On April 19 2008 12:36 azndsh wrote:
if you fix 5 coordinates, the remaining 2 coordinates determine a unit square... thus 632
if you fix 5 coordinates, the remaining 2 coordinates determine a unit square... thus 632
^there
| Blogs > micronesia |
|
micronesia
United States24776 Posts
On April 19 2008 12:36 azndsh wrote: if you fix 5 coordinates, the remaining 2 coordinates determine a unit square... thus 632 ^there | ||
|
azndsh
United States4447 Posts
definitely 672 | ||
|
micronesia
United States24776 Posts
On April 19 2008 16:10 azndsh wrote: typo, sorry hehe definitely 672 All right, cool. I know how to calculate the answer to it (I showed the work above, sort of) but I have absolutely no understanding of it. | ||
|
sigma_x
Australia285 Posts
Sum( C(n,k)C(k,m) ) = 2^(n-m) C(n,m) for an "n" dimensional cube within which we select "m" dimensional cubes. So, for example, a 3-dimensional cube has 6 2-dimensional cubes (faces). We shall prove the identity by double-counting. The right hand side is readily solvable by graph theory. Each n dimensional cube has 2^n vertices. Now, there are C(n,m) ways of selecting an m-th dimensional cube. Further, each selected m-th dimensional cube has 2^m vertices, so that we have 2^n * C(n,m)/2^m and we are done. For the second method, place the n-th dimensional cube on a Cartesian plane. Let the cube have side length 2. Then each vertex can be specified by (±1,±1,±1,...,±1,±1). Now, for any given vertex defined by a set of coordinates, we can change m of them (from +1 to -1 and vice versa) and the vertex will still lie on the m-th dimensional cube. Further, these two points define a unique m-cube (since this is the maximal number of coordinates that can be changed for both coordinates to lie on the same m-cube). Now, there are C(n,m) ways of selecting m coordinates (and assign them the value +1). For each of these, m of them can be changed to -1 C(m,m) ways. Together, these two points define a unique m-cube. Thus, there are C(n,m)C(m,m) ways of doing this. Now, do the same by assigning m+1 coordinates +1, C(n,m+1) ways. For each of these, m of them can be changed to -1, C(m+1,m) ways. Thus, there are C(n,m+1)C(m+1,m) ways of doing this. We repeat this process until we are assigning n coordinates +1, of which we change m of them to -1. Since this process is mutually disjoint, and covers all m-cubes, the total number of m-dimensional cubes in an n-dimensional cube is Sum( C(n,k)C(k,m) ). This completes the proof and explains why your process works. Edit: lol, i didn't actually answer the question. Let n=7, m=2, and we have C(7,2)C(2,2)+C(7,3)C(3,2)+C(7,4)C(4,2)+C(7,5)C(5,2)+C(7,6)C(6,2)+C(7,7)C(7,2)=21x1 + 35x3 + 35x6 + 21x10 + 7x15 + 1x21 = 2^5 C(7,2) = 672 | ||
|
micronesia
United States24776 Posts
On April 20 2008 16:49 sigma_x wrote: What your method exhibits is the following: Sum( C(n,k)C(k,m) ) = 2^(n-m) C(n,m) for an "n" dimensional cube within which we select "m" dimensional cubes. So, for example, a 3-dimensional cube has 6 2-dimensional cubes (faces). We shall prove the identity by double-counting. The right hand side is readily solvable by graph theory. Each n dimensional cube has 2^n vertices. Now, there are C(n,m) ways of selecting an m-th dimensional cube. Further, each selected m-th dimensional cube has 2^m vertices, so that we have 2^n * C(n,m)/2^m and we are done. For the second method, place the n-th dimensional cube on a Cartesian plane. Let the cube have side length 2. Then each vertex can be specified by (±1,±1,±1,...,±1,±1). Now, for any given vertex defined by a set of coordinates, we can change m of them (from +1 to -1 and vice versa) and the vertex will still lie on the m-th dimensional cube. Further, these two points define a unique m-cube (since this is the maximal number of coordinates that can be changed for both coordinates to lie on the same m-cube). Now, there are C(n,m) ways of selecting m coordinates (and assign them the value +1). For each of these, m of them can be changed to -1 C(m,m) ways. Together, these two points define a unique m-cube. Thus, there are C(n,m)C(m,m) ways of doing this. Now, do the same by assigning m+1 coordinates +1, C(n,m+1) ways. For each of these, m of them can be changed to -1, C(m+1,m) ways. Thus, there are C(n,m+1)C(m+1,m) ways of doing this. We repeat this process until we are assigning n coordinates +1, of which we change m of them to -1. Since this process is mutually disjoint, and covers all m-cubes, the total number of m-dimensional cubes in an n-dimensional cube is Sum( C(n,k)C(k,m) ). This completes the proof and explains why your process works. Edit: lol, i didn't actually answer the question. Let n=7, m=2, and we have C(7,2)C(2,2)+C(7,3)C(3,2)+C(7,4)C(4,2)+C(7,5)C(5,2)+C(7,6)C(6,2)+C(7,7)C(7,2)=21x1 + 35x3 + 35x6 + 21x10 + 7x15 + 1x21 = 2^5 C(7,2) = 672 Every time someone attempts to explain this to people who haven't learned about it already, it seems to result in them going 'HUH?' I guess only those who study the proper topics in college will get what the hell this means. | ||
| ||
StarCraft 2 StarCraft: Brood War Britney Dota 2Rain Calm Shuttle Mini Hyuk actioN Horang2 Rush BeSt [ Show more ] Counter-Strike Other Games Organizations
StarCraft 2 • HeavenSC StarCraft: Brood War• intothetv • AfreecaTV YouTube • Kozan • IndyKCrew • LaughNgamezSOOP • Migwel • sooper7s League of Legends Other Games |
|
uThermal 2v2 Circuit
Maestros of the Game
ByuN vs herO
Rogue vs Bunny
Replay Cast
Replay Cast
WardiTV Spring Champion…
OSC
Maestros of the Game
Serral vs Percival
SHIN vs ShoWTimE
Replay Cast
uThermal 2v2 Circuit
Maestros of the Game
Clem vs Lambo
Zoun vs SKillous
[ Show More ] Replay Cast
Solar vs Classic
uThermal 2v2 Circuit
Grudge Match
FlaShFTW vs A.Alm
GSL
herO vs Rogue
Maru vs Cure
Patches Events
uThermal 2v2 Circuit
BSL
Replay Cast
Monday Night Weeklies
Sparkling Tuna Cup
Replay Cast
Kung Fu Cup
|
|
|