hydraSlav's
Ostrakon Agency Gallente Federation
290
|
Posted - 2013.10.11 14:15:00 -
[1] - Quote
Haerr wrote: The way stacking penalties work in EVE ( and Dust, you can check the stacking penalties against Cardiac and Kinetics and they will follow this pattern. ) is like this:
For positive modifiers: Base * Skill Modifier * Skill Modifier * (1+Module_Modifier) * (1+Module_Modifier*Stacking Penalty) * (1+Module_Modifier*Stacking Penalty)GǪ
For negative ( reductions ) modifiers: Base * Skill Modifier * Skill Modifier * (1-Module_Modifier) * (1-Module_Modifier*Stacking Penalty) * (1-Module_Modifier*Stacking Penalty)GǪ
The stacking penalty increasing for each additional module.
The stacking penalty can be approximated by = 0.5^(((n-1)/2.22292081)^2) where n is the number of the module. This makes the modules work with the following efficacy. 1st module = 100% 2nd Gëê 86.91% 3rd Gëê 57.06% 4th Gëê 28.30% 5th Gëê 10.60%
If we apply this to a medium suit, which has a base scan profile of 50 dB, we get the following:
Base = 50 dB Skill Modifier = 2% reduction in scan profile per level of 'Profile Dampening' (10% reduction at level 5) Complex Profile Dampener = 25% reduction in scan profile Enhanced Profile Dampener = 20% reduction in scan profile Standard Profile Dampener = 15% reduction in scan profile
So a medium suit with level 5 in profile dampening and 2 complex profile dampeners will have: 50 * 0.9 * (1-0.25) * (1-0.25*0.8691) Gëê 26.42 dB scan profile
This is clearly lower than the 28 dB scan precision of a prototype active scanner and the suit will not be detected.
However if you have a medium suit with level 5 in profile dampening and 1 complex profile dampener and 1 enhanced profile dampener you will get the following: 50 * 0.9 * (1-0.25) * (1-0.20*0.8691) Gëê 27.88 dB scan profile
This is also clearly lower than the 28 dB scan precision of a prototype active scanner and the suit should not have been detected, how ever it is detected. ( we tried a few different prototype scanners and all of them detect this suit. )
Now we assume the reason for it being detected is because of the "armour bug".
In game: Complex armour plate has a movement penalty of 5%. Complex ferroscale plate has NO movement penalty how ever we assume it actually has one that is set to 0%. ( The calculation will show why we think this. ) Medium suit: 5.00 m/s 1st module, complex armour plate: 4.75 m/s 2nd module, complex armour plate: 4.54 m/s
Now the weird bit, we put on a ferroscale plate.
3rd module, complex ferroscale plate: 4.64 m/s
A few quick calculations to show stacking penalties: 5.00 * (1-0.05) = 4.75 5.00 * (1-0.05) * (1-0.05*0.8691) Gëê 4.54
Now for the weird bit: 5.00 * (1-0) * (1-0.05*0.8691) * (1-0.05*0.5706) Gëê 4.64
As you can see here both the assumption that the ferroscale actually has a movement penalty of 0% and that it is calculated 'first in line' in the stack is confirmed.
Now supposedly EVE ( and Dust ) are to apply the largest modifiers first and then the smaller ones. This leads us back to the example of a medium suit with 1 complex- & 1 enhanced profile dampener being detected by a prototype active scanner. You see if you apply the stacking penalty in the "wrong" order you will get: 50 * 0.9 * (1-0.20) * (1-0.25*0.8691) Gëê 28.18 dB scan profile
Which would explain why the suit is, currently, being detected by a prototype active scanner.
I can confirm that the above is all correct. I've tested and re-tested the "order" of modules for stacking penalty multiple times when developing DFT, and yes Dust simply sorts them from highest to lowest (numerically), which means for negative values (penalties), the hardest penalty gets the most stacking reduction, which is contrary to bonuses and normal logic. It should be fairly simple to make it work "correctly" by having an ABS function in front, but i am sure CCP has their reasons.
Interesting observation about Ferroscales counting as a 0% penalty. I could replicate the result by assigning -0.0000001% penalty to Ferroscales in DFT. Unfortunately can't use a penalty of 0% in DFT as very deep down in the formulas a zero value is used to determine if there is no penalty/bonus assigned.
Did DEVs confirm this to be a bug that will be fixed? Or intended? Should i just add -0.0000001% to DFT if it's staying like that for unforeseeable future?
Also, do you know if Basic Reactive Plates exhibit the same behavior (i am away from home for several days, can't check)?Other reactive plates have a small penalty, but Basic have none (according to stats). Just replace Ferroscale with Basic Reactive Plates in your example.
BTW, on topic of movement, did you know that the movement bonus and movement penalty are actually 2 separate attributes, and stack independently of each other, although they affect the same characteristic? Read about it here.
P.S. You are the first i've seen here using Aenigma's stacking formula. I use that too in DFT. Don't know why it so unknown compared to the other formula in all wikis. |
hydraSlav's
Ostrakon Agency Gallente Federation
303
|
Posted - 2013.10.23 14:15:00 -
[3] - Quote
Haerr wrote:hydraSlav's wrote:Oh, just noticed that DFT has Scan Profile as a non-stacking attribute (i guess it didn't mention stacking in the description). Your results suggest that it is stacking. Is this pretty much the consensus of the community? Also what about Scan Radius (it is also currently marked as non-stacking in DFT) Me and a corpmate did a bunch of testing with active and passive scanning and yes 'Precision Enhancers', 'Profile Dampeners' and 'Range Amplifiers' all suffer from stacking penalties. Edit 1: I do not know if anyone got a confirmation that this is a bug from a dev, to be fair I do not remember anyone saying that they heard back from any dev regarding this. I am certain, how ever, that it was reported back when ferroscales were introduced. Edit 2: It was confirmed as a defect by 'CCP Cmdr Wang' back in july. Link to weekly bug reports: https://forums.dust514.com/default.aspx?g=posts&m=1057188#post1057188 - '[BUG] Ferroscale speed bonus?' 'xxwhitedevilxx M' reported the bug on the forums in the begining of july: https://forums.dust514.com/default.aspx?g=posts&m=1009079#post1009079
Thank you Haerr
I've released DFT 3.4.2 which enables stacking penalty on those modules, thanks to your evidence.
It also replicates the "armor bug" by giving it a minuscule negative penalty. I should have caught that one sooner If you find more such issues, please post them in the DFT thread so i can take care of it.
Cheers |