Free and Open Source real time strategy game with a new take on micro-management

Difference between revisions of "Solution to the guard area distribution problem"

From Globulation2

Jump to: navigation, search
(New page: == The problem == Guard areas have the problem that they share one common gradient per team. This results in all the warriors going to the closest area ignoring the distant ones. == A pos...)
 
(A possible solution)
Line 6: Line 6:
* count of warriors assigned
* count of warriors assigned
* size
* size
* saturation = count of warriors assigned / size
so those regions have a saturation (warrior-count / size) that warriors could take into account when picking a gradient. Of course it will be non-trivial to optimize distances so that east-warriors don't get assigned to west-regions while west-warriors walk to the east-regions while at the same time guarantee distant regions to get saturated but stupidly long paths through simply picking the least saturated region in range would still be better than what we have now in b4.
warriors could take the saturation into account when picking a gradient.

== Problem ==
Of course it will be non-trivial to optimize distances so that east-warriors don't get assigned to west-regions while west-warriors walk to the east-regions and at the same time guarantee distant regions to get saturated.

== Conclusion ==
Stupidly long paths through simply picking the least saturated region in range would still be better than what we have now in b4. Going for the above approach would be a first step to turn the so called guard areas into actual guard areas.

Revision as of 22:45, 6 June 2010

The problem

Guard areas have the problem that they share one common gradient per team. This results in all the warriors going to the closest area ignoring the distant ones.

A possible solution

In order to get a more even distribution but at the same time keep the extra gradient count low, the guard area gradient should be split up into several guard region gradients. One for each connected patch of guard areas. Each such region would need to get some attributes such as:

  • count of warriors assigned
  • size
  • saturation = count of warriors assigned / size

warriors could take the saturation into account when picking a gradient.

Problem

Of course it will be non-trivial to optimize distances so that east-warriors don't get assigned to west-regions while west-warriors walk to the east-regions and at the same time guarantee distant regions to get saturated.

Conclusion

Stupidly long paths through simply picking the least saturated region in range would still be better than what we have now in b4. Going for the above approach would be a first step to turn the so called guard areas into actual guard areas.

Misc