Minerals: An experimental approach to saturation. - Page 2
| Forum Index > SC2 General |
|
Pandain
United States12989 Posts
| ||
|
LordOFswe
Sweden27 Posts
On June 04 2010 06:54 Piousflea wrote: What matters to a real game is the incremental income per worker. If the first probe harvests 40 minerals/min, how many do you get from probe #17, 18, etc? On June 03 2010 18:58 LordOFswe wrote: at 1 probe /patch it will harvest 42 mineral/min and probe (335 minerals total/min) 100% of M.cap.* at 2 probe /patch it will harvest 40 mineral/min and probe (639 minerals total/min) 95% of M.cap.* at 3 probe /patch it will harvest 33 mineral/min and probe (791 minerals total/min) 79% of M.cap.* at 4 probe /patch it will harvest 25 mineral/min and probe (811 minerals total/min) 60% of M.cap.* at 5 probe /patch it will harvest 21 mineral/min and probe (835 minerals total/min) 50% of M.cap.* at 6 probe /patch it will harvest 18 mineral/min and probe (839 minerals total/min) 43% of M.cap.* *M.cap. = % of maximum capacity Assuming 8 mineral patches you can calculate the difference: With 16 probes the income is 640/min (8patches with 2 probes on each= 8x80x2) With 17 drones you will have: 7 patches with 2 drones (7x40x2=560) 1 patch with 3 drones (1x33x3=99) Gives a total income of: 659 (+19minerals for the 17th drone) and so on... (did a program for the calculation out here it is) + Show Spoiler + So with 16 probes the income is 640/min 17 probes the income is 658/min (+18minerals/min)* 18 probes the income is 677/min(+19minerals/min) 19 probes the income is 696/min(+19minerals/min) ... 24 probes the income is 791/min(+19minerals/min) 25 probes the income is 794/min(+3minerals/min) 26 probes the income is 796/min(+2minerals/min) 27 probes the income is 799/min(+3minerals/min) 28 probes the income is 801/min(+2minerals/min) and so on (the +2/+3 is due to rounded numbers in my calculator actually value is more likely +2.5) *the difference here and in above calculation is also due to rounded off numbers in my first post *Edit* Corrected some miss calculations | ||
|
Disastorm
United States922 Posts
| ||
|
LordOFswe
Sweden27 Posts
On June 04 2010 07:38 Disastorm wrote: Since this seems to be a harvesting thread, just wanted to point out that it was discovered awhile back that if you put your hatcheries/nexus/cc one hex farther than normal from the resources, you actually get resources faster when it is fully saturated due to more synchronized timing of the workers getting to the minerals. I just thought thats pretty interesting. I think it applies to any point after 16 workers. Cooolio, have to test that... | ||
|
hacpee
United States752 Posts
On June 03 2010 23:01 PJA wrote: What exactly is the point of the curve fitting? I mean, we already know from the other 20 threads on the topic that the last 8 workers are far less efficient than the first 16. What's special about the last 8 being fit by a cubic? Why not cubic since it is usually more accurate? | ||
|
Cheticus
United States83 Posts
On June 04 2010 06:54 Piousflea wrote: What matters to a real game is the incremental income per worker. If the first probe harvests 40 minerals/min, how many do you get from probe #17, 18, etc? As far as the paper goes, your curve fittings make no sense. The linear portion of the curve has a non-zero intercept. At 0 workers you lose minerals, and the 1st worker is less effective than the 2nd worker. The cubic (why cubic?) portion of the curve eventually turns downward, meaning that if you extrapolate to 100 workers, you again lose minerals. You can use a spreadsheet to fit any curve to any data set. It's only a valid technique if your curve has some basis in reality. The most mathematically sound approach is to use intercepts and asymptotes that make logical sense. That is, you need a curve that starts at 0 income for 0 workers and eventually levels off at a fixed income with infinite workers. One logical approach would be a two-part linear-hyperbolic curve: W<16: I = a*W W>=16: I = b + ( (c * W) /(d + W) ) This function starts at 0, and trends toward (b + c) as W goes to infinity. Ideally, you would also want to constrain the second curve to keep a smooth slope. The reason the linear fit is as it is because I performed the fit in a least squares sense on the data I took for one worker through four workers, and that is the result. If you look at where the data picks up again at the 16th' worker, it's almost on the line, which indicates that it's likely a reasonable fit. The reason I used a cubic fit is that for 16-24 workers it looked pretty close. The cubic fit is least squares also. Don't extrapolate it, since it's only a description of data in a specific range. Also, non-integer workers don't make any sense, so don't interpolate between workers. I didn't think there was reason to have a smooth slope or differentiability or any of that jazz because I'm not making any claims based on the curve fits, only that these are the results that I found: that they match what we already know, and that this is a calibration that describes those results. If you're interested in creating your own fits, I've included a the processed (time and mineral differences already taken (read: not raw)) data I took for my fits below, feel free to use it if you want to take it further (I've never done a hyperbolic fit): W Test1 2 3 4 1 55 55 55 50 2 115 120 115 120 3 175 170 180 170 4 240 235 230 240 W Test: 1 2 3 4 16 935 930 935 940 17 975 970 960 965 18 990 970 980 995 19 1025 1035 1020 1030 20 1075 1075 1065 1075 21 1085 1095 1095 1105 22 1100 1110 1130 1110 23 1130 1115 1135 1130 24 1060 1135 1125 1120 Also, Disastorm--that's an interesting idea I'd never even considered. In general the situation wouldn't arise where it would be beneficial UNLESS you were in a spot where you had maybe your main mined out and natural close to mined out, and then were going to transfer most of your workers from main and natural to a new third expansion. I suppose it makes sense, based on how much time a worker wastes just chilling at the mineral patch when there's three harvesting it. | ||
|
iSTime
1579 Posts
On June 04 2010 08:04 hacpee wrote: Why not cubic since it is usually more accurate? My main complaint is that he just gives a bunch of data with no conclusions. Obviously more workers over 16 aren't as efficient as the first 16, this has been shown dozens of times already. How is anything in this thread more useful than using the search function? | ||
|
Piousflea
United States259 Posts
If your goal is to "fit the data as close as possible", you might as well just use a data table for 1-24 workers, with empirically measured means and standard deviations. In most real-life physics or engineering situations you'd use empiric data. The entire purpose of using a mathematical model is to simplify the data to some kind of "smooth curve" or theoretical framework. Pulling formulae out of Excel without a rationale is offensive to my personal sense of mathematical taste. | ||
|
Sanasante
United States321 Posts
| ||
|
Crixus
Canada110 Posts
first set of 8 workers = 350 minerals/min second set of 8 workers = 350 minerals/min third set of 8 workers = 100 minerals/min (i.e. u typically see people with 700 minerals/min or 800 minerals/min from one base) which is why when people transfer probes to expo, they move all but the first 2 sets of probes to avoid effect of diminishing returns | ||
|
Hidden_MotiveS
Canada2562 Posts
The main thing to note is to just get 3 workers per patch. After about 24 the increase in mineral harvesting is flat (experimentally). Knowing this can be useful in determining whether you want a new base or not. Before 20, the increase in mineral harvest rate is nearly linear. This means that if you want to be aggressive and take a lot of bases without fully saturating them, then 2.5 workers per mineral patch is a better number. Now in the heat of battle.... just keep making workers is all you need to know. | ||
|
Sanguinarius
United States3427 Posts
| ||
|
virgozero
Canada412 Posts
| ||
| ||