ScalarN/Version History

• 0.011d.115 (September 21, 2018)
  • RoboRumble ‒ APS: 88.59% (5th), PL: 1136-19 (17th), Survival: 95.53%
  • add more 1nn view (most recent & all)

• 0.011d.113 (September 21, 2018)
  • RoboRumble ‒ APS: 88.28% (7th), PL: 1131-24 (22nd), Survival: 95.23%
  • add more 1nn view (recent 20)

• 0.011d.107 (September 20, 2018)
  • RoboRumble ‒ APS: 88.52% (5th), PL: 1139-16 (14th), Survival: 95.52%
  • Faster decay in movement; add more 1nn tree

• 0.011d.96 (September 16, 2018)
  • RoboRumble ‒ APS: 88.52% (5th), PL: 1137-18 (16th), Survival: 95.45%
  • Only match enemy power when far away
  • Detect rammers and increase anti-rammer power distance threshold

• 0.011d.92 (September 16, 2018)
  • RoboRumble ‒ APS: 88.33% (7th), PL: 1138-17 (15th), Survival: 95.27%
  • Tune gun
    • Special handling of disabled bots
    • Match rolling average enemy power
    • Only match enemy power when I'm not doing good
    • Undercut power even earlier

• 0.011d.91 (September 15, 2018)
  • RoboRumble ‒ APS: 88.16% (9th), PL: 1140-15 (14th), Survival: 94.83%
  • Add more good power tiers

• 0.011d.90 (September 15, 2018)
  • RoboRumble ‒ APS: 87.88% (9th), PL: 1138-17 (15th), Survival: 95.34%
  • Tune energy management
    • Use 0.1 power earlier
    • Match lower enemy power
    • Round to nearest “good power” (with optimal damage per tick/energy)
    • Undercut power earlier

• 0.011d.87 (September 15, 2018)
  • RoboRumble ‒ APS: 88% (9th), PL: 1134-21 (21st), Survival: 93.17%
  • Use precise intersection in gun; tune for TC score

• 0.011d.80 (September 10, 2018)
  • RoboRumble ‒ APS: 87.72% (9th), PL: 1130-25 (22nd), Survival: 92.82%
  • Tune against rambots: fire 3.0 when near; cut precise gf to [-1, +1].

• 0.011d.78 (September 09, 2018)
  • RoboRumble ‒ APS: 87.18% (11th), PL: 1132-23 (22nd), Survival: 92.16%
  • finally N gets a vivid color
  • add a duel gun
  • tune mc2k7

• 0.011c.32 (September 06, 2018)
  • RoboRumble ‒ APS: 86.55% (13th), PL: 1125-30 (26th), Survival: 91.37%
  • Tune for MC2k7
    • Surf multiple trees like ScalarBot!
  • Vote 14.06 (2nd) after ShieldNext

• 0.011c (September 02, 2018)
  • RoboRumble ‒ APS: 84.71% (25th), PL: 1117-38 (35th), Survival: 89.7%
  • 1v1 release
    • Delay pif waves until hit
    • Update waves on new bullet recorded
    • Detect wall damage in 1v1 and melee
    • Very simple surf tree

• 0.011a (August 29, 2018)
  • MeleeRumble ‒ APS: 71.36% (3rd), PL: 400-4 (5th), Survival: 64.98%
  • Use Manhattan distance in gun

• 0.011 (August 28, 2018)
  • MeleeRumble ‒ APS: 71.23% (3rd), PL: 400-4 (5th), Survival: 64.06%
  • State-of-the-art wall hit avoidance (best used together with wall smoothing)
  • Use intersect distance instead of current distance in melee surfing
  • Wave Surfing in 1v1

• 0.01 (August 26, 2018)
  • MeleeRumble ‒ APS: 70.29% (3rd), PL: 395-9 (10th), Survival: 45.97%
  • Initial release

• 0.01 phase 4 (August 26, 2018 – August 26, 2018)
  • Melee movement between minimal risk and wave surfing (surfing hot yet)

• 0.01 phase 3 (August 20, 2018 – August 26, 2018)
  • State-of-the-art swarm knn pif melee gun
    • Very simple but very effective now

• 0.01 phase 2 (August 03, 2018 – August 05, 2018)
  • Simple minimal risk movement influenced by Neuromancer, Firestarter, Diamond and HawkOnFire and more bots.

• 0.01 phase 1 (July 20, 2018 – August 03, 2018)
  • State-of-the-art melee radar that maximizes scan rate
    • Unified design for melee and duel
    • Graceful handling of out-ranged robot, supporting even 5000x5000 battle field!

• 0.01 phase 0 (July 09, 2018 – August 19, 2018)
  • Bot Development Framework
    • Lightweight Log4j style logging for robot and corresponding Intellij plugin
    • BaseRobot — AdvancedRobot alternative, decoupling bot implementation from robocode
    • Subject/Listener style component communication, and bind together automatically via ListenerRegister module!
    • Fully redesigned Unit module that provides a unified model for all robots in battlefield
  • Math and Data Structures
    • 2D vector library V inspired by PyTorch; Math library U that unifies math/geom API.
    • Primitive Collections Framework that feels the same as JCF but much faster and compact for primitives.
    • Brand new Xor's KdTree.