EVE Musings

 Posted by at 16:25  No Responses »
Aug 072009
 

I mentioned a few posts back that I was trying to solve a complex integral for damage calculations in EVE Online.  I was doing so in the interests of identifying the “best” overall ship for tackling PvE, specifically L4 agent missions.  My current ship (a T2-fitted Apocalypse) does a fine job of it and I’ve never had to warp out of a mission when I didn’t do something stupid to aggro the whole pocket.  However, if I can be using something better, I’d like to know it.

The problem with the approach I was using, as pointed out by Fraser, is that there are many, many additional factors beyond simple DPS.  In particular, the targets themselves play a big role.  Their size, velocity, angular velocity relative to your ship, and the size of your own weapons all play into the damage calculation (if you’re curious what the full formula is, check it out).  The Raven, for example, is an incredibly common PvE battleship because of how cheap it is to purchase and fast it is to train for at a basic level.  Its main weapons are cruise missiles, which are intended for large targets like battleships.  Consequently, they do less damage against small, nimble targets that can out-run their explosions.  As a result, Ravens often do well to fit target painters, which artificially inflate the apparent size of a ship.  Raw DPS calculations won’t account for this.

I’m not really sure how to resolve it.  I do think there’s an answer — and a generic one, at that — but I’m just not sure what it is.

Aug 062009
 

Been a while since my last post.  Since then, a ton of stuff has happened.

  • We fully moved-in to our house (though we’re still only about 25% unpacked).
  • We set a date for the wedding (8/7) and have picked a location for the reception, which may double as the wedding site too.
  • We selected/customized Cody’s engagement ring and matching wedding band
  • My company laid off about 25% of its work force (a layoff I rather miraculously was not a part of).
  • The Vampire game has resumed.

I’m probably forgetting a few things, but those are the big highlights.


One of the reasons I haven’t posted often of late is that it seems a bit of a chore to go to the blog page, log in, write up a post, etc, etc. I’ve recently implemented an easier method of posting that I think should make posting a more frequent occurrence. I’ve also got to get over my internal reluctance to post a battery of short posts as I think of things to say. If Twitter has proven anything, it’s that people enjoy hearing about the exploits of others in short bites. I don’t think I’ll ever hop on the Twitter bandwagon, though (famous last words…).


I finally got LaTeX-style rendering working on my wikis. It’s not that this is particularly difficult to do, but rather I had never had a server setup that would allow me to make the necessary changes to support it before. The particular implementation I’m using right now is MimeTeX. I had to do some custom hackery to make it work (specifically, my server did not seem content to create image links with some of the formatting required by TeX, so I wrote a PHP “middleman” that stands between the MediaWiki math engine and the MimeTeX CGI to properly handle formatting), but it’s great fun.


The major motivating factor in getting the TeX support to work is that I wanted to explore the idea of “damage potential” in EVE Online. Because of the way damage works in EVE, specifically with turrets, a given ship using a given type of gun is going to do the most damage at close range, and then see that damage falloff gradually as the target gets farther and farther away. This isn’t accounting for aspects of the target, which also play a role. The formula for this falloff is known and can be calculated, but I wanted to see how different ships stacked up to one another when they were compared.

I decided that the best way to do this would be to integrate the falloff curve (i.e. find the area bound by the DPS graph for the ship). Of course, this led to about 15 hours of wrestling with a truly atrocious integral. After consulting with Wolfram’s online integrator, engineers at work, the think tank at SDN, my dad, Cody, and Dr. Math,it became clear that the only way to solve the integral was via approximation and a computer.  I wrote up a Python script to do the integral and started getting good results.  I’m not really sure how valid they are, though.  Mathematically, they’re sound, but I’m not sure about their practical application.


I think that’s about it for now.