Difference between revisions of "User:Tkiesel/Dev Notebook"

From Robowiki
Jump to navigation Jump to search
m (Voting. Fun. :))
m (updated)
Line 2: Line 2:
  
 
# Continue tracking down why there are so many skipped turns.
 
# Continue tracking down why there are so many skipped turns.
#* I used an approximation via <math>e^x = \lim_{n \to +\infty}(1+\frac{x}{n})^n</math>, with n = 128  It helped a lot, and it only gets serious error at 4-5+ bandwidths away. Also, the approximation drops off faster than the real thing, which is more acceptable to me than lingering too high.
+
#* I used an approximation via <math>e^x = \lim_{n \to +\infty}(1+\frac{x}{n})^n</math>, with variable power-of-2 choice of n based on input x. Keeps error < 5%, and performance high. Skipped turn issue seems to be resolved... for now.
# Virtual gun array time!
+
# Virtual gun array
 
#* How to choose a gun?   
 
#* How to choose a gun?   
#** Current VirtualGunManager uses Condorcet voting among several selectors to choose the best gun. Needs lots of work.. but then again, so does my brand new learning gun!
+
#** Current VirtualGunManager uses Condorcet voting among several selectors (rolling normalized hit rates with different depths) to choose the best gun.
 +
#*** Works quite well, changes guns appropriately when [[Dookious]] engages/disengages its flattener.
 +
#*** However, no big rumble score gain over rev0073b with main gun.
 +
#**** I probably just need to continue to develop/tune the guns themselves (the common code they're based upon) and choose only 2-4 of them to minimize churn.
 +
# Flattener
 +
#* Put it in a separate object so it's not hanging out at 12 different places in Wheels.java
 +
#* Really think through some theory on when/how to engage.
 +
# Bullet Power/fire choice

Revision as of 19:51, 1 July 2012

DeBroglie Roadmap / Development notepad / Brainstorming Sandbox

  1. Continue tracking down why there are so many skipped turns.
    • I used an approximation via <math>e^x = \lim_{n \to +\infty}(1+\frac{x}{n})^n</math>, with variable power-of-2 choice of n based on input x. Keeps error < 5%, and performance high. Skipped turn issue seems to be resolved... for now.
  2. Virtual gun array
    • How to choose a gun?
      • Current VirtualGunManager uses Condorcet voting among several selectors (rolling normalized hit rates with different depths) to choose the best gun.
        • Works quite well, changes guns appropriately when Dookious engages/disengages its flattener.
        • However, no big rumble score gain over rev0073b with main gun.
          • I probably just need to continue to develop/tune the guns themselves (the common code they're based upon) and choose only 2-4 of them to minimize churn.
  3. Flattener
    • Put it in a separate object so it's not hanging out at 12 different places in Wheels.java
    • Really think through some theory on when/how to engage.
  4. Bullet Power/fire choice