Re: Finding Clusters
- To: mathgroup at smc.vnet.net
- Subject: [mg117111] Re: Finding Clusters
- From: Daniel Lichtblau <danl at wolfram.com>
- Date: Thu, 10 Mar 2011 06:07:54 -0500 (EST)
Martin Vavpotic( wrote:
> Hello everyone.
>
> I have a large array of numbers (1000 =D72). I wish to find clusters in
> that mesh of numbers but the FindClusters method is giving me a hard
> time. I think it's because of those non-numeric leftovers from NSolve
> (x -> and y ->).
>
> {{x -> 9.95043, y -> 0.201587}, {x -> -0.0507389,
> y -> 0.39451}, {x -> -0.0502288 + 0.192544 I,
> y -> 0.199992 - 2.42154*10^-6 I}, {x -> -0.0502288 - 0.192544 I,
> y -> 0.199992 + 2.42154*10^-6 I}, <<992>>, {x -> -0.0507405,
> y -> 0.394519}, {x -> -0.0502287 + 0.192544 I,
> y -> 0.2}, {x -> -0.0502287 - 0.192544 I,
> y -> 0.2}, {x -> -0.0507405, y -> 0.00548135}}
>
> I know that if you write something like %[[1,2,2]] you get only the
> numbers but I have a whole array here. How do I get the whole array to
> lose those y -> ?
>
> Help anyone?
>
Have a look at Documentation Center > Rule. This will strongly suggest
the following approach:
{x, y} /. {{x -> 9.95043, y -> 0.201587}, {x -> -0.0507389,
y -> 0.39451}, {x -> -0.0502288 + 0.192544 I,
y -> 0.199992 - 2.42154*10^-6 I}, {x -> -0.0502288 - 0.192544 I,
y -> 0.199992 + 2.42154*10^-6 I}, {x -> -0.0507405,
y -> 0.394519}, {x -> -0.0502287 + 0.192544 I,
y -> 0.2}, {x -> -0.0502287 - 0.192544 I,
y -> 0.2}, {x -> -0.0507405, y -> 0.00548135}}
Out[21]= {{9.95043, 0.201587}, {-0.0507389,
0.39451}, {-0.0502288 + 0.192544 I,
0.199992- 2.42154*10^-6 I}, {-0.0502288 -
0.192544 I,
0.199992 + 2.42154*10^-6 I}, {-0.0507405,
0.394519}, {-0.0502287 + 0.192544 I, 0.2}, {-0.0502287 - 0.192544 I,
0.2}, {-0.0507405, 0.00548135}}
Depending on your specific needs, you might also want to lop (chop) off
the imaginary parts. That can be done either by using Re[] or
Chop[...,tol] where tol gives the tolerance, that is, magnitude, beneath
which values will be replaced by zero.
Daniel Lichtblau
Wolfram Research