列出6位数中的所有3位数组合

时间:2016-04-14 16:00:48

标签: r combinations combinatorics

我们说我有6个数字:

a <- c(1,2,3,4,5,6)

我想列出这6个数字的所有可能的3位数组合,包括重复。

期望的结果如下所示:

1 1 1
1 2 3
1 2 4
... 

我不想包含具有相同3个数字但顺序不同的元素:

例如

1 2 3
3 2 1

应该排除其中一个

4 个答案:

答案 0 :(得分:5)

来自combinations的{​​{1}}功能可以执行此操作:

gtools
library(gtools)
combinations(n=6, r=3, v=a, repeats.allowed=TRUE)

答案 1 :(得分:2)

SELECT page FROM Page page INNER JOIN page.permissions permission INNER JOIN permission.user user WITH IDENTITY(user) = :userId 返回组合的data.frame,从您提供的每个集合中选择一个。如果expand.grid1 2 3不同,请对其进行子集化以获得所需的行。

3 2 1

答案 2 :(得分:2)

下面是一个允许您在输出上指定约束的通用函数。例如,我有很多情况需要,因为我需要给定集合的所有n元组,使得产品小于给定的界限。在我编写此函数之前,我被迫使用combinations并搜索符合我条件的那些行。这花费了大量的时间和大量的记忆。

Combo  <- function(n,r,v=1:n,li=10^8,fun1="prod",fun2="<",repeats.allowed=FALSE) {
    ## where fun1 is a general function such as "prod", "sum", "sd", etc.
    ## and fun2 is a comparison operator such as "<", "<=", ">", "==",  etc.

    myfun <- match.fun(FUN = fun1)
    operator1 <- match.fun(FUN = fun2)
    operator2 <- match.fun(FUN = fun2)
    myv <- sort(v)

    if (fun2 %in% c(">",">=")) {
        myv <- rev(myv)
        TheLim <- min(v)
    } else {
        TheLim <- max(v)
        if (fun2 == "==") {
            operator1 <- match.fun(FUN = "<=")
        }
    }

    if (!repeats.allowed) {
        m <- matrix(numeric(0),combinat::nCm(n,r),r)
        v1 <- myv; n1 <- length(v); t <- TRUE; count <- 0L

        while (t) {
            t <- operator1(myfun(v1[1:r]),li)
            while (t && length(v1)>=r) {
                t_1 <- operator2(myfun(v1[1:r]),li)
                if (t_1) {count <- count+1L; m[count,] <- v1[1:r]}
                v1 <- v1[-r]
                t <- operator1(myfun(v1[1:r],na.rm=TRUE),li)
            }
            if (t) {
                s <- 1:length(v1)
                mymax <- myv[n1-(r-s)]
                t1 <- which(!v1==mymax)
                if (length(t1)>0) {
                    e <- max(t1)
                    v1[e] <- myv[which(myv==v1[e])+1L]
                    v1 <- c(v1[1:e],myv[(which(myv==v1[e])+1L):n1])
                } else {
                    return(m[!is.na(m[,1]),])
                }
            } else {
                r1 <- r-1L
                while (r1>=1L && !t) {
                    v1[r1] <- myv[which(myv==v1[r1])+1L]
                    if (v1[r1]==TheLim) {r1 <- r1-1L; next}
                    v1 <- c(v1[1:r1],myv[(which(myv==v1[r1])+1L):n1])
                    t <- operator1(myfun(v1[1:r],na.rm=TRUE),li) && length(v1)>=r
                    r1 <- r1-1L
                }
                if (!t) {return(m[!is.na(m[,1]),])}
            }
        }
    } else {
        MySet <- 1:n 
        for (i in 1:(r-1L)) {MySet <- sapply(1:n, function(x) sum(MySet[1:x]))}
        m <- matrix(numeric(0),nrow=MySet[n],ncol=r)
        v1 <- c(rep(myv[1], r),myv[2:n]); n1 <- length(v); t <- TRUE; count <- 0L

        while (t) {
            t <- operator1(myfun(v1[1:r]),li)
            while (t && length(v1)>=r) {
                t_1 <- operator2(myfun(v1[1:r]),li)
                if (t_1) {count <- count+1L; m[count,] <- v1[1:r]}
                v1 <- v1[-r]
                t <- operator1(myfun(v1[1:r],na.rm=TRUE),li)
            }
            if (t) {
                s <- 1:length(v1)
                t1 <- which(!v1==TheLim)
                if (length(t1)>0) {
                    e <- max(t1)
                    v1[e] <- myv[which(myv==v1[e])+1L]
                    tSize <- r - length(myv[1:e])
                    if (!v1[e]==TheLim) {
                        v1 <- c(v1[1:e],rep(v1[e],tSize),myv[(which(myv==v1[e])+1L):n1])
                    } else {
                        v1 <- c(v1[1:e],rep(v1[e],tSize))
                    }
                } else {
                    return(m[!is.na(m[,1]),])
                }
            } else {
                r1 <- r-1L
                while (r1>=1L && !t) {
                    if (v1[r1]==TheLim) {r1 <- r1-1L; next}
                    v1[r1] <- myv[which(myv==v1[r1])+1L]
                    tSize <- r - length(myv[1:r1])
                    v1 <- c(v1[1:r1],rep(v1[r1],tSize),myv[(which(myv==v1[r1])+1L):n1])
                    t <- operator1(myfun(v1[1:r],na.rm=TRUE),li) && length(v1)>=r
                    r1 <- r1-1L
                }
                if (!t) {return(m[!is.na(m[,1]),])}
            }
        }
    }
}

以下是一些例子:

## return all 3-tuple combinations of 1 through 6 such 
## that the PRODUCT is less than 10
> Combo(n=6, r=3, v=1:6, li=10, fun1="prod", fun2="<", repeats.allowed=TRUE)
       [,1] [,2] [,3]
 [1,]    1    1    1
 [2,]    1    1    2
         .    .    .
[10,]    1    3    3
[11,]    2    2    2

## return all 3-tuple combinations of 1 through 6 such 
## that the SUM is less than 10
> Combo(n=6, r=3, v=1:6, li=10, fun1="sum", fun2="<", repeats.allowed=TRUE)
       [,1] [,2] [,3]
 [1,]    1    1    1
 [2,]    1    1    2
 [3,]    1    1    3
         .    .    .
[20,]    2    3    3
[21,]    2    3    4
[22,]    3    3    3

以下是一些涉及素数的很酷的例子:

> library(numbers)
> myps <- Primes(1000)
> system.time(t1 <- Combo(n=length(myps), r=3, v=myps, li=10^5, fun1="prod",  fun2="<", repeats.allowed=TRUE))
user  system elapsed 
0.18    0.00    0.18 
> nrow(t1)
[1] 13465

> set.seed(42)
> t1[sample(nrow(t1),5),]
     [,1] [,2] [,3]
[1,]   13   31  197
[2,]   17   19  167
[3,]    2  131  227
[4,]   11   11  751
[5,]    5   31  151

> object.size(t1)
323360 bytes

> system.time(t2 <- combinations(n=length(myps), r=3, v=myps, repeats.allowed=TRUE))
user  system elapsed 
3.63    0.00    3.68
> nrow(t2)
[1] 804440

> system.time(t3 <- t2[which(sapply(1:nrow(t2), function(x) prod(t2[x,]) < 10^5)),])
user  system elapsed 
1.55    0.00    1.54 

> nrow(t3)
[1] 13465

> object.size(t2)
19306760 bytes

正如您所看到的,Combo功能更快,并且一步完成,而combinations/sapply二重奏很慢(超过5秒)和两个笨重的步骤。 Combo函数还返回一个几乎小60倍的对象。

这是另一个很酷的例子。假设您要查找前168个素数的所有3元组(即素数<1000),使得标准差小于50.没问题(使用与上面相同的设置):

> system.time(t1 <- Combo(n=length(myps), r=3, v=myps, li=50, fun1="sd",  fun2="<", repeats.allowed=TRUE))
user  system elapsed 
1.49    0.00    1.48

> system.time(t3 <- t2[which(sapply(1:nrow(t2), function(x) sd(t2[x,]) < 50)),])
user  system elapsed 
19.89    0.00   19.89 

> nrow(t1)
[1] 22906

> nrow(t3)
[1] 22906

> all(t3==t1)
[1] TRUE

应该注意,所有功能组合都不起作用。例如,如果您允许fun1="sd"fun2=">",则上述代码将返回0个匹配项。干杯!

答案 3 :(得分:1)

一种简单的方法是使用三个for循环。这会导致你想要的结果吗?

for (x in 1:6) { 
    for (y in x:6) {
        for (z in y:6) {
            print(paste(x,y,z))
        }
    }
}