Apr 2 2016, 6:56 pm
|
|
Nice.
|
KHUD that I use for Elora 2
// This is the engine. |
Outline code, sharing random stuff :P
obj |
My first scripting language I made back in 2013, you never stop improving tho :D
Examples of the language new obj1 /obj/Cookie : Rename A -1 cookies 0 bottles 99 The parser atom/movable |
if(passed) |
It is on a very rare occasion one has a function that long and is not doing something tremendously wrong. You have not happened upon one of those situation.
|
In response to Popisfizzy
|
|
Popisfizzy wrote:
It is on a very rare occasion one has a function that long and is not doing something tremendously wrong. You have not happened upon one of those situation. xD I made this and a bit more in a week with no internet around 3 or 4 years ago. Being the first time it allowed me to see what I needed to work around if I was to build it from scratch again and again improving each time. Makes it alot easier to understand how to do things :) |
[12:29:27 PM] Kozuma3: So I thought about AI in DM and I came up with an idea |
In response to Kozuma3
|
|
Why would you have a concrete representation when an abstract representation would work equally well with fewer resources? On top of that, that idea is already in use: chess AI works by looking ahead at possible child nodes in the game tree of the current state in the board and taking the approach that it finds most optimal. Thus, it looks forward a number of steps at and finds the choice it "feels" is best.
|
In response to Popisfizzy
|
|
Popisfizzy wrote:
Why would you have a concrete representation when an abstract representation would work equally well with fewer resources? Would have to write my own Bump and such procs causing more overhead than needed. On top of that, that idea is already in use: chess AI works by looking ahead at possible child nodes in the game tree of the current state in the board and taking the approach that it finds most optimal. Thus, it looks forward a number of steps at and finds the choice it "feels" is best. Yea, but while doing such could find and mark paths that would allow other AI to do such actions following the results making it so you'd only need to run it when needed. |
In response to Kozuma3
|
|
Kozuma3 wrote:
Would have to write my own Bump and such procs causing more overhead than needed. There are still ways of doing it other than recreating the entire map, doubling your memory and CPU footprint. You simply have to be a tiny bit creative. Yea, but while doing such could find and mark paths that would allow other AI to do such actions following the results making it so you'd only need to run it when needed. Caching or precomputing. Use them. |
In response to Popisfizzy
|
|
Popisfizzy wrote:
There are still ways of doing it other than recreating the entire map, doubling your memory and CPU footprint. You simply have to be a tiny bit creative. Nothing would be as efficient and I could easily modify this further and to allow for more than just Bump() ^^ Caching or precomputing. Use them. idk how or wat those exactly are, I just make stuff |
In response to Kozuma3
|
|
4/20 yesterday, no relation.
I had a random thought, would it be possible to somehow use multiple dream servers/dd's to host a single server to pass the single core limit? o-o Like you could sync then up via a config file from a master one or something idk |
I wonder what would happen if you used Distributed_Computing with Precomputation for AI to act leik different parts of the brain or soffin.
|