Hi!

Below is link to PP packet teraforming chances calculator (Excel).

http://www.interlogic.com.ua/StarsClub/Downloads/PP_Terra_Ca lculator.zip (12kB).

Any comments are welcome.

Report message to a moderator

Report message to a moderator

Hi!

Dogthinkers wrote on Mon, 13 March 2006 04:34

As far as I see, you only retrieve about 10% of the minerals used for this purpose... Does that sound about right? So you will get a mean terraforming of just over 10% (permaforming 1%) with a 2100kt packet, at a cost of 150 resources and with about 1900kt of minerals lost.

(Minerals recovered are 1/3. From 2100 you lose 1400, not 1900.)

Think in another way: how much cost your mines have? How many mines you need to mine lost 1400kT in 5 years? Add these resources to packet launching cost, and you will have "actual" cost" of terraforming in resources equivalent.

Quote:

Hmm, I wonder how high I can crank mining efficiency and keep the race viable through using packet forming to prepare the yellows and low greens.

Well, try to leave some leftover resources into mines, if your PP is not -f While you building initial factories, these additional mines would do their work. Also, it is not beneficial at all to packet-terraform just from start (unless you have unusual start). Better mass drivers speed up things - you probably would not want to waste minerals because decay, and warp 5 packets are really really slow, so that is another reason why it is bad idea to teraform early in the game. So, for me it looks like it is better to establish far colonies only at good green worlds, and wait a bit with other "lesser" greens and yelows (if universe is not very tense). Then terraform and colonize all close "lesser" worlds around after you get more minerals. -f race should benefit the most from this.

I did not played PP much, so I might be wrong.

Report message to a moderator

Tomasoid wrote on Mon, 13 March 2006 17:46

Hi! Dogthinkers wrote on Mon, 13 March 2006 04:34 As far as I see, you only retrieve about 10% of the minerals used for this purpose... Does that sound about right? So you will get a mean terraforming of just over 10% (permaforming 1%) with a 2100kt packet, at a cost of 150 resources and with about 1900kt of minerals lost. (Minerals recovered are 1/3. From 2100 you lose 1400, not 1900.) Think in another way: how much cost your mines have? How many mines you need to mine lost 1400kT in 5 years? Add these resources to packet launching cost, and you will have "actual" cost" of terraforming in resources equivalent.

I just lobbed a 10,010kt Ironium packet at warp 5 at a colonised, but undefended world. When I recolonised it there was 1147kt of ironium on the surface. Under 12% recovered

I then lobbed a 5040kt Boranium packet at warp 5 at an uncolonised world. When I colonised it there was 566kt on the surface. Under 12% recovered.

I then lobbed a 2030kt Boranium packet at warp 5 at a colonised (no defenses, no driver) world with high enough population to survive. 226kt were recovered. Just over 11% recovered.

I then lobbed a 350/350/350kt mixed packet. 38kt of each mineral was recovered. A touch under 11% recovered.



The empirical evidence here is that approximately 11% of the minerals are recovered (at worlds without drivers or defenses.)

(Obviously these throws were with PP and thrown slowly, so no degredation took place in flight.)

[Updated on: Mon, 13 March 2006 20:29]

Report message to a moderator

Hmm... See help file, "Guts of mass driver". Also, see other articles related to PP race. There is always 1/3 mentioned.

It is surprise, you seems uncovered one more discrepancy between rules and practice! Is it the same for uninhabited planets?

It may be also becasue following: there was mentioned that inhabited planet automatically disperses 1/3 in the atmosphere. It looks like 1/3 of THAT is dropped on the inhabited planet. 1/9 is very close to what you got .

Report message to a moderator

Report message to a moderator

I did not subtract the amounts of the colonisers... Santa maria, so only about 20kt... So amount recovered is still between 11% and 12%. If I remember correctly, on the strikes on inhabited worlds I lifted any surface minerals before the packet hit, so fingures should be pretty accurate.

You'll see that some of the drops were on uninhabited worlds, some on inhabited worlds, and some on inhabited worlds that became uninhabited as a result

The % amount recovered was similar in all cases, but not quite identical. I suspect the differences are more to do with rounding errors or some small random factor, rather than anything to do with the status of the colony. It seems that once again the help file is a work of creative art, rather than description of actual game mechanic.

Report message to a moderator

Dogthinkers wrote on Wed, 15 March 2006 01:43

I did not subtract the amounts of the colonisers...

Then your tests clearly show that recovered is 1/9 not 1/3. Just divide minerals by 9 and seems the results match. 1/9 is 11.11% so its between 11% and 12%
Oh well ... 3 times bigger losses make PP 3 times less interesting.

Report message to a moderator

Report message to a moderator

Report message to a moderator

Report message to a moderator

[Updated on: Tue, 18 April 2006 04:43]

Report message to a moderator

Report message to a moderator

[Updated on: Tue, 18 April 2006 09:30]

Report message to a moderator

Tomasoid wrote on Tue, 18 April 2006 09:21

Some explanation about why 8x350 << 1x2800: Chances to get teraforming from 0% to 6% are far better for 8x350 vs 1x2800. Chances to get from 28% to 32% also exist, but very small. So, in general, average is the same, but you have more risk to get little teraforming when sending 8x350. Taking that risk is not very profitable - you get only very little chances to hit jackpot by that (from 28% to 32%).

Still confused about this. I'll run some tests again using your calculator and see if it makes more sense

Report message to a moderator

Report message to a moderator

PricklyPea wrote on Tue, 18 April 2006 17:41

Tomasoid wrote on Tue, 18 April 2006 09:21 Some explanation about why 8x350 << 1x2800: Chances to get teraforming from 0% to 6% are far better for 8x350 vs 1x2800. Chances to get from 28% to 32% also exist, but very small. So, in general, average is the same, but you have more risk to get little teraforming when sending 8x350. Taking that risk is not very profitable - you get only very little chances to hit jackpot by that (from 28% to 32%). Still confused about this. I'll run some tests again using your calculator and see if it makes more sense

No need to. I ran "live" tests with real packet hits. Just download file http://www.interlogic.com.ua/StarsClub/Downloads/PP_Tera_Tes t_Results.zip and look to the "Test results - Multiple hits" tab. For 10x140: Select cells with chances to teraform below 5% and make a sum. Chances to get below 5% are 0.097916484. Now go to tab "Test results - 400-1500" in the same file and do the same for 1x1400. Chances for it are 0.087892848. No big difference, but it exists. Now look again to 10x140 and see if there is any hit above 14%. From almost 23k trials I got only 15% max and only 2 times. So, why select 10x140 vs 1x1400 for? What you get more risk to get below 5% for?

See also following thread:
http://starsautohost.org/sahforum/index.php?t=msg&th=283 1&start=0&rid=844&S=bd7c0bb0f4f1e91cf1d08e571800 0782
There are few messages discussing that.

In general, it is possible to calculate chances for multiple packets, BTW, with easy, but only if all packet sizes are equal. If needed, I can improve calculator and add section to calculate multiple packet hit chances. This would make file a bit heavier because it would require to use a matrix for such calculation (table at least 100x100 cells). However, is it needed? The difference is quite small.

Report message to a moderator

PricklyPea wrote on Tue, 18 April 2006 15:59

What I struggle to understand is why 2x700kt == 1x1400kt but 2x350kt != 1x700kt?

Some simple statistics (well it was in my day )

Firstly, a 700kT packet is the same as 7x100kT packets.
That's because by definition that's exactly what happens when the terraforming takes place. Any rock is broken down into 100kT chunks and each chunk is calculated separately.
So 2x700kT packets gives you 14x50% chances of terra, and so does 1x1400kT packet, so they are the same.

However a 350kT packet has 3x50% chances and 1x25% chance.
So 2x350kT packets has 6x50% chances and 2x25% chances.
This is not the same as 7x50% chances.
The average is the same, but the chances of any specific amount of terra is different.

To show it simply consider the chances of getting 0 clicks of terra -
for 1x700 packet = 0.5 ^ 7 = 1/128
for 2x350 packets = 0.5 ^ 6 x 0.75 x 0.75 = 9/1024 (which is bigger than 1/128)
So the second case has more chance of getting no terra, despite it also having more chance of getting more terra.

n.b.
This is a bit like the "square effect" (where 6x6 > 5x7) because the chances af getting x clicks is some combination of the product of the individual chances, whereas the average number of clicks of terra is more like the sum of the individual chances.
For the same sum (average) you get the highest product when all the instances are the same (square effect).

Report message to a moderator

PricklyPea wrote on Tue, 18 April 2006 17:59

Did a test using 2x350 vs 1x700 I found that chance of getting 1% terraforming was 5.86% vs 5.47% but chance of 3% terra was 26.95% vs 27.34%. What I struggle to understand is why 2x700kt == 1x1400kt but 2x350kt != 1x700kt? Also, at the danger of combining only 2 data points plus your assertion that the average remains the same, then sending lots of small packets appears to give a larger spread of values with a flatter distribution. Anyone here good with excel who know how to convert the spreadsheet into one which can calculate the distribution for say 3x700kt packets so that it's possible to plot a comparison between different amounts?

I already did it I had it at home, but lost ( I can re-make it again if needed. It needs a matrix at least 100x100 to calculate, and limitation is that only the same size packets (i.e. 5x300, not 2x200+4x300 or something like that) are possible to calculate that way exactly.

You're right that distribution is flattened for multiple packets. It is because there are chances to get something above usual with multiple packets. 2x350 have chances to give you 8% teraforming, while 1x700 have no such chances In a large numebr of hits run, average is the same, however it is more risk to get less teraforming than 1/3 of maximum.

Report message to a moderator

Report message to a moderator

Report message to a moderator

Report message to a moderator

Report message to a moderator