, is somehow better than a fruit fly, with 10^largenumber generations per unit of time?

>>2407113
then i just got this, which was amazing. it almost failed hard at the very beginning though, because it basically spawned with the front wheel pointing straight up in the air. it almost stalled out but then the car tipped over so both wheels were on the ground and then it was perfect.

>>2407118
No, what he's saying is that longer lived individuals are somehow better than shorter lived ones, which is equally stupid.

(same as)>>2407118
Actually, I would say that having more generations (in THIS simulation, at least), is more a measure of success than longest distance traveled. More generations = more doubling of distance in the same amount of time. I would say that the goal is to double distance, not to travel for an arbitrarily "long" distance.

>>2407141
generations over time in this simulation implies that the faster a generation is created, the less time each generation lasted.

suppose you could get generation 20 to average the longest distance you could travel in the time allotted.

this would be the most efficient you could get, akin to extending an individual's life span to its furthest possible extent.

in which case, generations should last longer and longer during the "refinement", and should then have some general generation time length in duration, between generations.

at that point, due to the arbitrary time constraint of this simulation, "negative" effects should eventually decrease the generation length and the efficiency of each individual.

this sim is alright, but kinda meh, to be honest.

This part fucks my cars up every single time

Whoever said to start with 100% mutation is probably a dumbass. I opened two windows, set one to 100% and one to 0. The 100% never got past 20, not even once. The 0% group is doing better than the original group I started when I first saw the thread.

>>2407187
same

>>2405697
I choked laughing when I read that.

For some reason saying something is "powered by autism" makes me lol pretty hard.

>>2407197
lower the rate, the more likely it is to retain the good things. starts out slow, but ultimately each generation has better and better average times.

I'm preety happy with my results given the generation. I played to gen 40, but gave up and started again.

144 point kills most of my cars...

I never got past to 220.

Please notice the multiple pun in the filename

>>2407233
not to be a ball buster, but 223 was your best

>mfw when one part of the course has a bunch of spikes that requires large wheels to pass, followed by a steep hill that requires small wheels for enough acceleration to make it up

It would be interesting to see whether there is an optimal mutation rate such that the carts converge the fastest on the highest possible score

>>2407241
I meant more or less. Geez /sci/...

220 is this shit:
>>2407187

Uh, can someone explain how this thing works?

What's interesting is the features that have evolved on my cars. Most of them have a front horn and a rear backbrace (I've noticed that cars with these tend to flip over backwards less)

The simulation is only running in the tab or window I have selected.
Must be a Firefox thing.

>>2407282
think its a flash thing, actually

The wheels on the bus go derp derp derp....

My genes must be retarded. Haven't seen a score above 26.
I'm on generation 13. Feels bad man.

>>2407271
read the thread

I think the horns tend to evolve because they're an easy way to adjust the center of gravity without changing the length of the car.

5% mutation rate is the way to go

First cart to beat my 158 hump, then proceeds to run into an insurmountable slope.

>>2407248
that's why there should be 3 wheels

>>2407248
>>2407233 here

There is a slope where most of my cars climb really slowly but without stopping and they resist spikes just fine. I got past the 220 mark now, I'm at 240. I seriously believe you can just keep going and going...

>generation 49732
>Porsche 911 design appears

It doesn't prove the theory of evolution (as it is currently proposed) to be true. NOT religiousfag.

>>2407282
download the flash

>>2407322
you guys should try lower mutations.

every time i do 1%, by generation 6-7, i'm already hitting 200+ for a distance. and hitting that huge 220-ish distance hill.

im getting 150-200ses now, but i still have a third of my cars doing 0-10. looking at other people's everything is much more average. what am i donig wrong?

What does the red line represent?

>>2407363
only thing you control is mutation rate.

what is your mutation rate?

Going to bed now and I'm leaving 4 sims running.
I'm breaking out the flamethrower tomorrow morning to put all of these abominations out of their misery.

>>2407372
it's a graph of how well you're car generations are doing over time, the steeper it goes up, the better you're car's doing

>>2407372
>>2407394
How about the black one? Just a weighted trendline or something?

>>2407330
I don't think so. If you treat each part of a course as a filter that allows only some range of length/wheelsize/etc to pass through and throw in an infinite amount of random filters, you're going to be filtering out everything you can toss at it.

Also,I've found that using a low mutation rate tends you lock you into your current strategy. If you hit a plateau you can't get past, then that means your current strategy probably isn't working and should consider upping the mutation rate.

>>2407372
I'm guessing the red graph represents the highest scores for each generation.

>>2407431
maybe the red is your best cars, and black all your degenerates. I don't really know

Guys, this is easy if you actually use the mutation rate as a tool.

High mutation: Shit will either go really right, or really wrong.
Low mutation: Shit will likely stay the same.

Start off with a high mutation rate, then every generation lower it by a regular increment. Once you get to around ten percent mutation, watch what happens during each generation. Do you like what happened that generation? Lower your mutation rate. Do you not like what happened that generation? Increase your mutation rate. Also, running multiple instances like a previous anon advised is smart, so you get a feel for good stuff and bad stuff.

/ramble

>>2407340
mine's getting there

>>2407454
put your cursor over the tab of the webpage, it explains what the graphs show

"colors show the crossover and mutation for each member of the population"

>>2407459
I was doing fine, but I raised my mutation rate and it all went to shit.

I say keep it 1% and continue to play. I can't believe I'm watching this shit for so long now..

>>2407454
I think red is highest, and black is average. For shiggles, I raised the mutation rate up to 20% for a few generations, then back down to 3%. You can see the results on my graph.

open this site while watching your cars fail for cheap lulz.

http://encyclopediadramatica.com/Derp

>>2407233
Here.

In 3 generations I'm much better now, most of my cars can get past the 144 mark and I'm stuck in the Himalaias at 240. But it will get past that one day. KEEP IT LOW, 1% ONLY.

>>2407549
forgot my cap

I've spent the last 4 hours on this and i needed to write a paper for english today. goddamnit /v/ and /sci/

>>2407565
Yeah. It's the fucking powder toy thing all over again.

So guys, what're your highest scores so far?

>>2407636
i'm pretty much stuck at the 350 point barrier

>>2407636
That's all dependent on the terrain. The one I'm currently running is capped at 250 because of rough terrain right before a steep slope.

Running 3 tabs of that 1%, 3% and 5%
1% max score 220.
3% max score 165.
5% max score 152.
Last time I ran both 1 and 5 and five won, we'll see how it goes.

550 here most of them developed a horn because they are a bunch of inbreds (3% mutation rate for 10 generations)

apparently upside down hydraulics are a superior gene

>>2407680
notice the backwards mounted suspension in the front, i tried to remove it, but it somehow worked for the very rough terrain. 18 showed this kind of suspension after 15 generations.

Currently failing because of a halfpipe like structure at 250 òó

>>2407708
>>2407680
hivemind

This is MUCH better than when bb blogged it two years ago:

http://boingboing.net/2008/12/08/genetic-algorithm-ev.html

We need a version of this where the environment slowly changes as well.

>>2407806
yes, also different number of wheels should be added

>>2406134

>>2407881

MOTHERFUCKING CAPTCHA

CRUISE CONTROLL FOR COOL

I saw a version of this program before but instead of cars it was moving limbs. I ran it for like 2 days and ended up with quadrupeds and bipedal kangaroo looking things. anyone know about that one

Jеw.

Is this flash based or something else? Seems very interesting so I'm gonna try to make my own game, pretty much like this, but a little varied.

>>2406134
OH GOD I LOL'd.

>>2408027
ofcourse it's flashbased. swf is a flash extension

>>2408027
>>2408086
Okay then. Does anyone know of a tutorial on making one of these games? or any basic tutorials on how to make the wheels turn and have the car move and all the physics involved? I really wanna make my own game like this

>>2408098
>genetic algorithm
>hai gays I wanna make my own :D I don't know how to code anything but it seems easy

>>2408104
It's not genetic algorithm really. That just makes it sound harder than it really is. Hereditary programming doesn't seem that hard. Anyone wanna link me or explain?

>>2408104
Besides, I know it's not easy, and I never assumed I could learn it quickly. But someone made it, so I will be able to also with enough time and studying of it

>>2408121
>>2408109
>>2408098

http://www.kongregate.com/labs

be sure to post it on here when you finish it

Anyone got more than 250? I can't get through those mountains

>>2408121
>>2408109
Goddamnit, kids these days...

>mfw I get a great design in the first generation

>>2408109
>>2408121

step 1: download flash
step 2: download eclipse/flex
step 3: complete > 9000 tutorials
step 4: ???????
step 5: profit

>>2408237
Lucky bastard. I had a mediocre design by 4th generation and it was just good enough to show barely any improvement after 7 more generations.

Its nicer to play with the mutation more than getting a good design, it doesn't take too long to surface. For example, I left it on at 2% Mutation for 16 generations. Then I increased it to 15%. After 5 or so generations, the designs deteriorated.

This must be what god does all day

I'm just going to... you know... euthanize my computer now.

>>2408501
Lol'd. We've all imagined the universe as a simulation, but never as a flash toy.

How do you speedhack this?

>>2409761

>>2406863

While I don't know much about genetic algorithms, I'm reaching the point where I think the simulation has generated terrain which cannot be surmounted. 70 generations since stability, and over 100 generations since they started reaching this point, the best I've witnessed one manage was to get a lucky bounce and have one wheel just on top of the top of the hill as it stopped.

It could also just be a case of "you can't get there from here" in which case cranking up the mutation rate to something crazy and trying to find another aptitude peak with smaller wheels with the torque to climb it.

>>2409308
Don't worry about it. It doesn't even have neurons. The ones that do have neurons though, you should save.

I still keep gene data for an old GA that simulated neurons on my flashdrive. I will revive them one day.

This fucking bump

>>2409858
Just let it keep running until it evolves intelligence and then hacks the hill away.

THAT FUCKING SLOPE

I have high hopes for this breed.

FOREVER ALONE

derpmobile

master race right here

>>2410114
>distance cropped away
Don't think so.

I'm a shark!

full powaa!

Does anyone know what the kill condition is?

It is unfortunate to see your dude going over a hill, landing on the back wheel, and in the process of falling onto the front wheel and it "dies"

LOL

>>2410166

Seems to be distance vs time. I'm not entirely sure though either. Catching too much air is a loss as well.

>>2410166
i think u dies when go to slow

this own all my cars every time >_<

>>2406579

In GenePool you've got all those things ou're asking for. It's pretty awesome.

>>2409874
>>2409874
>>2409874
>>2409874
>>2409874

THIS

Gen 13 and i can't get over that goddamn bump

can someone get the source code for this game. I really want to make it better and add sexual breeding and retard removal

dat fuckin bump

Beat this: #0 : 38.6

why the fuck do tires change randomly places and sizes ? I had wonderful tires, car got far with them, and then next gen had tires like shit. I thought good traits were passed ?

tl;dr: someone who is a crossover of /g/ & /sci/, please make a better one.

Well...

>>2410331
Gen 30 had 5 times 166

I remember when this first came out. Looked a lot shittier, and even after leaving the window open for 10 hours my cars were still pretty damn stupid and useless. Was it made any better?

>>2410331
How come everyone has got a huge ass exponential growth in development ?

I wish there was a button for

"put this car out of its misery"

>>2406134

>>2410351
I first had 100%
I have lowered it to 1% after a few steps from 50 to 40 to 20 to 10 to 5.

Gen 140 with 3% mutation.
LOWEST score is 102.7.
More than 3/4 of the scores are above 200.
Top score is 289.9.
U jelly, normalfags?

>>2410367
a pic of one of the cars

>>2410386
another one
u all so jelly that I can feel it

>>2410400
damnit, forgot pic
got dubs though - there's always something good in bad things

>>2410367
>>2410386
>98.9

>>2410422
>Gen 140 has 102.7 lowest
>pics are of gen 141
>you complain about 98.9
>mfw you are full retard

One of my cars could ride upside down.
Too bad it was the 166 boundarie that killed it...

278.8

Are there better Genetic Algorithms out there?

>>2410490
311 \o/

>>2410676
Shouldn't you just continue with homework?

>>2410682
probably, but this very entertaining

>>2410676
>Comic Sans
>Comic Sans everywhere
The level of retardedness you have just exhibited is unacceptable on /sci/. Please change your fonts to something more mature, or leave /sci/ forever.

>>2410704
and what would be "more mature"?

>left it on overnight, still haven't passed 208

I mad

>>2410676
>Comic sans as system font

WHAT THE FUCK.

>>2410713
Anything else. An awesome example is droid sans.

i got 270.7 on generation 9 with zero mutation rate

>>2410722
>look up droid sans
http://code.google.com/webfonts/family?family=Droid%20Sans
>serifs
>serifs everywhere

>>2410722

>>2410773
u just mad because it's the best font

>>2410792
u just mad because you know its a shitty font

>>2410864
u just mad because my tastes > your tastes

>>2410869
u just mad because you have nothing better to do than argue with me over the Internet on 4chan

>>2410888
u just mad because you don't have either

>>2410895
u just mad because you just mad

>>2410928
u just mad because your madder

interesting but also quite boring because it evolves very slowly

DAT HILL

Anybody else having a problem where the road turns to 2 bumps (essentially stopping your car), big cliff followed by another bump (stopping the car again) and then the steepest hill known to man (so that your car doesn't have enough power to get over it)? At around score 215?

This is hilarious.

It's like "If I have 20 cars and they fuck, and their children fuck, and so on and so forth, how long does it take for me to produce a working car?"

look at my speedo!!

Somebody please rewrite it so that the number of wheels varies from one to four and the cars run simultaneously. Look into Flash decompilers for getting the source code.

>mfw my best car in every generation shows up in gen 8 with both its wheels on the same axle

>>2412390
Not at all. Not. At. All.

/sarcasm