GresSuffurd/Version History

From RoboWiki
< GresSuffurd(Difference between revisions)
Jump to: navigation, search
(Version history of GresSuffurd: 0.3.25 stats, 0.3.26 explanation)
(0.3.26 stats)
 
Line 7: Line 7:
 
** Simplification (partly rewrite) so new features can be added more easily.
 
** Simplification (partly rewrite) so new features can be added more easily.
 
*'''0.3.26''' (20171211) gun: '''GV 0.3.3''' move: '''WS 0.3.18'''
 
*'''0.3.26''' (20171211) gun: '''GV 0.3.3''' move: '''WS 0.3.18'''
** '''APS: tbd (xxth), PL: tbd (xxth), Survival: tbd'''
+
** '''{{RumbleStatsDefault|link=http://literumble.appspot.com/BotDetails?game=roborumble&name=gh.GresSuffurd%200.3.26|rumble=RoboRumble|scorelabel=APS|score=86.3|rank=13th|win=1128|loss=29|plrank=24th|glicko2=N/A|pwin=97.49|vote=1.73|anpp=90.65|score2label=Survival|score2=92.69}}'''
 
** Multidimensional dangerarray instead of accumulated single arrays, only against good bots
 
** Multidimensional dangerarray instead of accumulated single arrays, only against good bots
 
*'''0.3.25''' (20171123) gun: '''GV 0.3.3''' move: '''WS 0.3.17'''
 
*'''0.3.25''' (20171123) gun: '''GV 0.3.3''' move: '''WS 0.3.17'''

Latest revision as of 12:41, 25 February 2018

[edit] Version history of GresSuffurd

  • 0.4.0 (201710xx) gun: GV 0.4.0 move: WS 0.4.0
    • APS: tbd (xxth), PL: tbd (xxth), Survival: tbd
    • Simplification (partly rewrite) so new features can be added more easily.
  • 0.3.26 (20171211) gun: GV 0.3.3 move: WS 0.3.18
    • RoboRumble ‒ APS: 86.3% (13th), PL: 1128-29 (24th), Survival: 92.69%
    • Multidimensional dangerarray instead of accumulated single arrays, only against good bots
  • 0.3.25 (20171123) gun: GV 0.3.3 move: WS 0.3.17
    • RoboRumble ‒ APS: 86.35% (12th), PL: 1126-28 (23rd), Survival: 92.78%
    • Take surfingdanger on t+1, to avoid edgeshots
    • Only better in PL it seems
  • 0.3.24 (20171118) gun: GV 0.3.3 move: WS 0.3.16
    • RoboRumble ‒ APS: 86.26% (13th), PL: 1120-32 (28th), Survival: 92.68%
    • back to 0.3.20. Replace virtual 'real danger' waves with virtual HOT waves.
    • The '70% against every rammer' release. +21% total against all 20 rammers => 0.02% APS gain
  • 0.3.23 (20171027) gun: GV 0.3.3 move: WS 0.3.15
    • RoboRumble ‒ APS: 86% (16th), PL: 1114-34 (33rd), Survival: 92.45%
    • Increase number of surfing bins from 37 to 137 (part 3). Now it should be better than 0.3.20.
    • Not . . . so back to the drawing board
  • 0.3.22 (20171027) gun: GV 0.3.3 move: WS 0.3.14
    • RoboRumble ‒ APS: 86.1% (14th), PL: 1116-35 (31st), Survival: 92.6%
    • Count every would-be hitted bin instead of fixed 5 bins piramidshaped (part 2)
  • 0.3.21 (20171026) gun: GV 0.3.3 move: WS 0.3.13
    • RoboRumble ‒ APS: 86.29% (12th), PL: 1117-34 (31st), Survival: 92.7%
    • Calculate surfingdanger over botwidth bins instead of fixed 5 bins piramidshaped (part 1)
  • 0.3.20 (20171021) gun: GV 0.3.3 move: WS 0.3.12
    • RoboRumble ‒ APS: 86.32% (13th), PL: 1120-32 (29th), Survival: 92.79%
    • Improve against non-firing bots, pea-shooting bots and extends Robot, also in endgame.
    • On par with 0.3.19. Crushing dragonbyte.Neutrino and better against cx.mini.Cigaret, but also some lesser side-effects.
  • 0.3.19 (20171015) gun: GV 0.3.3 move: WS 0.3.11
    • APS: 86.27% (11th), PL: 1130-29 (25th), Survival: 92.69%
    • Preferred distance 495, even better, good point to go on with real changes.
  • 0.3.18 (20171014) gun: GV 0.3.3 move: WS 0.3.11
    • APS: 79.41% (68th), PL: 1066-94 (81st), Survival: 86.89%
    • Preferred distance 180, but indeed diving in way to steep.
  • 0.3.17 (20171015) gun: GV 0.3.3 move: WS 0.3.11
    • APS: 86.1% (14th), PL: 1123-37 (32nd), Survival: 92.62%
    • Preferred distance 450, better than expected.
  • 0.3.16 (20171014) gun: GV 0.3.3 move: WS 0.3.11
    • APS: 85.8% (16th), PL: 1100-41 (53rd), Survival: 92.44%
    • Preferred distance 360, quite some interesing results . . .
  • 0.3.15 (20171014) gun: GV 0.3.3 move: WS 0.3.11
    • APS: 84.71% (23rd), PL: 1104-56 (49th), Survival: 91.66%
    • Half the preferred distance (so 270), see what an eventual dynamic distancing can do.
    • Too close, 1.2 APS loss. But maybe something changed since 2009 with 740 bots.
  • 0.3.14 (20171004) stats after 5 years . . .
    • RoboRumble ‒ APS: 85.9% (15th), PL: 1127-32 (28th), Survival: 92.53%

[edit] Old rankings

  • 0.3.14 (20121224) gun: GV 0.3.3 move: WS 0.3.10
    • APS: 85.57% (8th), PL: 945-21 (19th), Survival: 92.59%
    • calc acceleration on right tick, append with constant for xx ticks.
    • Nice Christmas present: 0.6 APS just for adding acceleration to my movementprofile.
  • 0.3.13 (20121221) gun: GV 0.3.3 move: WS 0.3.9
    • APS: 85.22% (8th), PL: 940-26 (22nd), Survival: 92.25%
    • add extra uni-dimensional array about acceleration to movement
  • 0.3.12 (20121007) gun: GV 0.3.3 move: WS 0.3.8
    • APS: 84.82% (11th), PL: 920-27 (23rd), Survival: 91.8% (0.15 APS higher than 0.3.11)
    • use bulletlist to choose bulletpower (due to remark from Skilgannon)
    • NO adaption of bulletpower when enemy has more energy left.
  • 0.3.11 (20111029) gun: GV 0.3.2 move: WS 0.3.8
    • APS: 84.06% (10th), PL: 814-24 (21st), Survival: 91.24% = on par with 0.3.8
    • back to 0.3.9, bugfixing danger calculation (was 0.5*(bin-1), 1.5*(bin), duh)
    • dangercalculation now 0.2*(bin-2), 0.6*(bin-1), 1.0*(bin) etc. iso 0.5*(bin-1), 1.0*(bin), 0.5*(bin+1)
  • 0.3.10 (20111018) gun: GV 0.3.2 move: WS 0.3.7
    • APS: 83.83% (14th), PL: 812-22 (19th), Survival: 91.27%
    • use BulletHitBullet info to go to the created 'safe spot' (although whole wave is safe). Small step before implementing Bullet Shadows
  • 0.3.9 (20111015) gun: GV 0.3.2 move: WS 0.3.6
    • APS: 83.92% (11th), PL: 805-28 (27th), Survival: 91.27%
    • normalise surfingdanger with average danger for that wave (necessary for 0.3.10)
    • surfingdanger dependent on bulletpower (just because I think it is logical)
  • 0.3.8 (20110922) gun: GV 0.3.2 move: WS 0.3.3
    • APS: 83.95% (9th), PL: 805-24 (20th), Survival: 91.24%
    • back to 0.3.5, for normal gun, increase influence of non-bullet waves
    • for decaying gun, decrease influence of non-bullet waves
  • 0.3.7 (20110813) gun: GV 0.3.1 move: WS 0.3.5
    • APS: 83.25% (12th), PL: 780-36 (35th), Survival: 90.34%
    • use 'classic' binsmoothing (1/sqr(bindiff)) iso fixed small pyramid (0.2, 0.6, 1.0, 0.6, 0.2)
  • 0.3.6 (20110813) gun: GV 0.3.1 move: WS 0.3.4
    • withdrawn
    • use 'classic' binsmoothing, but due to auto-cast to int, it recorded only the single hitted bin
  • 0.3.5 (20110710) gun: GV 0.3.1 move: WS 0.3.3
    • APS: 84.1% (8th), PL: 789-25 (23rd), Survival: 91.48%
    • back to 0.3.2
    • widen surfingdanger from 1 bin to (bin + (((bin-1)+(bin+1))/2)) (small pyramid)
  • 0.3.2b (20110708) gun: GV 0.3.1 move: WS 0.3.0
    • APS: 83.82% (11th), PL: 791-24 (20th), Survival: 91.28%
    • re-introduce due to switch to Robocode 1.7.3.0
  • 0.3.4 (20110612) gun: GV 0.3.1 move: WS 0.3.2
    • APS: 83.63% (11th), PL: 773-31 (30th), Survival: 90.83%
    • re-introduce binfactor to counter possible problems at extreme MEA (MEA * 1.07)
  • 0.3.3 (20110610) gun: GV 0.3.1 move: WS 0.3.1
    • APS: 83.55% (11th), PL: 770-36 (34th), Survival: 90.73%
    • increase bins from 37 to 73
    • when hit, count every position (bin) that would have been hit fully (iso 0.2, 0.6, 1.0, 0.6, 0.2)
    • when calculating risks, take the accumulated risks of botwidth iso just 1 bin
      • ouch, did it became to predictable or are there any bugs??
  • 0.3.2 (20110522) gun: GV 0.3.1 move: WS 0.3.0
    • APS: 83.99% (9th), PL: 783-21 (16th), Survival: 91.57%
    • when enemy is killed, assume it fired if it was able to
    • more consistent behaviour if no real enemywave is present
  • 0.3.1 (20110504) gun: GV 0.3.1 move: WS 0.2.10
    • APS: 83.95% (9th), PL: 782-21 (18th), Survival: 91.56%
    • revert ignore of bullet-hit-bullet in gun (what was I thinking . . .)
    • movement graphics inspired by (and peeked at) voidious.Diamond
    • move at minimum angle (11.6 degrees) at very close ranges
  • 0.3.0 (20110426) gun: GV 0.3.0 move: WS 0.2.9
    • APS: 83.75% (11th), PL: 776-30 (28th), Survival: 91.15%
    • general cleanup
    • add cornersegment to wallattribute
    • some bulletpower/energymanagement issues
    • bugfix/improve precise MEA in close range/nearwall situations
    • mark bullet-hit-bullet waves as non-bullet waves
      • Comment: to much changes in one time, now I have to check which change(s) caused the loss in APS
  • 0.2.29 (20110412) gun: GV 0.2.20 move: WS 0.2.9
    • APS: 83.84% (10th), PL: 778-21 (16th), Survival: 91.38%
    • Replace 'distance' attribute with 'absolute lateral velocity' attribute in gun
  • 0.2.28 (20110315) gun: GV 0.2.19 move: WS 0.2.9
    • APS: 83.64% (11th), PL: 770-26 (25th), Survival: 91.3%
    • Use estimated bearing next tick i.s.o bearing this tick when aiming gun (lining up bullets with virtual guns)
    • I feared for a 'performance enhancing bug', but it turned out better than expected
  • 0.2.27 (20110228) gun: GV 0.2.18 move: WS 0.2.9
    • APS: 83.42% (11th), PL: 760-34 (33rd), Survival: 90.92%
    • Simulate precise MEA in gun
    • re-enabled the BulletCatcher tweak'
  • 0.2.26 (20110131) gun: GV 0.2.17 move: WS 0.2.9
    • APS: 83.15% (11th), PL: 754-30 (29th), Survival: 90.6%
    • Use Precise Intersection in the gun
    • small tweaks f.e. against stop-and-go
    • accidentally disabled the 'BulletCatcher tweak'
  • 0.2.25 (20110126) gun: GV 0.2.16 move: WS 0.2.9
    • APS: 83.05% (14th), PL: 749-35 (33rd), Survival: 90.56%
    • Updated ancient gungraphics, found and fixed missing 'killing bullet' processing
  • 0.2.24 (20110125) gun: GV 0.2.15 move: WS 0.2.9
    • APS: 82.92% (14th), PL: 754-32 (31st), Survival: 90.47%
    • Adding second gun, just the same gun but with some decay. Good enough for 11th place ??
      • Comment: less improvement than I hoped and expected, maybe some 'bad luck' battles?
  • 0.2.23 (20110115) gun: GV 0.2.14 move: WS 0.2.9
    • APS: 82.84% (15th), PL: 751-33 (33rd), Survival: 90.44%
    • further BasicSurfer bug exploitation, energy management improvements
  • 0.2.22 (20100414) gun: GV 0.2.13 move: WS 0.2.9
    • APS: 82.8% (15th), PL: 729-32 (33rd), Survival: 89.69%
    • version 0.2.20 firing 1.95 power bullets (BasicSurfer bug exploitation)
  • 0.2.21 (20100112) gun: GV 0.2.12 move: WS 0.2.10
    • APS: 81.86% (22nd), PL: 714-46 (42nd), Survival: 88.72%
    • remove wallsegment in surfing once again, only abs velocity is left.
  • 0.2.20 (20091026) gun: GV 0.2.12 move: WS 0.2.9
    • APS: 82.09% (18th), PL: 707-34 (34th), Survival: 89.27%
    • increase bins in gun from 37 to 73 for accuracy, check at least [botwidth] bins.
    • Comment: accuracy means better against weak, worse against topbots
  • 0.2.19 (20091017) gun: GV 0.2.10 move: WS 0.2.9
    • APS: 82.01% (18th), PL: 707-33 (33rd), Survival: 89.14%
    • back to 0.2.17
    • increase surfingdistance from 480 to 540
    • Comment: only marginal better, but I'll keep it
  • 0.2.18 (20091005) gun: GV 0.2.11 move: WS 0.2.8
    • APS: 81.44% (23rd), PL: 709-38 (28th), Survival: 88.71%
    • half the influence for non-firing waves
    • Comment: wrong bet
  • 0.2.17 (20090925) gun: GV 0.2.10 move: WS 0.2.8
    • APS: 81.88% (20th), PL: 707-34 (33rd), Survival: 89.3%
    • no energysaving against rammers
    • use multiple unidimensional surfbuffers instead of one multidimensional.
    • Comment: first release to enter the top-20 !
  • 0.2.16 (20090919) gun: GV 0.2.9 move: WS 0.2.7
    • APS: 79.61 (37th) Surv: 87.99 PL: 708-32
    • tweaking wallsegment: set treshold a lot higher, but try keep PL score.
  • 0.2.15 (20090917) gun: GV 0.2.9 move: WS 0.2.6
    • APS: 79.58 (38th) Surv: 87.88 PL: 707-34
    • tweaking wallsegment: only use it when treshold per round is met. Should help against simpler targetters.
  • 0.2.14 (20090916) gun: GV 0.2.9 move: WS 0.2.5
    • APS: 81.52 (27th) Surv: 89.30 PL: 703-37
    • adding wallsegment to surfing (third time), now with same calc as in gun
  • 0.2.13 (20090914) gun: GV 0.2.9 move: WS 0.2.4
    • APS: 81.84 (21st) Surv: 89.04 PL: 701-39
    • reduce the inprecision of the fireangle to +/- 0.6 pixel
    • change the wallsegment calc from 'linear extends current direction' to 'orbital from my position'.
  • 0.2.12 (20090905) gun: GV 0.2.8 move: WS 0.2.4
    • APS: 80.73 (32th) Surv: 87.58 PL: 686-53
    • bugfix in bulletdetection. Also sees bullets from accelerating bots that suddenly brake (thx BulletCatcher)
    • add a small random inprecision to my fireangle
  • 0.2.11 (20090426) gun: GV 0.2.7 move: WS 0.2.3
    • APS: 80.75 (29th) ELO: 2053 PL: 648-51
  • 0.2.4 (20090426) gun: GV 0.2.5 move: WS 0.1.7
    • APS: 80.28 (33rd) ELO: 2043 PL: 638-60

[edit] Ancient rankings

  • 0.2.11 (20080830) gun: GV 0.2.7 move: WS 0.2.3
    • Rating: 2050 (8th) (due to messed up ranking), PL: 596-44 (40th)
    • keep a 'firing-window' to (further) eliminate enemy ghost-bullets
    • anti-ram: when calculating the GF -1/1 bins, assume enemy is chasing me at constant distance
  • 0.2.10 (20080216) gun: GV 0.2.7 move: WS 0.2.2
    • Rating: 2041 (25th), PL: 581-41 (39th)
    • add 'distance last 14 ticks' segment in gun
    • more conservative energy-management in endgame
  • DookiSuffurd 0.2.9 (20071009) gun: DookiLightning 1.554 move: WS 0.2.2
    • Rating: 2078 (5th), PL: 561-15 (12th)
    • what to improve in the [gun]
    • what to improve in [surfing]
  • 0.2.9 (20070612) gun: GV 0.2.6 move: WS 0.2.2
    • Rating: 2015 (21st), PL: 509-35 (33rd)
    • remove wall segmentation of movement
    • what is left is 0.2.4 with some minor changes
  • 0.2.8 (20070608) gun: GV 0.2.6 move: WS 0.2.1
    • Rating: 2013 (22nd), PL: 487-36 (35th)
    • remove distance segmentation of movement
    • re-introduce wall segmentation of movement (other implementation)
  • 0.2.7 (20070523) gun: GV 0.2.6 move: WS 0.2.0
    • Rating: 2014 (22th), PL: 487-36 (34th)
    • segment movement also on distance
    • slight preference for current direction
    • get rid of 'GF1-protection' multiplyfactor (gun)
    • no decay on gunstats
  • 0.2.6 (20070317) gun: GV 0.2.5 move: WS 0.1.9
    • Rating: 2017 (20th), PL: 479-29 (29th)
    • rollback rolling the surfingstats
    • get rid of 'GF1-protection' multiplyfactor (movement)
  • 0.2.5 (20070311) gun: GV 0.2.5 move: WS 0.1.8
    • Rating: 2010 (21th), PL: 481-27 (28th)
    • rolling the surfingstats (0.7*old + 0.3*new)
    • lower rating but better in PL
  • 0.2.4 (20070222) gun: GV 0.2.5 move: WS 0.1.7
    • Rating: 2019 (20th), PL: 474-34 (34th)
    • change wall-segmentation (gun) from 3 to 5 bins
  • 0.2.3 (20070208) gun: GV 0.2.4 move: WS 0.1.7
    • Rating: 2009 (22nd), PL: 468-38 (36th)
    • do not fire virtual waves if opponent did not move yet
    • use information from onBulletHitBullet
  • 0.2.2 (20070125) gun: GV 0.2.3 move: WS 0.1.6
    • Rating: 2001 (21st), PL: 456-39 (36th)
    • small decay in gunstats
    • always have a surfable (real or simulated) wave
  • 0.2.1 (20070117) gun: GV 0.2.2 move: WS 0.1.5
    • Rating: 2000 (21nd), PL: 448-40 (37th) member of The2000Club
    • bugfix weighting of waves
  • 0.2.0 (20070114) gun: GV 0.2.2 move: WS 0.1.4
    • Rating: 1992 (22nd), PL: 444-43 (38th)
    • removed nearwall segmentation
    • also evaluate second wave
  • 0.1.9 (20070102) gun: GV 0.2.2 move: WS 0.1.3
    • Rating: 1980 (23nd), PL: 439-43 (40th)
    • segment movement also on nearwall (3 segments)
  • 0.1.8 (20061212) gun: GV 0.2.2 move: WS 0.1.2
    • Rating: 1982 (22nd), PL: 436-44 (41st)
    • reverted anti-ram energymanagement, segment movement on absolute currentvelocity (3 segments)
  • 0.1.7 (20061127) gun: GV 0.2.1 move: WS 0.1.1
    • Rating: 1946 (30th), PL: 428-50 (46th)
    • no stop-position evaluation and adapted energymanagement when enemy is close (anti-ram)
  • 0.1.6 (20061117) gun: GV 0.2.0 move: WS 0.1.0
    • Rating: 1946 (30th), PL: 429-48 (43rd)
    • also use non-bullet waves with 0.1 weighting
  • 0.1.5 (20061028) gun: GV 0.1.5 move: WS 0.1.0
    • Rating: 1938 (33rd), PL: 414-57 (50th)
    • proper onDeath handling (David Alves) and 'onWin handling'
    • slowdown when wallhit is eminent
  • 0.1.4 (20061024) gun: GV 0.1.4 move: WS 0.0.8
    • Rating: 1931 (33th), PL: 418-49 (45th)
    • bugfixing stop-position
  • 0.1.3 (20061023) gun: GV 0.1.4 move: WS 0.0.7
    • Rating: 1925 (33th), PL: 415-51 (47th)
    • continue direction when evaluations are equal
    • evaluate stop-position
    • use 5-bin wide hitmark instead of 3-bin (no 'classical' binsmoothing)
  • 0.1.2 (20061014) gun: GV 0.1.4 move: WS 0.0.5
    • Rating: 1900 (40th), PL: 413-52 (49th)
    • keep preferred distance, this automatically includes simple dive-in protection
    • circle enemy when no wave is airborne
  • 0.1.1 (20061007) gun: GV 0.1.4 move: WS 0.0.4
    • Rating: 1854 (49th), PL: 395-61 (56th)
    • record all hitted bins and fire on highest concentration
    • fire HOT and no waves on disabled enemies
  • 0.1.0 (20060926) gun: GV 0.1.1 move: WS 0.0.4
    • Rating: 1828 (54th), PL: 384-73 (65th)
    • basic wavesurfing and basic segmented guessfactor gun
Personal tools