Shiva/Versions

Versions of Shiva

2.1

• Score : ????
• maintaining 2.0
• bugfix

2.0

• Score : 2018
• maintaining 2.0beta2 -12
• gunning:
  • now tries to fool the adaptive movement of its enemy
  • save shooting at the end of the round
  • some other (little) tweaks and bugfixes
• changes its color from red to black as the blood of its victim does.

2.0beta2

• Score : 2029
• maintaining 2.0beta (1) +23
• firePower is 1.9

2.0beta

• Score : 2006 RECALLED
• maintaining 1.10.3 -14
• more guess factor segments
• prepopulated movement stats
• default firepower is 2.1

1.12.1

• Score : 2002 RECALLED
• maintaining 1.10.3 -18
• now fires bullets with power 3 when enemy is closer than 320
• default firepower set to 2.1
• expected to be better against the weak and weaker against good bots

1.12

• Score : 2021
• maintaining 1.10.3 +1
• tweaked segmentation of my gun

1.11.1

• experimantal version
• Score: 2004 RECALLED
• maintaining 1.10.3 -17

1.10.3

• Score: 2020
• maintaining 1.10.2 +10
• again tweaks

1.10.2

• Score: 2010
• maintaining 1.1 +3
• tweaks

1.1

• Score: 2007 2K!!!!!!!!!
• maintaining 1.02 +18
• bugfix
• new gun

1.02

• Score: 1989
• maintaining 1.0 +4
• runs faster
• new EnemyWave segmentation

1.0

• Score: 1985
• maintaining 0.91 +5
• added velocity segment to EnemyWave segmentation

0.91

• Score: 1980
• maintaining 0.9 +10
• new weightning of segmentet EnemyWave stats

0.9

• Score: 1970
• maintaining 0.8 -7
• big bugfix in dodging

0.8

• Score: 1977
• maintaining 0.70.1 +30
• bugfix
• changed weightning

0.70.1

• Score: 1947
• maintaining 0.7 -1
• removed victory dance

0.7

• Score: 1948
• maintaining 0.5 +45
• FOUND GIANT BUG in bin smothing

0.6.X ... some broken. RECALLED

0.5

• Score: 1903 (>1900 !!!!)
• maintaining 0.41.1 +26
• anti ramming methode

0.41.1

• Score: 1877
• maintaining 0.41 +3
• new fire-power-calculation

0.41 Score: 1874

• maintaining 0.2 +16
• now surfes wave that will hit next and not wave that is nearest.

0.4

• not public
• tweak

0.3

• not public
• bugfix and tweak

0.2

• Score: 1858
• first release

---

I think it's a mistake to fire power 3 in the range 150 -> 320. It'll lower your accuracy too much. -- PEZ

You ment in the range of 0 -> 320 didnt you?? Anyway i am not sure how it will affect my rankings, but with my current segmentation i had to take 320 if i wanted it to increase from 160. (my fast learing array segments the distance like this : distance/160 :/ )--deathcon

I meant from 150 -> 320. Just guessing. Not too far from 160. =) I'm pretty sure you'll lose rating points by that change. But we'll see, won't we? -- PEZ

Once again you were right. Want to compete with rozu - the robocode-oracle ??? :D --deathcon

Nah! But you lost about as many points as I gained when going the other direction. Among everything fuzzy in Robocoding this is now non-fuzzy I'd say. Power 2 is better than 3 for all ranges but the very closest. Good to know. -- PEZ

Hey! Does that mean that you gained 23 points just by using firepower 1.9 instead of 2.1? No other changes? -- PEZ

That's it. --deathcon

Amazing. I know I have experimented a lot and at the moment CC uses firepower 1.91. =) But I wouldn't have guessed that it could make such a difference going from 2.1 to 1.9. You know my theory says power 1.7x should be best. But I haven't managed to prove that in real points. This is another nice little PerfomanceMystery?. -- PEZ

I think, that you cant say what firepower is BEST in general. You always have to find a good firepower that fits to your prefered fighting distance AND that of your enemy. Of cause it depends on the strenght of your enemy and your hit rate as well. So once more TESTING,TESTING,TESTING ... :D --deathcon

Well, everybody who really test it seems to end up on 1.9. -- PEZ

You may be right that 1.9 fits to many current top-bots in the current rumble (current opponents) - many bots we have there dont mind about their fighting distance, so top bots can keep their 500-600. They also have good hit rates so 1.9 seems ok. But if your bots fights closer to your enemy 2.3 would take you higher one the ranking table.

Uploaded an updates ShivaRRGC to see how the change of guess factor segments from version 1.10.3 to 2.0beta2 affected the result.

--deathcon