Author 
Topic 
Dr Corvax 
Posted  2008.08.11 07:27:00  [ 1]

Kazuo Ishiguro House of Marbles

Posted  2008.08.11 10:01:00  [ 2]
Good work. Have you considered turning your attention to the align time formula (i.e. time for a ship to get into warp from a standing start)? There were some cases that were never adequately covered by the existing one: t = ln(0.2)*agility*mass/1,000,000
In particular, ships with oversized plates and freighters seem to differ somewhat from the predicted times. 
Catelco La Isla del Mono

Posted  2008.08.11 10:15:00  [ 3]
Just a pointless though:
As the formula presented by the OP may be accurate (I don't know), I find that is too much complicated computationaly speaking (sqrt, exp...) to be calculated for every ship, every tick... May the real formula be more simplystic... or maybe CCP uses such kind of formulas with fixedpoint math or .... well who knows and who cares?
Anyways, nice research 
Gypsio III Dirty Filthy Perverts

Posted  2008.08.11 11:23:00  [ 4]
Good stuff. But does anyone have a formula that describes the rate of shield recharge? 
Mona X Caldari Missions Mining and Mayhem Northern Coalition. 
Posted  2008.08.11 11:39:00  [ 5]
It's the same. 
Dr Corvax 
Posted  2008.08.11 11:48:00  [ 6]
I think so. It's seems that shield and capacitor restoration has same algorithm. 
Wil Smithx Minmatar Brutor Tribe 
Posted  2008.08.11 11:59:00  [ 7]
Originally by: Catelco Just a pointless though:
As the formula presented by the OP may be accurate (I don't know), I find that is too much complicated computationaly speaking (sqrt, exp...) to be calculated for every ship, every tick... May the real formula be more simplystic... or maybe CCP uses such kind of formulas with fixedpoint math or .... well who knows and who cares?
Anyways, nice research
Wouldn't be hard for a comp to do a calc like that would it? While its somewhat more long winded to do on paper think of all the calcs eve has to do anyway, like tracking, momentum, stuff like that... 
Akita T Caldari Navy Volunteer Task Force

Posted  2008.08.11 12:14:00  [ 8]

Catelco La Isla del Mono

Posted  2008.08.11 12:23:00  [ 9]
Square roots, exponentials and divisions with floating point numbers are "expensive" for the CPU. While additions or differences can be processed in a few CPU cycles, complex operations may need hundreds of cyles IIRC.
Doing a single calculation won't hurt. But recalculating capacitors, and shields for every ship active in eve, along with the other hundreds of calcs (tracking, drones, warping, moving...), in every "game" tick >may< represent a noticeable difference.
I repeat, I don't know how internally CCP manages such calculations, but I would guess that with the "need for speed" initiative, thoose calculations would be much more lightly. Or just the Tranquility servers are much more dependant on database optimisation than aritmethic optimizations

Gypsio III Dirty Filthy Perverts

Posted  2008.08.11 12:49:00  [ 10]
Awesome. I've always wanted to write an equation that describes the contribution of passive regen to buffer tanks whilst taking n DPS. It sounds a bit fiddly but I'm sure it's doable. 
Derek Sigres 
Posted  2008.08.11 13:25:00  [ 11]
Originally by: Wil Smithx
Originally by: Catelco Just a pointless though:
As the formula presented by the OP may be accurate (I don't know), I find that is too much complicated computationaly speaking (sqrt, exp...) to be calculated for every ship, every tick... May the real formula be more simplystic... or maybe CCP uses such kind of formulas with fixedpoint math or .... well who knows and who cares?
Anyways, nice research
Wouldn't be hard for a comp to do a calc like that would it?
While its somewhat more long winded to do on paper think of all the calcs eve has to do anyway, like tracking, momentum, stuff like that...
You have to realize that the single implementation of the calculation is almost a negligable expense of computing power. However, Eve consistantly has tens of thousands of ships in space that are doing this calculation (or some other calculation with similar behavior) along with countless others. Square Roots are less computationally expensive these days then they have been historically but it's still a math function you don't want to call for a constantly executed equation simply because of it's complexity. You'll start to realize how big of a pain some of math that people bandy about is once you take more advanced mathamatics. 
Gripen 
Posted  2008.08.11 13:30:00  [ 12]

Dr Corvax 
Posted  2008.08.11 14:11:00  [ 13]
Originally by: Gripen
Originally by: Dr Corvax
I have read a topic about research of the capacitor recharging http://myeve.eveonline.com/ingameboard.asp?a=topic&threadID=116993 As I can see, no final formula of capacitor recharging have been discovered.
You may have missed it but it was posted on second page. And it gives correct results but with your formula peak recharge is shifted (Actual peak is at 29.5%).
I have grabbed video to find peak of capacitor recharging. UI updating once per second and it is not enough to catch the peak. Speed of recharge is same at 25% and at 30%. Please tell me, how can I try to find peek moment more precisely? P.S.: Formula from previous topic use trigonometric. Such functions is also "hard" to compute. 
Akita T Caldari Navy Volunteer Task Force

Posted  2008.08.11 14:57:00  [ 14]
Originally by: Dr Corvax Please tell me, how can I try to find peek moment more precisely?
Big capacitor, slow recharge time. I'd recommend something like Dominix full of capacitor batteries and shield power relays Start/stop warps until cap depleted, wait for it to recharge. Might take a damned long while, but gives awesomely precise readings. 
Akita T Caldari Navy Volunteer Task Force

Posted  2008.08.11 15:04:00  [ 15]
Originally by: Dr Corvax Formula from previous topic use trigonometric. Such functions is also "hard" to compute.
You should look at the turret tohit and damage calculation formula. 0.5^(something), linear distance in 3d space, transversal (or, well, angular, whatever) speed in 3d space, call to random numbers, plenty of divisions and a couple of squares, etc. Now, imagine that's used EVERY TIME a turret makes a shot... with some ships that have 8 guns firing every 4 seconds, and I can bet they don't even cache distance/angular/sig and stuff at all from shot to shot... ouch. Compared to that, the cap recharge (which is calculated like, what, once a second, maybe 4 times per second tops) is trivial. If you were wondering why the game lags so badly... now you have some answers. 
Dr Corvax 
Posted  2008.08.11 15:51:00  [ 16]

Vessper Indicium Technologies Hephaestus Forge Alliance 
Posted  2008.08.11 19:54:00  [ 17]
This is some great work Corvax Your equation matches up to several sets of observed results perfectly (taking into account rounding errors from looking at rounded data in game!) 
Chequer Bones 
Posted  2008.08.11 21:12:00  [ 18]
They may be using table driven data, so say a lookup of 10000 values with 0.01% increments. Find the %capacitor remaining, look up the Delta.

Dr Corvax 
Posted  2008.08.12 10:33:00  [ 19]
Edited by: Dr Corvax on 13/08/2008 05:35:52Edited by: Dr Corvax on 12/08/2008 11:01:20I have made some tests to find peak of capacitor recharging. Raven with 5 Shield Power Relays and 5 Large Peroxide Capacitor Power Cells has capacitor capacity of 10695 units and capacitor recharge time 4142.8 seconds (with my skills). Here are the results.Some comments. I've write down capacitor's charge readouts for 270 seconds (from capacitor charge 17% to 33%). For each tick, we can easy find a capacitor's charge increment. Unfortunately, due to readouts rounding this is not enough to find peak. Nevertheless, we can easily find a moving average (I used 120 points) to isolate the peak. Finally, peak recharge rate is located on 25.2% of capacitor's charge. All of this is in excel file. P.S.: Mathematically, peak should be on 25% exactly. This can be found from formula (get the second time derivative C''(t) and find extremum). 
Dr Corvax 
Posted  2008.08.16 19:39:00  [ 20]
Edited by: Dr Corvax on 17/08/2008 08:46:23 Originally by: Gripen .... but with your formula peak recharge is shifted (Actual peak is at 29.5%).
Gripen, can you give some comments to my test. Am I right and peak is at 25% ? 