In one of the discussions on ZTMs and the need for them, SWATH was given about a 97-98% accuracy rating and SupG's ZTM slightly higher. The question I have is how would a person determine the accuracy of a ZTM script? I have written a ZTM script that would appear to give a very high accuracy, but I am guessing because I am not sure how to validate it. If anyone has any ideas for a validation method, I would appreciate it.

Also, what is acceptable performance for a 5k ztm with a 200ms ping? I usually run ztm after I have ran my moves down and have never really timed a ztm.

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

Im just taking a guess but I would think that you could compare your results to the data in t-edit or somewhere in TWGS, I would think that that info is probably displayed somewhere in TWGS?

You simply need to compare warp data as recorded in TWX before and after a 100% probe or explore, or info from t-edit.

+EP+

have 2 databases for same game. on one make observer account and use CIM to get all the warps. in the other, do your ztm. observer will be 100%, see how much yours differs.

Slim

You know, I've never really thought about how many missed warps make up 1%. If the average sector has 4 warps, then a ztm routine could miss 199 warps in a 5K sector game and still be over 99% accurate. That's pretty scary, especially when consider how many warps are missed if it is 97% accurate.

+EP+

I run a ztm on dialup in a 5k sector game in about 80-90 minutes, I think. I normally miss 4-10 warps total.

k, thanks to everyone that has replied. The method I used did not work near as well as I thought it would, so it is back to working on the script. The time required for a 5k sector game was close to 2 hours and it appears that I need to work on both the speed and accuracy. Until I can improve both, I will stick with SupG's script.

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

That is very high accuracy. I use twassist, which will map 5K in less than an hour, but last time I ran it against an observer CIM it missed 88 warps out of a game total of 13189 warps.

quote:Originally posted by Harley Nuss (teamEIS)

I run a ztm on dialup in a 5k sector game in about 80-90 minutes, I think. I normally miss 4-10 warps total.

quote:Originally posted by ElderProphet

You simply need to compare warp data as recorded in TWX before and after a 100% probe or explore, or info from t-edit.

+EP+


I ran the ztm in a 5k sector game and recorded the warp data - then eprobed 100% as you suggested. The file compare resulted in 6 missed sectors in the warp data for the adjacent sectors. The time frame is the next thing I have to look at to improve speed.

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

Nice. Knowing nothing about your routine, let me offer two pieces of advice. First, use the TWX database to track the warp data. No sense in your script and the database both keeping track of warps, plus it will decrease the code and complexity of your script... probably significantly. Second, keep the commands queued on the server. This will keep the delay between plots to .5 seconds. If you send a plot, await the response, then send the next plot... then each plot will take about .7 seconds or more. Queuing the commands will also allow dial-up users to ztm as fast as low-latency connected users.

Now, having said all of that... unless you've come up with some very clever ztm method, you are probably reinventing the wheel. SupG's public routine is roughly based on Rev's technique, which is extremely efficient insofar as ztm methods go. You aren't likely to come up with a better combination of speed and accuracy, so you should focus on one or the other. So find out why you missed those 6 adjacents, and try and come up with a technique that will catch even those. Many players would run a ztm that took a bit longer if it were more accurate.

+EP+

quote:Originally posted by ElderProphet

Nice. Knowing nothing about your routine, let me offer two pieces of advice. First, use the TWX database to track the warp data. No sense in your script and the database both keeping track of warps, plus it will decrease the code and complexity of your script... probably significantly. Second, keep the commands queued on the server. This will keep the delay between plots to .5 seconds. If you send a plot, await the response, then send the next plot... then each plot will take about .7 seconds or more. Queuing the commands will also allow dial-up users to ztm as fast as low-latency connected users.

Now, having said all of that... unless you've come up with some very clever ztm method, you are probably reinventing the wheel. SupG's public routine is roughly based on Rev's technique, which is extremely efficient insofar as ztm methods go. You aren't likely to come up with a better combination of speed and accuracy, so you should focus on one or the other. So find out why you missed those 6 adjacents, and try and come up with a technique that will catch even those. Many players would run a ztm that took a bit longer if it were more accurate.

+EP+


Thanks for the reply and the information on queing the commands. I am using the TWX database to track the warp data as that seems to be the easiest way to do the script.

You are most likely right about reinventing the wheel, but I started writing it mostly as a challenge to get one to work - writing scripts is part of the fun of Twars to me. I do like SupG's script since it was the best one I have found. I know his script is a lot more complicated by looking at the code, and probably therefore more complete in some areas, but I am hoping to obtain some data at the time of the ZTM that I usually have to wait and get later.

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

While we are on the topic of ZTM. It was my understanding that TWX captures all data that it sees. Therefore i made the assumption that a ZTM run from Zoc (Rev's Script) would act as a ZTM for TWX as well. This way killing 2 birds with one stone. A ZTM for all of my REXX scripts and at the same time grabbing a warp data for TWX.
Now on to the problem. I ran a ZTM (Rev's Script) on a 5k universe (TEST GAME), I then proceeded to probe 100% of the univerese. What i found was that the warp data for TWX is incorrect?
ie: calculating a warp path internally on the TWX database results in an incorrect path?
Any ideas?

Help
Zoso

Are you saying you found the data inconsistent while trying to probe, or after? I've seen times when the route reported by TWX was different than the route reported from a CIM query, but it was always due to incomplete warp data, or a corrupt database. If you have 100% explored, try hitting ^I to ensure TWX has the full warp data, then see if TWX's routes match those of a CIM route query. If they do not, try creating a new TWX database and retesting. Also, while TWX captures much of what it sees, it does not capture all of it.

The technique (breadth-first search) used by TWX's getCourse is the same technique used by TradeWars for course plotting, so if the data captured from the ZTM is accurate, then the warp paths will be identical.

+EP+

Yes, What i'm saying is that after probing 100% (with a full ZTM), if i plot a course ($,D,C) the results are not only wrong but do not match the (C)omputer, (F)Plot Course from Zoc.
So it would appear that the Warp Data is not complete. Although i am still unclear on the ZTM question, whether a ZTM script run in REXX will allow TWX to capture an accurate warpdata? Or do i require to run 2 seperate ZTM's? One REXX and one TWX?

Zoso

quote:Originally posted by Zoso

Yes, What i'm saying is that after probing 100% (with a full ZTM), if i plot a course ($,D,C) the results are not only wrong but do not match the (C)omputer, (F)Plot Course from Zoc.
So it would appear that the Warp Data is not complete. Although i am still unclear on the ZTM question, whether a ZTM script run in REXX will allow TWX to capture an accurate warpdata? Or do i require to run 2 seperate ZTM's? One REXX and one TWX?

Zoso


Zoso,
If you are running ZOC and connecting to TWX, then you don't need to run the ztm twice. However, after probing, you should pull down a CIM (^i) like EP suggested to pull in all the warp data. That should get you all the warps. I have discovered that after probing the universe, you need to run a CIM, since twx only stores the warp data that it sees. (the probes don't go through every possible warp.)