www.ClassicTW.com
http://www.classictw.com/

ZTM
http://www.classictw.com/viewtopic.php?f=13&t=11791		 Page 1 of 2
Author:  Davros [ Fri Oct 15, 2004 9:33 pm ]
Post subject: 
Hi All,

I'm curious how the ZTM macros work. Specifically, what optimizations do they use? If you have to do a ZTM of every sector to every other sector in a 5000 sector universe, thats 5000*5000 = 25000000 courses to check.

At about 1 pair per second, that would take something like 128 days to check all the possible courses.

How are the ZTM macros out there currently optimizing this? Do they find all the single-direction warps, or just bi-directional warps?

I tried searching around websites and these forums, but don't see it discussed anywhere.

Thanks!
Author:  ElderProphet [ Fri Oct 15, 2004 10:32 pm ]
Post subject: 
You're saying macros, but I think you mean scripts.

By plotting enough courses from arbitrary sector to arbitrary sector, you'll determine nearly all of the warps for every sector. The way that you determine whether a warp is 1-way is to compare warps out vs. warps in, then test any discrepancies. If a warp appears in both lists, then it's a 2-way warp. Let's say sectors 111, 222, and 333 have 2 warps as follows:

Sector Warps
111 222 333
222 111 999
333 123 456

We can see that it is a 2-way between 111 & 222, but we're unsure about 111 to 333. So your script must then test all one ways by attempting to plot the opposite course. In this example, you would plot from 333 to 111. If the course is 333->111, then it's a 2-way. Otherwise, it's a 1-way.

There are more complex ways to control a ZTM, but this covers the basics. Shout back if you want more info.

+EP+
Author:  Davros [ Sat Oct 16, 2004 3:10 am ]
Post subject: 
Thanks for the answer (and you're right, I mean script, thanks for the clarification).

But I'm still left with a lot of questions. I'm sure that will determine if a warp is 1-way, but it doesn't seem to do much for reducing all the sectors I need to check.

That is, if i do:

for 'from' = 1 to 5000
for 'to' = 1 to 5000
plot course and keep track of all the hops I discover
end
end

That's a really long loop.. So, how can you avoid any of the iterations? The only optimization I can think of is that I can stop if I've found already found 6 warps out of my 'from' sector. But many sectors (most) won't have that many warps out, so I never know when to stop. I pretty much have to check a route between it and every other sector. That takes a long time.....

So, how can I reduce the number of courses I need to plot?

Do the scripts that people use uncover a full map of the universe, or are they just uncovering 'most' of the map?

Thanks!
Author:  Harley Nuss [ Sat Oct 16, 2004 6:18 am ]
Post subject: 
Most ZTM scripts miss warps, the number missed depending on the depth of the scan. One thing to keep in mind, if you plot say sector 95 to sector 85 and get this:

quote:
95 > (14697) > (5855) > (10950) > (17594) > (7002) > (8117) > (3890) > (1406) >
(4265) > (15768) > (11757) > (85)


You now no longer have to plot 14697 to 5855, 5855 to 10950, etc. By optimizing your plots based on information recieved from previous plots, you will greatly reduce the total number you need to make.

The most popular current ztm method makes use of avoids in plotting courses. Basically, you make one pass finding at least one warp out from every sector. I do this by plotting one sector with no known warps to another sector with no known warps. Once you have at least one warp for every sector, you then do a second cycle in which you test every sector that has only 1 known warp (the first pass will give you multiple warps for some sectors as you criss-cross the universe). You place an avoid on the 1 know warp then plot from that sector to another sector in the universe. I tend to pick another possible 1-warp sector. If you plot out, you now have a second warp. If you don't plot out, you know it's likely a deadend, or at least part of a bubble. You then clear the avoid and plot the next sector. Once you have done all the 1-warps, you repeat for the 2-warps then 3-warps. At this point, in most universes, you will have caught most of the warps in the game, but you can continue for 4 and 5 warp sectors if you so choose. After this is done, you then need to run one more check for 1-way warps as EP descibed.

If you feel confused about any of this, please say what so someone else or myself can try to clarify.
Author:  Traitor [ Sat Oct 16, 2004 6:45 am ]
Post subject: 
quote:Originally posted by Davros

Thanks for the answer (and you're right, I mean script, thanks for the clarification).

But I'm still left with a lot of questions. I'm sure that will determine if a warp is 1-way, but it doesn't seem to do much for reducing all the sectors I need to check.

That is, if i do:

for 'from' = 1 to 5000
for 'to' = 1 to 5000
plot course and keep track of all the hops I discover
end
end

That's a really long loop.. So, how can you avoid any of the iterations? The only optimization I can think of is that I can stop if I've found already found 6 warps out of my 'from' sector. But many sectors (most) won't have that many warps out, so I never know when to stop. I pretty much have to check a route between it and every other sector. That takes a long time.....

So, how can I reduce the number of courses I need to plot?

Do the scripts that people use uncover a full map of the universe, or are they just uncovering 'most' of the map?

Thanks!

Dang you Harley, beat me too it!

Really the best way to understand how it works is to watch one in action. It's not unlike watching paint dry, but from that, you should be able to figure out how to write your own.

The Reverend has a good one for ZOC http://www.thereverend.org/tradewars/
SupG has a good one for TWX http://www.scripterstavern.com/ (but it seems to be down at the moment. sigh...)
Some of the helpers have a built in ZTMs too.
Author:  ElderProphet [ Sat Oct 16, 2004 4:01 pm ]
Post subject: 
The method that Harley describes is very time efficient, but maybe the following will help too. Since you want to learn the basics, let me explain how a simple, slow, very accurate ZTM works. This is essentially the method that Xide's 1_ZTM uses.

You pick a target sector, let's say Stardock in Sector 1000.
You plot from 1 to 1000, and say the warp data looks like this: 1 > 2 > 9 > 52 > 1000
You have just captured 1 warp for each of the first 4 hops. You then void the first hop, sector 2, and replot. You repeat this until no path is found. You now know all warps out of sector 1. Clear all voids.
Now plot from sector 2 to 1000, but start by voiding any known warps out of sector 2. Set voids as above until no path is found.
Clear voids and continue with sectors 3-5000. Just don't try to plot from 1000 to 1000 or you'll be asked "So what's the point?".
As before, you'll still have to check one-ways.

+EP+
Author:  Davros [ Sun Oct 17, 2004 4:07 pm ]
Post subject: 
Ah! Very cool, i hadn't thought of using avoids, that's exactly the sort of advice I was looking for. Thanks so much for the responses!
Author:  Slim Shady [ Sun Oct 17, 2004 4:12 pm ]
Post subject: 
i have a ZTM macro, i dunno what you guys are talkin about. i hit F1 and i have a 20k universe in under a minute. (100%) i can't tell you how, cause its part of my "super game-ending, uber-leet, auto-afk-kill, do-everything-in-one, ppt/pdrop at the same time, unstoppable-gridder/and unreal photoner in one, make newbies run away in fear" script pack. (yes, super macros come with my script pack) :P
sigh.. im tired. davros, ignore what i say, listen to them. they are all smart guys with far less sarcasm than such as myself. also they happen to know a ton about the game.
ill throw in a real .02 though, i think traitor said it, fire up a ZTM and watch it go. you'll watch, you'll understand it to start. then a bit later you'll get confused and then you come ask questions to these types of people and in no time you'll be learning alot :)

Slim

mail check or money order for slim's "super game-ending, uber-leet, auto-afk-kill, do-everything-in-one, ppt/pdrop at the same time, unstoppable-gridder/and unreal photoner in one, make newbies run away in fear" script pack to... well me of course :P

heh
Author:  Speed Demon [ Mon Oct 18, 2004 2:33 pm ]
Post subject: 
Slim I sent you my $19.95 payment over a month ago when are you going to send me the uber-leet script pack you sold to me? It seems you sent it out to everyone else according to all the complaints I have been hearing lately.

No really this is interesting stuff and its good to know but I really hope you not trying to write a ztm script there are so many really good ones out there you can use I would put my time into writing some other type of script. Maybe you can help Slim finish up his uber-leet script pack.
Author:  Slim Shady [ Thu Oct 21, 2004 3:13 am ]
Post subject: 
heh, the crap that flows out of my fingers when i am tired :)

Slim
Author:  Promethius [ Thu Oct 21, 2004 7:36 pm ]
Post subject: 
quote:Originally posted by Slim Shady

heh, the crap that flows out of my fingers when i am tired :)

Slim


Then we must assume that you never get on the computer and post until you are tired.... ;)
Author:  Slim Shady [ Thu Oct 21, 2004 9:36 pm ]
Post subject: 
that would be a good assumption

Slim
Author:  ElderProphet [ Fri Oct 22, 2004 12:03 am ]
Post subject: 
I actually enjoy playing with ZTM scripts. Don't think for a second that the time a ZTM routine takes to complete translates into accuracy. Some of the slowest have terrible accuracy, while some of the fastest miss a tiny percentage of warps. I ran a well known ZTM on a 5k stock bang that took 3 Hrs, 28 mins and missed 28 warps. I then ran a fairly simple ZTM I created, on the same universe, that took 1 Hr, 35 mins and missed zero warps. So don't make the assumption that popular or readily available ZTM routines are all good. Test them.

How about you? Would you rather use a 45 min routine that missed 30 warps, or a 2 Hr routine that missed none?

+EP+
Author:  Harley Nuss [ Sat Oct 23, 2004 3:47 am ]
Post subject: 
I would contend you can't make a 2 hour routine that consistently missed 0 warps. You pretty much have to plot every sector to every sector to guarantee 100% accuracy. As far as public ztm routines go for accuracy, I would stick with either supg's ztm or rev's ztm. All others I've seen (including swath, ck's, twx's built in) miss quite a few warps.
Author:  ElderProphet [ Sat Oct 23, 2004 6:30 pm ]
Post subject: 
I was hoping I'd found a method that consistently missed zero warps, so I banged a 20K universe, 100 Gold bubbles, 100-199 sectors each. My routine took 7 Hrs and change, missed 27 warps out of 53,765. Good, but not perfect.

I agree absolutely about SupG's and Rev's, and they still take half the time.

+EP+
Page 1 of 2 All times are UTC - 5 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/