User talk:D414/Diary of a Robocoder

Great work! I'm very curious how it would do when combined with nano-sized melee movement. In case it helps (and you haven't already done it) you could reduce the code size by

• Removing the gameTime > 1 check.
• Use a constant state instead of makeState(...) when the opponent wasn't scanned for consecutive ticks.
• Instead of replacing a random table entry, just keep replacing the last one when the table is full.
• Don't add the weight/5 addition to the midpoint aim bin.

I imagine these changes would only slightly degrade performance while saving a decent amount of code size, but I kept them in Mirage because I had the space.

Thank you!

For a while I was noticing quite a drop in performance without the gameTime > 1 check. It was driving me nuts because I couldn't figure out why but trying it now it has minimal impact. I'm still confused but won't say no to another 5 bytes! That gets me down to 665 bytes without any movement code or colours.

I've tried it with Quantum, Lib and Infinity's movements so far. Quantum's movement isn't a great fit, presumably because it moves about and changes target quite frequently. Lib's movement wasn't quite as good as expected, not much (if any) better than just doing Infinity's box movement without seeking the edge which itself might still be good enough to hit top five in the micro melee.

Sprout's movement gave some pretty exciting results. I've not tested extensively but it wouldn't surprise me if it challenged Wallaby. Gotta squeeze a bit more out for that to fit though.

That is quite the result for your first bot :). Avoiding where other bots are (probably) going to be rather than where they are now is a really neat idea that maybe could me useful for melee bots of all codesizes.

Congrats! Glad to see new (and mighty!) competitors in melee sector, there’s still plenty to explore here. Maybe even a new general melee king in the near future;)

I definitely have my eye on your crown but I've got micro and mini to go before I start a megabot! Before I get to any of that I need to build a framework that'll let me tune my bots properly which will probably take longer to develop than Quantum did :(

Thank you! I've got a few ideas to improve on Quantum's implementation of the concept, although most/all of them will have to wait until I start working on a microbot :)

I'm definitely standing on the shoulders of giants with this entry though, which reminds me that I need to create a page for Quantum and repackage it with source code asap so that I can credit where it's due.

Looks like Quantum might lose the nano melee crown once scores stabalise some more. I'm struggling to find the change that would warrant a v0.3.0 release and it'll be a long time before I have a framework to help me fine tune parameters to mark v1.0.0.

That said, even if I do lose the nano crown I'm still incredibly proud of Quantum's survival score - it smashes the other nanos out of the park on survival - it's currently 4th on survival in micro and 10th in mini. In fact, it currently has a better survival score in the mini category than it does in nano.

And there goes my nano crown. Looks like I'm going to have to drop the colours :(