From: spooq <spoooq@...>
Date: 2006-11-02T23:39:57+09:00
Subject: Civ4/integration question and practice for the golfers

------=_Part_4055_32291534.1162478393326
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

For reasons not clear even to myself, I'm trying to predict the growth and
production of a city in CivIV (c.f. CivFanatics <http://www.civfanatics.com>
).

Each city has an integer population and a number of tiles (actually 20 but
reduced here to 10 to speed-up testing). Each population point (P) can work
one tile in a given turn. Tiles have three attributes excitingly named
F, G and H. I am attempting to enumerate all interesting combinations
of worked tiles for any given city size. Output is in gnuplot format.

Here's the code (fore!).

puts "Starting #{Time.now}"

numArray = (1..10).to_a

# Generate tiles
tiles = numArray.inject([]) { |acc, node| acc + [[rand(4),rand(4), rand(4)]]
}

# Generate unique combinations for given size
File::open("output.txt", "w") { |file|
  file << "#P F G H\n"
  numArray.inject([[]]) { |acc, node|
    acc.inject([]) { |acc, injectedNodes| acc + [injectedNodes,
injectedNodes + [node]] }
  }.collect { |c|
    c.inject([0,0,0,0]) { |acc, node| [acc[0].succ, acc[1] +
tiles[node-1][0], acc[2] + tiles[node-1][1], acc[3] + tiles[node-1][2]] }
  }.uniq.sort.inject(1) { |acc, node|
    file << "\n" if node[0] == acc
    file << " " << node.join(" ") << "\n"
    node[0] == acc ? acc.succ : acc
  }
}

puts "Finished #{Time.now}"

My problem is ensuring that only efficient combinations are output. If we
look at a doctored example...

#P A B C
0 0 0 0

1 0 0 0
1 2 0 0            <- This is not interesting,
1 2 0 3            <- when this is possible.
1 0 1 0            <- However this still is, even though A+B+C is less than
the line above.

Can anyone suggest a nice golfed algorithm to get rid of the useless
combinations? Remember that any of the three resources might be deemed most
important, e.g. 1 1 0 0 is still interesting when 1 0 9 9 exists.

When this bit gets nailed down, the next step is integrating permutations of
combinations over time and solving for maximum G and/or H, given that
acquiring enough F will increase P. It's probably best to assume no
micro-management, i.e. when a city grows, the combination chosen for the new
size remains in place until the next growth.

Thanks for getting this far, all comments welcome.

------=_Part_4055_32291534.1162478393326--