|
On April 26 2010 17:27 Arrian wrote:
So, just for kicks, if you were to put those together to get numbers for gas/s per mineral invested ( * 1000 )
1(1) 5.92
2(1) 8.57
3(1) 8.88
3(2) 7.43
4(2) 8.71
5(2) 8.32
6(2) 8.75
3 workers on one gas would seem to be the most investment-efficient way to mine... but why isn't it the same rate as 6(2)? Just discrepancy in distance between the two geysers? I guess this really just tells us, "saturate the closest geyser first". Kinda obvious, I guess, but I never really thought about it.
|
If you want less error and more actually significant figures the best way is to just stick the appropriate number of workers in the gas and see how long it takes to mine out the gas (and of course assume that the mining rate is continuous). This also much more clearly reveals whether there is a difference between different situations since a small difference compounds over the long time it takes to mine out an entire geyser.
I definitely like the effort and SC2 needs more analysis on resource gathering especially with its implications on mapmaking.
|
so 4 workers on 2 gas is more efficient than 6? Cool.
|
Articles like this make me feel inadequate when it comes to math :/
At least there are kind people like the OP who spend their time and effort for the community :D
|
United States889 Posts
On April 27 2010 00:20 Kickchon wrote:Nice analysation, however I recently did those measurements myself and as your results differed from mine I have redone my experiments and I would like to share my findings. My experiments were done on Blistering Sands on the southwest main on faster speed. I measured every race for the southern geyser and the northern geyser for protoss additionally to compare both. Table in spoiler: + Show Spoiler +My results have the following implications: 1. All 3 races gather at the same speed. (Requires creep for the zerg, but the hatchery is there anyway) 2. The northern geyser might be a little bit more efficient because of its distance. Even if it is insignificant building the northern geyser first has no drawback. 3. Maybe the most drastic: The diminishing returns of the 3rd worker are nearly non-existant on both geysers. This should demonstrate the differences on different maps more clearly. And in conclusion the statement 3(2) > 3(1) might not be generalizable and would require further research. Edit: I just noticed I should have done 4 workers as well. :/
Nice work
I don't think our results are in conflict because they all seem to vary appropriately based on the difference in geyser position. The differences in rates seem to be a function exclusively on that. And yeah, the 3(2)>3(1) seems to be heavily dependent on the geyser position, to the point where the mineral investment would almost always make 3(1) a better choice than 3(2)
|
United States41911 Posts
Geysers are only 75!?? Then why have I been waiting for 100 every game.
|
Would be interesting to add how many minerals/s a drone collects, anyone?
|
On April 27 2010 02:06 KwarK wrote:
Geysers are only 75!?? Then why have I been waiting for 100 every game.
Yeah. I noticed that for Terran after a few games. I said to my friend, "Wtf; my reaper rush can come so much faster now."
Nice analysis. It's not too beneficial for my current loose build orders but I can see it being key with more development of precise build orders.
|
nice information
it seems this is mostly a lesson in common sense, but a good lesson nonetheless
|
Your definitely going to be one of the professors at starcraft university
|
Nice,
I was thinking about 'What is the best gas strat' today =)
Thanks for write up!
|
As others have said, there is no science or maths here, just some annotated arithmetic.
Furthermore it makes absolutely no sense to regress a function on a domain for which it is undefined. Quite a lot of pretentious nonsense in this article.
|
|
|
Nice writeup. Enlightening, especially on the 3(2). Was this a single run or multiple tests on the same position?
|
You know, I've been wondering about this kind of stuff for a while and I'm glad you took the time to write an article on it! This really helps out a lot, especially the table where you can compare them all. I'll put this to good use; keep up the good work 
|
Arrian, this is a great analysis; one direct practical conclusion would be knowing how much to invest on gas mining while trying to FE (several tier 1.5 units require gas - and determing just how many tier 1.5 units one should build to FE successfully!)
but instead of including a "minerals invested" vs "rate of gas"
the logically correct comparison should be rate of minerals mined vs rate of gas mined; with x workers on gas, (total workers - x) on minerals of course
i can help you with the analysis if you would like!
|
Thanks for this. You put in a lot of effort, and it shows.
|
I agree that most of this article was unnecessary fluff, but there was one very important discovery that I actually think I will incorporate into my play: the fact that the third worker doesn't add as much as the first two.
Basically, early game it might give a significant advantage (in regards to net wealth) to get two refineries with two workers on one and one or two on the other rather than one refinery with three, and keep the leftover workers on minerals until gas is needed more urgently.
The dividing of gas in sc2 was definitely a good idea; it makes getting gas an actual choice. For map makers, I think you should consider expansions or even mains with only one or possibly three geysers.
|
It is not surprising that your rate of gas mining is proportional to the number of workers mining gas. Because your rate of collection is constant per worker, when you add an additional worker, you necessarily increase your rate of gas collection per time by some approximate exponential factor. For example, consider the graphs y = x for one worker, y = 2x for two workers, y = 3x etc. When you factor in build times for the workers and offset those linear functions by said build times you get something that looks like the exponential function. From kinematics you know there will be some max rate that is determined by the distance of the geyser from the main building, worker acceleration (do workers accelerate?) and speed etc.
Some specific comments:
The first graph is labeled incorrectly. Gas x time is not a graph of gas collection rates. The rate of gas collection per time is the slope at any point on the gas x time graph. I found the differently shaped graph points distracting given that you are trying to analyze the overall behavior of the graphs as opposed to comparing specific points between them.
The second graph isn't that helpful. Intervals of time to the geyser vs. workers on gas. It starts high and goes down. There's not a lot to say about it. A more useful graph would a graph of gas collection per second x workers/geyser b/c it can help you visualize what is significant about rates of resource collection: potential unit production. For example, if your gas collection per second is 20 gas/sec then you can calculate the optimal number of barracks to support continuous marauder production given that you know the marauder build time. I will leave this as an exercise for the reader
Also, it would be cool if you added your data files to the OP.
I hope you found my comments constructive ^_^ Thanks for taking the time to share your research and analysis!
|
|
|
|
|
|
|
|
EPT![[image loading]](/staff/Arrian/efficiency_table.png)
![[image loading]](http://img411.imageshack.us/img411/5067/collectionblisteringsan.jpg)
