[Patch 3.05.1: Karma Remake] General Discussion - Page 11

On March 29 2013 02:28 emperorchampion wrote:


Here are my thoughts on how to best approach this problem. It could be an interesting thought exercise, I kind of doubt it would yield any real meaningful results though.

This would work by calculating a bunch of functions (g, p, and m, ...) before hand then optimizing based off of them.

Variables
x1= health ; x2= armour ; x3= mr
g= estimated gold at time = t ; p= estimated phys damage at time = t ; m= estimated magic dmg at time = t

Objective Function
Maximize: Z= x1+ x2+ x3

Constraints
2.6x1+ 18x2+ 18x3 <= g ;total expenses less than total gold
(some linear function involving ehp vs physical) >= p ;make sure will survive physical damage
(some linear function involving ehp vs magic) >= m ; make sure will survive magic damage
...

Some thoughts:
-make x1, x2, and x3 all fractions ie, you should spend x1% of gold on health -- as it is now it's just spend x1 gold on health
-make p and m fractions of total damage; total damage would be a function of t, and depend on what you're playing
-g could be a function of t, that would involve some rate that you could calculate yourself based on: position (solo top, solo mid, ect), passive gold gain, last hitting ability, ect

-This would change what the optimal amount spent for each player, each role, and champ would be. So a pro player who might have a much higher gold value at a given time would get different stats

-a lot of functions would probably be piecewise, could account for power swings ie, if you could find that death cap was bought at a certain point in time on average there would be a spike in the m function at that point

-can add constraints based on regens

-maybe make it overall item buys, would be position specific by saying I want y% of damage stats and (1-y)% defense stats or something of the like, could account for different players and different levels of skill

-maybe maximize ehp?

Anyhow, I could go on. This is a very basic form that would obviously need a lot of work and research (research would need data, ie Riot data or parsable replays or something). As I said it probably wouldn't yield any interesting results because it's pretty intuitive anyhow, plus there are so many variables to consider that it would have to be calculated on the fly. Either that or you'd have to calculate a whole bunch of scenarios before hand. Maybe it could be useful as some sort of post game analysis, I dunno.

Servers pls come back!

On March 29 2013 02:58 sylverfyre wrote:
2.6 H, 18A, and 18M all imply that you're buying basic items to achieve your maximization of EHP, which ALSO isn't realistic, since combined items have better gold efficiency than that (for the most part)

On March 29 2013 02:53 thenexusp wrote:


Wrinkles in your plan:

- the equations for ehp are inherently nonlinear, due to having to multiply health and armor together.
- Your objective function seems wrong, you're generally not trying to maximize the sum of your health, armor and MR (in this case, just buy health! It's 7 times cheaper)
- You can't simply say you want to survive the physical damage and the magic damage, you have to be able to survive the sum of both. (e.g. if you have 2000 physical ehp and 1000 magic ehp, taking 1500 physical and 300 magic will still kill you)

There are several ways to frame the problem:
1) Given you will take X physical and Y magic damage, what is the cheapest way to buy health, armor, and MR to survive (burst damage)
2) Given you will take X% physical and (100-X)% magic damage, and given a budget G, how much health, armor and MR do you buy to survive the most damage before dying. (dps)
3) Given you will take X_p physical damage upfront and X_m magic damage upfront and then Y_p physical damage per second and Y_m magic damage per second, and a budget G, how much health, armor and MR do you buy to survive the longest time (maybe a more accurate representation of how real battles go)

For case (1), the equation looks like this:
h is health, a is armor, m is MR, h_0 is base health, a_0 is base armor, m_0 is base armor, X is incoming physical damage, Y is incoming magic damage

minimize 2.6h + 18a + 18m
given
X*100/(100+a_0+a) + Y*100/(100+m_0+m) < h_0 + h

On March 29 2013 03:44 Frolossus wrote:
"LoL Status ‏@lolstatus 1m

We have potentially found the root cause and are working on a solution that will allow us to deploy patch 3.5 as planned."
https://twitter.com/LoLStatus

does this mean that they are going through with the patch instead of reverting it and thus the servers will be down for even longer than they already have been? :/

On March 28 2013 23:03 Cedstick wrote:
fuk riot.

But for reals, I have nothing against the new art for Trundle, but I don't like how he's a bad dude now :< I like Good Guy Trundle.

On March 29 2013 03:49 WaveofShadow wrote:
Up to you guys if you want to bother but this post just popped up on Reddit and apparently despite all the work Neo + crew have put in we still go unrecognized.

http://www.reddit.com/r/leagueoflegends/comments/1b6lxe/some_useful_lol_websites_that_everyone_should/c945o3d

On March 29 2013 03:59 AsmodeusXI wrote:


Thank you for giving me a good opportunity to toot my own horn. =D

+ Show Spoiler +

On March 29 2013 04:02 WaveofShadow wrote:

Are you 'pissfacebukkakekilla?'

On March 29 2013 04:08 Dusty wrote:


LOL

but seriously, do we really want the average redditor to come to this forum? =o

On March 29 2013 03:49 WaveofShadow wrote:
Up to you guys if you want to bother but this post just popped up on Reddit and apparently despite all the work Neo + crew have put in we still go unrecognized.

http://www.reddit.com/r/leagueoflegends/comments/1b6lxe/some_useful_lol_websites_that_everyone_should/c945o3d

On March 29 2013 04:09 NeoIllusions wrote:

Sure, why not?