Home » Stars! 2.6/7 » The Academy » Colony models
| Colony models |
Tue, 16 May 2006 13:49  |
|
 PricklyPea |
 | Lieutenant |
Messages: 534
Registered: February 2005 | |
|
Does anybody have a colony model which supports terraforming? It's sometimes hard to figure out whether it would be better to build factories or terraforming.
With the race wizard points worked out, and planet value public, a model could be built. Has anybody done it?
Report message to a moderator
|
|
|
|
| Re: Colony models |
Tue, 16 May 2006 14:51  |
|
| Kotk | 
 | Commander |
Messages: 1227
Registered: May 2003 | |
|
What model? You mean spreadsheet? Or some program?
I havent found a way to predict what field the game will actually terraform if multiple available unless its yellow.
[Updated on: Tue, 16 May 2006 14:52] Report message to a moderator
|
|
|
|
| Re: Colony models |
Tue, 16 May 2006 15:24 |
|
| PricklyPea |
| Lieutenant |
Messages: 534
Registered: February 2005 | |
|
yes a spreadsheet or program. at the moment i use posey's spreadsheet for colonies, but it doesn't have the ability to set production queues for terraform.
it would be nice to get something that could try different scenarios and produce resource, mining and population curves so that you can compare them and pick the best curve for the particular situation.
Edit: how does it choose to terraform when it's green and does it definitely not use the same algorithm for yellow?
[Updated on: Tue, 16 May 2006 15:24] Report message to a moderator
|
|
|
|
| Re: Colony models |
Tue, 16 May 2006 15:27 |
|
| Kotk |
| Commander |
Messages: 1227
Registered: May 2003 | |
|
Yellows it terraforms always toward green first.
The problem is dont know what field it picks once its green.
Seems to be most away from ideal ... but not always.
[Updated on: Tue, 16 May 2006 15:30] Report message to a moderator
|
|
|
| |
| Re: Colony models |
Tue, 16 May 2006 17:14 |
|
|
|
| Quote: |
Seems to be most away from ideal ... but not always
|
I believe it is what gives the biggest improvement in hab. The farthest from ideal is often the best, but shift in a narrow field may do more good, especially if wide field isn't at very edge.
I personally use a rough guesswork approach, looking at what percent production gain I get from factories compared to gain in pop growth rate (longer term), and looking at how my pop levels are doing (lots surplus compared to factories or tight).
Pop over 25% level barely grows, so terraforming is sometimes as much about increasing max colonists as growth rate.
A planet at near pop capacity and where you won't be adding more pop won't be helped much by small amount of terraforming (grows slow no matter what) so often best to focus on factories till you can do a good terraforming push.
A planet with high needed mineral concentrations is more likely to get factories rather than terraforming.
HG race will tend to terraform at lower planet resource levels than HP race.
[Updated on: Tue, 16 May 2006 17:19] Report message to a moderator
|
|
|
| | |
| Re: Colony models |
Tue, 06 June 2006 09:04 |
|
|
|
Hi!
| Kotk wrote on Tue, 16 May 2006 22:27 |
Yellows it terraforms always toward green first.
The problem is dont know what field it picks once its green.
Seems to be most away from ideal ... but not always.
|
By my observations, it is the field that gives the most improvement hab %, and calculated with no accounting of decimal places in improvement% and no accounting that fuinal hab% is a multiply of particular hab%s. In one of my current games I often get a wide field teraformed instead of preferred narrow. Both 1 point teraforming estimates 1% of improvement, and wide field is gravity (first), so gravity is teraformed. However, if teraform gravity by 3 terformings, it gives 2% of improvement only, while narrow field teraforming by 3 would give at least 4%, sometimes more. But each particular terforming "click" is estimated searately, and estimated as 1% for both fields and all 3 teraformings, so all 3 go into gravoty instead of narrow field. So that is where confusion begins.
What I would like to know is exact formula of hab% calculation. I.e., if you have given hab range margins, and current hab value, what is hab% for that field? Even if you do not know exact way how teraforming field is picked up, such calculator would help a lot anyway in general.
WBR, Vlad
Report message to a moderator
|
|
|
| |
| Re: Colony models |
Tue, 06 June 2006 11:39 |
|
|
|
Thanks for the link. I'll try to create a program to calculate optimal production for planet economy (mines/factories/teraforming and pop growth for the best final result).
Q: Is it good idea to also account "external" things for that planet, such as mineral and pop transport? Also, it probably would be good if program tells when Germanium would be needed to ship. Looks like a simple AI for a single planet economy.
WBR, Vlad
Report message to a moderator
|
|
|
| | | | |
| Re: Colony models |
Thu, 20 July 2006 15:59 |
|
|
|
You get a fixed percentage increase by focusing on factories/mines. So for example a certain planet may gain 20% more factories if focus on them, if it has 10 factories and max pop for them, that will lead to 12 at cost of 20 resources, if it has 1000 factories will lead to 1200 at cost of 2000 resources.
20 resources would pay for 1/5 a terraform, but 2000 resources would pay for 20 terraformings. Each terraform gives a percentage increase.
An HP eccon tends to want lots of factories, then burst of terraforming before back to factories.
An HG eccon sometimes will want same as HP but sometimes will instead want terraforming first to make room for more pop workers to be dropped on to planet.
Report message to a moderator
|
|
|
| |
| Re: Colony models |
Fri, 21 July 2006 07:49 |
|
|
|
| m.a@stars wrote on Fri, 21 July 2006 12:22 |
| Tomasoid wrote on Thu, 20 July 2006 19:07 |
There are a lot of nuances like "Need build more factories/mines for year if we build terra as well - because pop would grow more and would be able to handle more - pop growing goes after production. Because we build more factories, we now cannot build enough tera for what factories/mines we bilt - we spent more resources for factories/mines." 
I solved this one riddle, but I feel more comes...
|
In my own experience with this kind of thing, I've usually found that the "absolute best" is rarely needed.  You can go with "good enough" and see little (if any) practical difference. And it should be far easier to code/run.  We're talking about hundreds of planets here, so execution time might get significant. 
|
I'll try to do it (have enough skill for that). If it would run too slow, I'll add an option to switch off optimizing to the smallest detail and leave only quick optimization.
What I'm more concerned now is how you would apply results of such programs for hundreds of planets?  There is no way to update X file from outside of Stars! as far as I know. The only way to automate this that comes to mind is some hack into keyboard and simulate keys pressing for instanse of Stars! application to apply all commands. But there would be issues with turns that contain ships building for you.
| Quote: |
Still, I'd love to see how you solved it. 
|
The algorithm is simple though - it balances between fact/mine building and terra building up until we get difference in 1 between 2 cases for terra building (or up until it repeats, though terra cost is quite high, so this case would be quite rare). Then I always choose to do more teraforming rather than build (unneeded) factories/mines. Then reserve resources for teraforming - all except partial building if it happend for teraforming. Then recalculate factories/mines up to available resources except reserved for terraforming. Algorithm is quick - it would not ever do more steps than teraforming we are going to do for planet next year initially.
BTW, it is something similar to what is made already in FreeStars for building with auto-alchemy option on.
I can send sources, but do you know Pascal? I'm wirting main algorithm in old Turbo Pascal on the old machine which I have access to at evenings at home when have free time (No way to install any other programming language there...)
BTW, thanks again for the hab% calculation algorithm! It is working great. You mentioned in comments, however, that there are some issues that are overcomed in Stars! with the center of hab range slightly shifted because even number of clicks in hab range. Do you have details of that, or they are not needed/not important?
Also, is it possible to get somewhere exact algorithm of selection what hab field is teraformed next year for 1 click? Is the algorithm in FreeStars sources (BestHabForTerraform function) correct?
...
WBR, Vlad
Report message to a moderator
|
|
|
| |
| Re: Colony models |
Wed, 26 July 2006 16:57 |
|
| yartrebo |
 | Petty Officer 3rd Class |
Messages: 43
Registered: July 2006
Location: North America | |
|
Assuming you're trying to maximize resource growth and you have a decent supply of germanium, compare how fast 100 resources of factories can add increase your resources by building new factories to twice how fast a tick of terraforming will increase your resource growth (by adding population). In the comparison, you double the resource growth from population because not only does population grow exponentially, but you get to use the extra resources the extra pop gives you.
If you have 100,000 people on a 50% hab world, a 20% PGR, 1 resource/1000 pop, and terraforming nets you 1% hab per tick, you growth will increase from 10,000 people/turn to 10,200 people/turn - netting an measly 0.2 resources per turn squared.
With 15/8/21 factory settings (HP), 100 resources in factories would produce 12.5 factories making 18.75 resources. Those 18 resources could build 2.25 factories/turn, which will make an extra 3.375 resources per turn squared.
3.375 is greater than twice 0.2, so the factories will give you more resources in the long run than terraforming. Of course, at some point the balance will shift in favor of terraforming, or you will start getting close to your factory limit, but the factories will let you terraform much faster and the planet will be built out much sooner.
As a general rule, the following characteristics of a race and planet encourage terraforming over factory building:
- high PGR
- high max planet capacity (ie., JOAT + OBRM).
- high population on planet (doubly so if over 25% max).
- poor factory settings (factory eff / factory cost mostly, but number operated will sometimes factor in).
- tight habitability bands.
- having total terraforming.
- many factories already built (to avoid hitting the limit on factories operated).
- germanium is in shortage
For low hab or yellow planets you might want to set the following autobuild queue:
Autobuild Factories up to 1020
Autobuild Max Terraform up to 1020%
The reasoning is that for low pop planets, factories are a much better investment than terraforming, which will take an anemic growth rate (low population limit compounded with low growth %age) and make it a still anemic growth rate. Only invest in terraforming if unable to build factories, probably due to insufficient population. A better idea is to import the needed population up to 100% capacity.
PS: If you will reach your limit on factories operated soon, you should terraform, as there is a major lag between terraforming and getting more population.
Report message to a moderator
|
|
|
|
| Re: Colony models |
Wed, 26 July 2006 20:22 |
|
| m.a@stars |
| Commander |
Messages: 2765
Registered: October 2004
Location: Third star to the left | |
|
| Tomasoid wrote on Fri, 21 July 2006 13:49 |
What I'm more concerned now is how you would apply results of such programs for hundreds of planets? There is no way to update X file from outside of Stars! as far as I know. The only way to automate this that comes to mind is some hack into keyboard and simulate keys pressing for instanse of Stars! application to apply all commands. But there would be issues with turns that contain ships building for you.
|
Heh, true, but that might change in the future, with either new x-file-tweaking tools or new versions of Stars (i.e Freestars)
| Quote: |
The algorithm is simple though - it balances between fact/mine building and terra building up until we get difference in 1 between 2 cases for terra building (or up until it repeats, though terra cost is quite high, so this case would be quite rare). Then I always choose to do more teraforming rather than build (unneeded) factories/mines. Then reserve resources for teraforming - all except partial building if it happend for teraforming. Then recalculate factories/mines up to available resources except reserved for terraforming. Algorithm is quick - it would not ever do more steps than teraforming we are going to do for planet next year initially.
|
Sounds good.
| Quote: |
BTW, it is something similar to what is made already in FreeStars for building with auto-alchemy option on.
I can send sources, but do you know Pascal? I'm wirting main algorithm in old Turbo Pascal on the old machine which I have access to at evenings at home when have free time (No way to install any other programming language there...)
|
Oh my, good ole TP lives on! By all means, send these precious sources. They will make a few old 'n hardened synapses happy! 
| Quote: |
BTW, thanks again for the hab% calculation algorithm! It is working great.
|
I only did the easy part (rewriting) You can thank the Jeffs for the original and that dogged russian dissassembler for the initial disclosure. 
| Quote: |
You mentioned in comments, however, that there are some issues that are overcomed in Stars! with the center of hab range slightly shifted because even number of clicks in hab range. Do you have details of that, or they are not needed/not important?
|
IIRC, current racewizard forces you to design hab ranges so that the center hab is exactly the same # of clicks away from the max and min. What comments I wrote were reflecting the fact of my taking that into account for rewriting/simplifying a significant portion of the algorithm, since the "original" version seemed geared to cope with non-range-centered ideal habs. 
| Quote: |
Also, is it possible to get somewhere exact algorithm of selection what hab field is teraformed next year for 1 click? Is the algorithm in FreeStars sources (BestHabForTerraform function) correct?
|
IIRC, Leit's explanation is as close to what Stars! actually does as anyone can guess.
...
So many Stars, so few Missiles!
In space no one can hear you scheme! Report message to a moderator
|
|
|
|
| Re: Colony models |
Thu, 27 July 2006 05:28 |
|
|
|
Hi!
| yartrebo wrote on Wed, 26 July 2006 23:57 |
Assuming you're trying to maximize resource growth and you have a decent supply of germanium...
|
Not realistic. I'm writing a tool for realistic cases when you usually have little of germanium.
| Quote: |
...In the comparison, you double the resource growth from population because not only does population grow exponentially, but you get to use the extra resources the extra pop gives you...
...If you have 100,000 people on a 50% hab world, a 20% PGR,...
|
As you pointed alerady, it depends on the player settings (grows vs factories effectivity etc.). However, the tricky part here is that it also depends on the planet habitability. If you read this thread again more accurately, we were talking mostly about yellow and poor hab planets where it is really hard to judge what is better to build. Balance between the teraforming and factories building here is always really slip.
| Quote: |
1 resource/1000 pop, and terraforming nets you 1% hab per tick, you growth will increase from 10,000 people/turn to 10,200 people/turn - netting an measly 0.2 resources per turn squared.
|
Well, in practice, if you have good factories effectiveness, factories usually give more. So building mines instead of terraforming would give you germanium for factories. Usually it ends up better than teraforming, but it depends on many things. This part is easy to calculate for each particular case manually, so I'll leave this issue out.
The moment when we switch from factories building to mines is a slippery one. Let's see on a 100% planet. Assume next year you get really short of germanium (we assume none is left) and you have no mines (or 10 initial mines as on HW). Next year you have no other choices but build mines. However, this way you would spend most of your resources on the mines building, include resources you get from just grown up pop. Tricky thing here is that if you build half factories and half mines at the previous year, next year you would be able to build more factories because germanium is mined more earlier. Sometimes, in case mine cost is expensive and factory requires 4 germ to build, you may end up in few next years with more factories and resources. It means that dropping off usual "max factiores then mines" rule at the previous year to germ shortage year might be better. An exact balance how much you should reserve for mines building at the previous year is not known for me yet, though I hope my program would be able to figure that out.
Now look: above is ONLY about balance between factories and mines building. Add here teraforming, and you have a 3-parameter balancing (3-dimentional), that is much more complex for hand calculations. That is what I'm writing a program for: not just contain a set of commonly known tactics and strategies, but find out the _best_ or close to the best tactic from all possible. I know this is NP-hard problem, but I hope I would be able to use some euristic algorithm together with functional optimization that would work in acceptable time - as I already said, I have enough skill for that.
| Quote: |
For low hab or yellow planets you might want to set the following autobuild queue:
Autobuild Factories up to 1020
Autobuild Max Terraform up to 1020%
The reasoning is that for low pop planets, factories are a much better investment than terraforming, which will take an anemic growth rate (low population limit compounded with low growth %age) and make it a still anemic growth rate. Only invest in terraforming if unable to build factories, probably due to insufficient population. A better idea is to import the needed population up to 100% capacity.
|
Well, what about -1% yellow? -2% yellow? -3%? -1% probably better to teraform first for obvious reasons. At what -XX% we should stop initial teraforming for reaching green and build factories as you suggested? There is a balance point in here which is different for different race settings _and_ different for different planets. This is quite hard to calculate. It depends on planet be
...
[Updated on: Thu, 27 July 2006 05:30]
WBR, Vlad
Report message to a moderator
|
|
|
| |
| Re: Colony models |
Fri, 28 July 2006 00:47 |
|
| JasonC |
 | Petty Officer 2nd Class
Stars! V.I.P |
Messages: 58
Registered: July 2006
Location: Arlington, MA | |
|
Well isn't this a fun problem?
I worked on this stuff a long time ago. I'll give you the rules of thumb I came up with and then get in to some of the provisos.
My standard Q was factories mines terra 5%. But as soon as a planet hit 500 resources I bumped the terra to the top of the q and just banged it all out in a couple of years.
This saves most of the terra for when the planet is big. The large terra build also cuts off the resources to factories and makes G pile up on the surface. Then I put in a base and the place is ready as an exporter-breeder.
A planet with low hab won't hit 500 resources before factories catch the pop and the autobuild turns green. When that happens it will terraform by default. The result is low habs do factories first.
Now, I changed this SOP for narrow immune hab HGs. They terra'ed at the 100 resource line by putting 1% max terra at the top of the Q as soon as they crossed 100 resources. That slows factory growth but it speeds their transition to breeder status.
For good yellows, -1 and -2 or so, HGs terra to green as soon as they land, HPs max the factories first. Deep yellows always do factories first.
Those were my rules of thumb. Why?
Well, in the long run the factories are going to catch the pop, not just on one planet but across the entire empire. At that point, economy will equal max resources per pop times total pop (that isn't in the air). Faster factories up does not raise the total reached at this point, it raises resources available for tech and ships etc.
So the real determiner of total econ by a given year is the race maximum resources per pop, the PGR, and the *achieved average hab*. The main key to a huge econ after the opening, in fact, is higher achieved pop growth through higher achieved average hab. And after the race design is set, managing the economy is mostly about getting that hab.
The real limit on hab achieved, however, is rarely the habitat rating of the minor worlds. Instead it is the onset of crowding. Crowding cuts off the PGR increase on the "breeders". But properly managed it leaves a line - a sustainable, linear rate of pop growth.
This is achieved by pop lifting - the breeders "hold" at 50% of capacity and send all their new pop to the lousy worlds, packing them to the brim. The growth achieved by all the breeders held at 50 and fully terra'ed, and all the minors crowded to the brim, vastly exceeds the growth achieved by letting everyone go up.
You might think it is better to hold at 33% of capacity because that is max PGR. But that is suboptimal. It gives up a third of the income of the best worlds, and will result in a much lower resource integral. The drop off in PGR going from 33 to 50 is tiny. You might think the pop would do as well on another world, but in practice this is not true. Besides lost travel time, the poor worlds aren't as developed as a "held" breeder, nor do their resources unlock for shipbuilding as easily, etc.
The maximum absolute PGR would be achieved holding at 25 as long as there is equal or better hab anywhere under the crowding limit - but that is way too low in practice, and alternate hab usually pretty poor when travel time is taken into account etc.
Now, how does this fit the 5% all of it rule? A soon to be breeder will experience crowding around the time it hits 500 resources. HGs get 2 to 3 resources per pop max and their factories typically haven't fully caught the pop yet. They will be around 200-250k therefore. Any breeder world will eventually terraform to around 80% hab. The world ready for the 5% terra line is therefore crossing the crowded threshold even after its done. HPs can have more like 4, but are slower to catch by 5 or 10 years. Similar effect - when a good world is ready to bang it all out, it is also in a position to transition to a breeder quality hab, and it will experience near maximum absolute pop growth right afterward.
The narrow HGs have to go sooner than this, though. For them, every initial
...
Report message to a moderator
|
|
|
| |
Goto Forum:
Current Time: Fri May 03 15:19:01 EDT 2024
|
Members
Search
Help
Register
Login
Home
stars has published closest algorithm so far:
) Also, I'm re-writing it a bit at the moment to split to pieces and optimize more to decrease calculations time, as well as read data from config file rather than from hard-coded constants. I think I'll finish this in a week or so. Most calculation things I already have - thanks to FreeStars sources!
If you are really *the* Jason Cawley welcome to the forum and welcome back to Stars! 
