Jump to content

Psi (Again)


Bomb Bloke

Recommended Posts

Here are my observations about psionic patterns to date. I'll update this post every now and then as I get more information.

 

This shows the results of mind control attacks from a soldier, against an alien, with the listed attributes:

 

Soldier		  |Alien			 |
Skill |Strength  |Skill   |Strength |Percent |Assumed Percent
------+----------+--------+---------+--------+---------------
100   |100	   |100	 |100	  |100	 |100
------+----------+--------+---------+--------+---------------
75	|75		|75	  |75	   |75	  |75
50	|50		|50	  |50	   |25	  |25
25	|25		|25	  |25	   |11	  |10
------+----------+--------+---------+--------+---------------
75	|75		|100	 |100	  |21	  |20
50	|50		|100	 |100	  |0	   |1
------+----------+--------+---------+--------+---------------
75	|100	   |100	 |100	  |93	  |95
50	|100	   |100	 |100	  |9	   |10
------+----------+--------+---------+--------+---------------
100   |75		|100	 |100	  |95	  |95
100   |50		|100	 |100	  |8	   |10

 

The two units were standing right next to each other. Both had 100 morale, and each test was made up of 100 trials.

 

It seems that psi skill is just as important as psi strength when it comes to attacking.

 

Note that a soldier with 100 skill/strength will NEVER fail a mind control attempt. However - the same is NOT true if an alien with 100 skill/strength attacks YOUR units!

 

I would assume that psi skill has little role to play in terms of psi defense, as most units don't have it.

Edited by Bomb Bloke
Link to comment
Share on other sites

Cool stuff, Bomb Bloke!

 

I took a quick look at your data, and am wondering how you arrived at the "Assumed Percent" column. Is it just an educated guess based on results, or do you know something I don't? :cool:

 

Trying to base an overall equation off this data is going to be a tad bit tricky, since all the conventional ones don't seem to hold up. (I even tried Danial's wiki suggestion of using a strict addition and it still doesn't work). Those trials where there is a very low probability of success are going to be the deal-breakers here. If those areas are researched in-depth we might just actually stumble on the correct equations.

 

I'll set something up myself and try to gather some more data on those "grey areas". That is, unless you have a saved game you would be willing to share. :)

 

- Zombie

Link to comment
Share on other sites

The Assumed Percent is exactly that - assumed. Because I'm only doing a hundred runs per setup, there's plenty of room for error, so a guess is as good as the results. :cool:

 

The idea is that when enough setups have been run through, the patterns will be easier to spot.

 

I have a save file, and it's attached... I didn't really plan on distributing it, so you'll have to excuse the fact that it's using my modified desert terrain.

 

In the save game folder is a small editer which'll let you quickly tweak the stats of the units in concern. They're easy enough to find, as they are standing right next to each other. :)

GAME_10.zip

Link to comment
Share on other sites

Thanks BB! That sure makes things easy!

 

Right, so I didn't have a whole lot of time to work on this today, but I managed to run a few trials where the Test Soldier had 50/50 and the Floater had 100/100. I picked this scenario because it had the lowest chance of success. The Psi-Amp has a per use charge of 25 TU and this fact coupled with the low chance of success and the max TU limit of 255 led me to take a shortcut. I just edited current TU to 250 and ran 10 trials at a clip. That really cut down the research time! This trick will not save any time if the chance of success is high, so that's why I aim low first. :)

 

Out of 1000 trials the soldier was unable to MC the Floater even one time. I'd have to say that would be a 0% chance. Maybe I'll try the 100/50, 100/100 scenario next. Here's hoping for 10%! :cool:

 

- Zombie

Link to comment
Share on other sites

Anyone want to try to see how much distance affects psi? I'm pretty sure it plays a big factor, at least when I'm using psi-newbies on regular alien grunts.

 

I'm on it at the moment but I have found that moral also effects the result for the aliens so it's not a quick job.

Link to comment
Share on other sites

I mean, I've seen it on the end where I'm mind-controlling the aliens. On the same mission, with the same person, mind controlling the same alien, the farther away the alien is from the soldier the less likely mind control will work. (Try a soldier of strength >80 and skill about 20)

I can only assume that this is also the case for the other way around.

Link to comment
Share on other sites

  • 2 weeks later...

I just did some testing on the conditions where the soldier had 50/50 and the Floater also had 50/50. From BB's preliminary results, he had a 25% success rate. Well, after 510 tries, the test soldier was only successful 130 times. That's approximately 25.5%. Close enough. From my previous test, when the soldier had 50/50 and the Floater 100/100, no attacks worked. Since the soldier has constant psi levels, I theorize the following results when the Floater has different numbers:

 

Psi Skill		  % Chance
 100/100			  0%
 90/90				5%
 80/80			   10%
 70/70			   15%
 60/60			   20%
 50/50			   25%
 40/40			   30%
 30/30			   35%
 20/20			   40%
 10/10			   45%
  0/0				50%

So for every 10 point decrease in Floater psionic skills, the chances for success increase by 5%. This is still theory yet and may only hold with a soldier having 50/50 Psionic Strength and Skill. Still, it is a start. I might try the 50/50 - 90/90 scenario next to see if the 5% guess is correct. Then it's on to the 50/50 - 0/0. :(

 

- Zombie

Link to comment
Share on other sites

Shouldn't you test with only 1 variable changing to begin with? Changing both the soldier's skill and strength might not give a very clear picture of what's going on. If you start with 100's all round and then decrease one stat only, it should paint a clearer picture... shouldn't it? :(
Link to comment
Share on other sites

For my first group of tests, I am keeping the soldiers psionic stats constant at 50/50 and moving the Floaters stats by ten. This starting point is based off of BB's preliminary results. It makes no sense to start out verifying that a soldier with a PsiStrength of 0 and Skill of 1 will fail every attack against a 100/100 floater. Nor does it make much sense to check what happens when a soldier has 100/100 and the Floater 0/0. Those trials will prove useless. However, where there is a chance of success

 

Once the basic structure of the 50/50 trial is understood, then I'll keep one of the floaters stats fixed and the other variable. Because there will be an approximate range to work off of, figuring out the next series will be much easier. :(

 

By my calculations there are 658,127,716 possible combinations to test out (101*254*101*254). Knowing where to start is anyones best guess with a project of this magnitude. The 50/50 set seemed to offer the best in terms of variety and range. Just bear with me here. I know what I'm doing. ;)

 

- Zombie

Link to comment
Share on other sites

  • 2 weeks later...

Zombie: I'd suggest two things.

 

1) bear in mind we're trying to come up with a formula, so it's always important to be able to pull out data sets where one variable changes and the other three are held steady, so we can at least see if the success rate is a straight line or curved relationship. (y=mx+b, hopefully).

 

2) Take a spread data set: eg start with strength scores of 20, 40, 60, 80, 100, which will reduce the number of trials required for your first run and very quickly tell you what the most interesting range is for your second run. On the second run (say 40<x<80 is where the action happens) divide the smaller interval into about 10 sections and repeat.

 

Much like you're suggesting - moving scores by 10 will reduce the sample size, making life easier, but you've still 10x20x10x20 points in the first run

 

Happy Christmas

JFG

Link to comment
Share on other sites

  • 2 weeks later...

Well, I had a chance to run a few psi trials over the holidays. The soldier remained a constant at 50/50 for all tests. The soldier had a 0% chance of success over 1000 trials. From this, I concluded that the 5% chance increase theory per 10/10 psionic skills increase was incorrect. So I forged on. With a 60/60 alien, the soldier had a 2.575% chance of controlling the alien (again, 1000 trials). Right now, I am wrapping up the tests with a 40/40 alien. Preliminary results indicate a 50% chance. I suppose it would be easier to summarize in a table:

 

Psi Skill		  % Chance
100/100				 0%
 90/90				 0%
 80/80				??
 70/70				??
 60/60			   2.6%
 50/50				25%
 40/40				50% (still underway)
 30/30				??
 20/20				??
 10/10				??
  0/0				 ??

So as you can see, there has been some progress. The 60/60 scenario was very interesting to run because of the low chances for success. A while back I also had to revise my gathering process a bit as I thought that the mission before might play a role in the current one (artifacts still in memory?). Therefore, no more strict reloading - I am now aborting and then loading from that screen instead as a precaution. :(

 

- Zombie

Link to comment
Share on other sites

Hey, a hundred trials might be just as useful as a thousand, while we're trying to find the shape of the curve... not that I want to stop you if you're having fun...

 

Whilst he's doing that he's not getting into trouble. By the way, has anyone done any tests on the "random" shots that troops make when they go beserk? It seems to me that whereever I have other troops standing they seem to be a target. It's not what I'd expect to happen in reality - shoot your own troops.

Link to comment
Share on other sites

Sometimes troopers spin around randomly (had a guy up on a roof do that, firing 255 TUs worth of laser rifle blasts).

 

Sometimes they target some poor sap, be it a civilian, an alien, or one of your own units. They generally keep going until they run out of TUs, regardless as to when their target dies.

Link to comment
Share on other sites

Hey, a hundred trials might be just as useful as a thousand, while we're trying to find the shape of the curve... not that I want to stop you if you're having fun...

I always found that 200 trials are the bare minimum for understanding how something works in X-COM. 1000 trials usually give the correct number of points needed for accurate comparisons. And since I'm all about accuracy... :(

 

The nice thing about these tests is that 1000 trials can be completed in a rather short time frame if the chances for success are low. By cranking (editing) current TU up to 250 it allows the soldier to attempt up to 10 trials per round. Since most attempts fail, this trick speeds things up tremendously.

 

- Zombie

Link to comment
Share on other sites

  • 4 weeks later...

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
  • Create New...