0.9.0.Ex (Experimental)

It's work complete, the playtesting forum rides off into the sunset.
Sunconure11
Posts: 16
Joined: Sun Oct 27, 2013 6:51 pm

Re: 0.9.0.Ex (Experimental)

Post by Sunconure11 » Sat Mar 26, 2016 9:53 pm

If I modify the world settings, and then load up the save again, I get this error thrown at me when I try to load it up:

http://pastebin.com/gG2iu06U

Skylimit
Posts: 361
Joined: Tue Jul 10, 2012 5:57 pm

Re: 0.9.0.Ex (Experimental)

Post by Skylimit » Sat Mar 26, 2016 11:46 pm

Some observations I made from long running simulations
1) meat seems to have not enough weight
2) sexual reproduction doesn't favour carnivores either

The adjustments to make are pretty extreme - I have had meat value up to 7.5 at one point, when carnivores were a bit too successful and driving everything to extinction

Also strange in long running simulations is that with 1-1-1 value, at one point the population goes to cap, at which point it decreases the weight
But then when adjusting the values after the simulation was started, you can safely put them back to 2.5-2.5-5.0
They never seem to reach cap again. In fact, with these values neither cap is reached nor is carnivores overpowered
It could be that once you have been at cap, that nature simply doesn't recover quickly enough, thereby autocontrolling the environment, and even if efficiency is more than doubled it still autocontrols the environment

Now to find the exact balance where carnivores and herbivores have equal chances, I'm unsure at this point
I had mainly ominvores 0 with the values 0.89-0.89-2.83
But that was after the simulation had already ran for a very long time, and unsure what the long term effect would have been when starting off with theses values

We have auto population control, keeping us from hitting cap, we have nature that auto-controls population, we have carnivores that drive population down, and then also go extinct when not carefully balanced, so finding the optimal values for long term simulations is not so trivial - I suspect they need to be adjusted after you reached cap and they were auto-adjusted, and also the startup values and late game values probably differ

edit : Or it could be (just a vague suspicion) that the startup values, the auto-adjusted values, and the values set after the sim started, somehow have different effect - the direction of the changes is ok, it's just the magnitude that seems off when setting values after startup - OR that food is abundant at startup, and then when eaten away never reaches that same abundance again - could be a logical explanation

Skylimit
Posts: 361
Joined: Tue Jul 10, 2012 5:57 pm

Re: 0.9.0.Ex (Experimental)

Post by Skylimit » Sun Mar 27, 2016 10:43 pm

I think I found a bug with the auto-adjust values for preventing cap
Those work well as long as you leave them at default values

Meat is undervalued, so it seems, so I start off with a 3:1 ratio meat:veggie
Now the auto-adjust leaves the 3 untouched, and then typically in the 1.5 worldsize with cap 2000 it would adjust the efficiency to 0.35

So, this resulted in interesting dynamics, however with meat overpowered a tiny bit
Herbivores -0.4 were very common, as were the 0.7 and 1 carnivores
But I was actually looking at 10:1 ratio or 3:0.35 ratio

So it seems, if you touch default values the ratio is not kept

So I then manually adjust back to 0.5/0.5/2 or roughly a 4:1 ratio meat: veggie
Which again gives interesting dynamics

So for now still looking to find the optimal ratio that would result in both -1 herbi and 1 carni in the same simulation

Note : both sexual and asexual reproduction must be allowed when trying to reproduce my results

Skylimit
Posts: 361
Joined: Tue Jul 10, 2012 5:57 pm

Re: 0.9.0.Ex (Experimental)

Post by Skylimit » Mon Mar 28, 2016 12:57 pm

I think the correct ratio, allowing for carnivores to specialize is around 3:1
Needs some more testing but I would assume a 2:1 ratio would result in average omnivore 0 and would never see specialization in either direction
Whilst a 1:1 ratio would see -0.9 herbivores, but even at 1:1 they are extremely rare
Only spotted them once

Also apparent are that carnivores are needed so that all species develop eyes and higher attack damage or speed

What is very strange in carnivore 1 is that they seem to be able to sustain in desolate places, and almost don't consume food (or not at all?) and yet their groups do grow steadily from 30 to 50 to 70 or more, at which point they sometimes start a raid
Looks like some kind of leak (3:1 ratio)
Does a carnivore 1 eat vegetation? Or is there a leak when eathing babies or something

So ratio's, more about them later :o)

What also strikes me - the carnivores maintain groups of around 50 or more
Can hang around for ever

Yet when they mingle, they eradicate entire populations, and then go extinct themselves
Sometimes it even almost leads to mass extinction (but never quite)

User avatar
Alexcat989
Posts: 2488
Joined: Wed Mar 11, 2015 2:35 am
Location: In a fictional world with a bunch of cute kitty people
Contact:

Re: 0.9.0.Ex (Experimental)

Post by Alexcat989 » Tue Mar 29, 2016 4:43 am

i love how your all testing the game without pay
SCREW YOU I AM THE ARCH-EMPOROR
Staff of the Official Discord
Image
Uses Cate/Cates pronouns
Image

User avatar
Quasar
Site Admin
Posts: 1731
Joined: Tue May 29, 2012 2:04 am

Re: 0.9.0.Ex (Experimental)

Post by Quasar » Tue Mar 29, 2016 11:27 am

Skylimit wrote:Load game crashes on me (even before I can make a selection)
Looks like it's the comma-as-decimal-point crash again. Will fix for the public release.
Skylimit wrote:So, this resulted in interesting dynamics, however with meat overpowered a tiny bit
Herbivores -0.4 were very common, as were the 0.7 and 1 carnivores
But I was actually looking at 10:1 ratio or 3:0.35 ratio

So it seems, if you touch default values the ratio is not kept
Yeah, it modifies all three values via simple addition, so to reduce meat from 3 to 2 would mean reducing vegetation from 1 to 0.

Will fix it to use multiplication for the public release.
Looks like some kind of leak (3:1 ratio)
Does a carnivore 1 eat vegetation? Or is there a leak when eathing babies or something
Good call. Values above 1 meat efficiency can theoretically cause an energy leak, since they mean it's possible for a +1 carnivore to get more energy from a corpse than the dead creature had in life. They would have to be very specialised to achieve this, and the automatic efficiency adjustment would probably stop them from abusing it to any great degree, but it's possible.

User avatar
Therminator
Posts: 220
Joined: Wed Apr 15, 2015 12:03 pm

Re: 0.9.0.Ex (Experimental)

Post by Therminator » Wed Mar 30, 2016 9:35 am

I'm not playtesting, but still following conversation, and I had to chime in on something.

Am I the only one who is conceptually disturbed about the idea of capping populations and blocking population explosions by altering what is for all intents and purposes the laws of physics in this simulation?
It's like simulating a bunch of animals, but noticing birds start to evolve, which give simulation problems, so you set gravity higher so they go extinct, and have gravity return to normal when they go. Inevitably getting birds again later. It strikes me as cheating, and as the ultimate interference in what is supposed to be a hands off simulator. You're basically punishing succesful creatures that play by the evolutionary rules. The ecologist and physicist in me is sad because of this :(
You basically created a negative feedback loop between biology and the laws of physics. This looks to me like a coding trick to optimize your simulation, rather than solving the actual problem of biological instability.

Because I am of the persuasion that you shouldn't just criticize, but also offer solutions and constructive criticism, here's my 2 cents.
The problem, in essense, is a carrying capacity that is too high. The total amount of energy in all plants is so high, that it can produce more creatures than the computer can handle.
If I have this correctly, carbon and energy now come from infinite stores (atmosphere and sun) into plant growth, into animals. Meaning a super high carrying capacity. Added to that, a lot of food is accessible to most creatures, with not much niche specialization. This is an invitation for viral behaviour with population explosions that directly influences a large part of the map.

My solution is this. Give them a better metabolism with at least one variable that is not infinite and that recycles. In my metabolism model, that would be nitrogen. Nitrogen is then the limiting factor to population sizes. So, if you take the maximal amount of creatures you want on a map, and you take the average amount of nitrogen contained in an animal, you can calculate the amount of nitrogen that can go into a single map. Increasing or lowering that value changes global fertility, which translates to carrying capacity, capping the population at that amount, not by failure to breed or inducing disease or less efficient eating, but just resource depletion.
That way, there is selection on becoming more efficient, which means selection on niche diversification, because with efficiency should come specialization.
If the various food sources require more specialization, this will have two effects: more clades living side by side without one taking over the other, and smaller population sizes because the large carrying capacity is broken into several even smaller ones. This, I believe, should cause more stability and more realism.

Natural_20
Posts: 1658
Joined: Mon Jul 14, 2014 5:25 am

Re: 0.9.0.Ex (Experimental)

Post by Natural_20 » Wed Mar 30, 2016 10:12 am

Therminator wrote:I'm not playtesting, but still following conversation, and I had to chime in on something.

Am I the only one who is conceptually disturbed about the idea of capping populations and blocking population explosions by altering what is for all intents and purposes the laws of physics in this simulation?
It's like simulating a bunch of animals, but noticing birds start to evolve, which give simulation problems, so you set gravity higher so they go extinct, and have gravity return to normal when they go. Inevitably getting birds again later. It strikes me as cheating, and as the ultimate interference in what is supposed to be a hands off simulator. You're basically punishing succesful creatures that play by the evolutionary rules. The ecologist and physicist in me is sad because of this :(
You basically created a negative feedback loop between biology and the laws of physics. This looks to me like a coding trick to optimize your simulation, rather than solving the actual problem of biological instability.

Because I am of the persuasion that you shouldn't just criticize, but also offer solutions and constructive criticism, here's my 2 cents.
The problem, in essense, is a carrying capacity that is too high. The total amount of energy in all plants is so high, that it can produce more creatures than the computer can handle.
If I have this correctly, carbon and energy now come from infinite stores (atmosphere and sun) into plant growth, into animals. Meaning a super high carrying capacity. Added to that, a lot of food is accessible to most creatures, with not much niche specialization. This is an invitation for viral behaviour with population explosions that directly influences a large part of the map.

My solution is this. Give them a better metabolism with at least one variable that is not infinite and that recycles. In my metabolism model, that would be nitrogen. Nitrogen is then the limiting factor to population sizes. So, if you take the maximal amount of creatures you want on a map, and you take the average amount of nitrogen contained in an animal, you can calculate the amount of nitrogen that can go into a single map. Increasing or lowering that value changes global fertility, which translates to carrying capacity, capping the population at that amount, not by failure to breed or inducing disease or less efficient eating, but just resource depletion.
That way, there is selection on becoming more efficient, which means selection on niche diversification, because with efficiency should come specialization.
If the various food sources require more specialization, this will have two effects: more clades living side by side without one taking over the other, and smaller population sizes because the large carrying capacity is broken into several even smaller ones. This, I believe, should cause more stability and more realism.

I agree that having infinite energy is a problem, and that tampering with food efficiency is kinda cheesy. There is room for improvement there.
Quasar wrote:Dammit, we forgot to sell the psycho drugs and now the chickens are ODing.

Skylimit
Posts: 361
Joined: Tue Jul 10, 2012 5:57 pm

Re: 0.9.0.Ex (Experimental)

Post by Skylimit » Wed Mar 30, 2016 4:45 pm

It is a matter of optimizing cpu resources
So you can have the maximum map size, and keep population somewhere between 200 and 1100
(as of 1500 my computer slows down)
The simulation runs best with mapsize 1.5, and creatures around 700/800

The problem is, they occasionally would boost, which either crashes the pc, or takes like 45 min to cool down

Now if you carefully tweak the startup conditions, the creatures never reach cap
Also, if you want to avoid the auto-adjust, set your cap at 2250 -> so it never actually adjusts

Or simply turn it off :D
Cause you do have this option

It is not cheating, it is optimizing resources

You can also set smaller mapsize, or lower efficiency to start with

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest