search

jsvazic / double-fanucci-ga

Automatically exported from code.google.com/p/double-fanucci-ga
1 stars 0 forks source link

The total card value is wrong if there are more than 3 suits in a given hand. #12

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Set 2 (32)
3 of Tops, 2 of Tops
2 of Ears
2 of Fromps

Is actually 28
Set 3 (18)
2 of Bugs, 1 of Bugs
1 of Zurfs
2 of Inkblots

is not 18 but actually 16

Original issue reported on code.google.com by jsva...@gmail.com on 9 May 2009 at 7:20

GoogleCodeExporter commented 9 years ago 
From Canis:

I ran into the same exact problem. You need to apply multiple modifiers to by 
card 
group. So for a quick idea...

int values[]
double modifiers[]
int bestValue
int bestGroup
for each card in hand
  values[card.group] += card.value
  if values[card.group] > bestValue
    bestValue = values[card.group]
    bestGroup = card.group
  end if
end for each

for each modifier in modifiers
  -modifier = 1
end for each

for g1 = 0, g1 < values.length
  for g2 = 0, g2 < values.length
    if g1 != g2 && g2 != bestGroup
      modifiers[g2] *= Group.getModifier(g1, g2)
    end if
  end for
end for

int value = 0;
for i = 0, i < values.length
  value += (values[i] * modifiers[i])
end for

That is the basic idea. You can make your genetic smarter by giving each card a 
team 
and only have cards that are on the same team on the same hand. This makes 
finiding 
the modifier easy as if the two groups are the same then the modifier is 1, if 
they 
are different then the modifier is .5. Since the 0 and -.5 modifiers are part 
of the 
other team those get ignored since a hand has a specific team. It should 
greatly 
speed up your genetic by ignoring cards that dont work well together and should 
make 
calculating hand value much easier.

Original comment by jsva...@gmail.com on 10 May 2009 at 3:43

GoogleCodeExporter commented 9 years ago 
This one is a pain.  Following the suggestion seems to have my numbers skewed 
for 
some reason.  I need to investigate further.

Original comment by jsva...@gmail.com on 11 May 2009 at 7:06

GoogleCodeExporter commented 9 years ago 
Still working on this one.  Major PITA.

Original comment by jsva...@gmail.com on 11 May 2009 at 7:58

GoogleCodeExporter commented 9 years ago 
Finally resolved.  The solution was to modify the getFitness() method of the 
FanucciChromosome class to discourage either having more than 3 cards from a 
single 
suit or having more than 2 groups of suits in a given hand.

Unfortunately the above algorithm does not work in all situations, so it was 
decided 
it would be better to go with the original suggestion in the forums, namely to 
avoid 
using more than 2 groups of cards per hand/slot.

Original comment by jsva...@gmail.com on 12 May 2009 at 4:02