thr3ads.net - R help - [R] comparing two matrices [Jan 2007]

If this information is useful, please help other people find it:
Share via:

Adrian Dusa

2007-Jan-20 22:14 UTC

[R] comparing two matrices

Dear helpeRs,

I have two matrices:
mat1 <- expand.grid(0:2, 0:2, 0:2)
mat2 <- aa[c(19, 16, 13, 24, 8), ]

where mat2 is always a subset of mat1

I need to find the corersponding row numbers in mat1 for each row in mat2.
For this I have the following code:

apply(mat2, 1, function(x) {
    which(apply(mat1, 1, function(y) {
        sum(x == y)
        }) == ncol(mat1))
    })

The code is vectorized, but I wonder if there is a simpler (hence faster) 
matrix computation that I miss.

Thank you,
Adrian


-- 
Adrian Dusa
Romanian Social Data Archive
1, Schitu Magureanu Bd
050025 Bucharest sector 5
Romania
Tel./Fax: +40 21 3126618 \
          +40 21 3120210 / int.101

Christos Hatzis

2007-Jan-20 22:56 UTC

head link

[R] comparing two matrices

Here is a slightly more compact version of your function which might run
faster (I did not test timings) since it does not use the sum:

apply(mat2, 1, function(x) which(apply(mat1, 1, function(y) all(x == y)) =TRUE))

-Christos
 
-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Adrian Dusa
Sent: Saturday, January 20, 2007 5:15 PM
To: r-help at stat.math.ethz.ch
Subject: [R] comparing two matrices


Dear helpeRs,

I have two matrices:
mat1 <- expand.grid(0:2, 0:2, 0:2)
mat2 <- aa[c(19, 16, 13, 24, 8), ]

where mat2 is always a subset of mat1

I need to find the corersponding row numbers in mat1 for each row in mat2.
For this I have the following code:

apply(mat2, 1, function(x) {
    which(apply(mat1, 1, function(y) {
        sum(x == y)
        }) == ncol(mat1))
    })

The code is vectorized, but I wonder if there is a simpler (hence faster)
matrix computation that I miss.

Thank you,
Adrian


--
Adrian Dusa
Romanian Social Data Archive
1, Schitu Magureanu Bd
050025 Bucharest sector 5
Romania
Tel./Fax: +40 21 3126618 \
          +40 21 3120210 / int.101

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Marc Schwartz

2007-Jan-20 23:06 UTC

head link

[R] comparing two matrices

On Sun, 2007-01-21 at 00:14 +0200, Adrian Dusa wrote:> Dear helpeRs,
> 
> I have two matrices:
> mat1 <- expand.grid(0:2, 0:2, 0:2)
> mat2 <- aa[c(19, 16, 13, 24, 8), ]
> 
> where mat2 is always a subset of mat1
> 
> I need to find the corersponding row numbers in mat1 for each row in mat2.
> For this I have the following code:
> 
> apply(mat2, 1, function(x) {
>     which(apply(mat1, 1, function(y) {
>         sum(x == y)
>         }) == ncol(mat1))
>     })
> 
> The code is vectorized, but I wonder if there is a simpler (hence faster) 
> matrix computation that I miss.
> 
> Thank you,
> Adrian

I have not fully tested this, but how about:

mat1 <- matrix(1:20, ncol = 4, byrow = TRUE)
mat2 <- matrix(1:60, ncol = 4, byrow = TRUE)
mat2 <- mat2[sample(15), ]
> mat1     [,1] [,2] [,3] [,4]
[1,]    1    2    3    4
[2,]    5    6    7    8
[3,]    9   10   11   12
[4,]   13   14   15   16
[5,]   17   18   19   20
> mat2      [,1] [,2] [,3] [,4]
 [1,]   13   14   15   16
 [2,]    5    6    7    8
 [3,]   41   42   43   44
 [4,]   17   18   19   20
 [5,]   21   22   23   24
 [6,]   25   26   27   28
 [7,]   53   54   55   56
 [8,]    9   10   11   12
 [9,]   57   58   59   60
[10,]   33   34   35   36
[11,]   49   50   51   52
[12,]   45   46   47   48
[13,]    1    2    3    4
[14,]   29   30   31   32
[15,]   37   38   39   40
> which(apply(matrix(mat2 %in% mat1, dim(mat2)), 1, all))[1]  1  2  4  8 13


HTH,

Marc Schwartz

Adrian Dusa

2007-Jan-21 09:53 UTC

head link

[R] comparing two matrices

Hello Marc and Dimitris,

There was an error in my first example (therefore not reproducible), so
mat1 <- expand.grid(0:2, 0:2, 0:2)
mat2 <- mat1[c(19, 16, 13, 24, 8), ]

Your solution works if and only if the elements in both matrices are unique. 
Unfortunately, it does not apply for my matrices where elements do repeat 
(only the rows are unique).
> which(apply(matrix(mat1 %in% mat2, dim(mat1)), 1, all))integer(0)
> which((mat1 %in% mat2)[1:nrow(mat1)])integer(0)


Another solution would be using base 3 operations:
mat1 <- expand.grid(0:2, 0:2, 0:2)[, 3:1]
mat2 <- mat1[c(19, 16, 13, 24, 8), ]

mylines <- mat2[, 1]
for (i in 2:ncol(mat2)) {mylines <- 3*mylines + mat2[, i]}
mylines + 1
[1] 19 16 13 24  8


I was still hoping for a direct matrix function to avoid the for() loop.
Thanks,
Adrian


On Sunday 21 January 2007 01:06, Marc Schwartz wrote:> On Sun, 2007-01-21 at 00:14 +0200, Adrian Dusa wrote:
> > Dear helpeRs,
> >
> > I have two matrices:
> > mat1 <- expand.grid(0:2, 0:2, 0:2)
> > mat2 <- aa[c(19, 16, 13, 24, 8), ]
> >
> > where mat2 is always a subset of mat1
> >
> > I need to find the corersponding row numbers in mat1 for each row in
> > mat2. For this I have the following code:
> >
> > apply(mat2, 1, function(x) {
> >     which(apply(mat1, 1, function(y) {
> >         sum(x == y)
> >         }) == ncol(mat1))
> >     })
> >
> > The code is vectorized, but I wonder if there is a simpler (hence
faster)
> > matrix computation that I miss.
> >
> > Thank you,
> > Adrian
>
> I have not fully tested this, but how about:
>
> mat1 <- matrix(1:20, ncol = 4, byrow = TRUE)
> mat2 <- matrix(1:60, ncol = 4, byrow = TRUE)
> mat2 <- mat2[sample(15), ]
>
> > mat1
>
>      [,1] [,2] [,3] [,4]
> [1,]    1    2    3    4
> [2,]    5    6    7    8
> [3,]    9   10   11   12
> [4,]   13   14   15   16
> [5,]   17   18   19   20
>
> > mat2
>
>       [,1] [,2] [,3] [,4]
>  [1,]   13   14   15   16
>  [2,]    5    6    7    8
>  [3,]   41   42   43   44
>  [4,]   17   18   19   20
>  [5,]   21   22   23   24
>  [6,]   25   26   27   28
>  [7,]   53   54   55   56
>  [8,]    9   10   11   12
>  [9,]   57   58   59   60
> [10,]   33   34   35   36
> [11,]   49   50   51   52
> [12,]   45   46   47   48
> [13,]    1    2    3    4
> [14,]   29   30   31   32
> [15,]   37   38   39   40
>
> > which(apply(matrix(mat2 %in% mat1, dim(mat2)), 1, all))
>
> [1]  1  2  4  8 13
>
>
> HTH,
>
> Marc Schwartz
-- 
Adrian Dusa
Romanian Social Data Archive
1, Schitu Magureanu Bd
050025 Bucharest sector 5
Romania
Tel./Fax: +40 21 3126618 \
          +40 21 3120210 / int.101

Dimitris Rizopoulos

2007-Jan-21 10:04 UTC

head link

[R] comparing two matrices

I think the following should work in your case:

mat1 <- data.matrix(expand.grid(0:2, 0:2, 0:2))
mat2 <- mat1[c(19, 16, 13, 24, 8), ]
############
ind1 <- apply(mat1, 1, paste, collapse = "/")
ind2 <- apply(mat2, 1, paste, collapse = "/")
match(ind2, ind1)


I hope it helps.

Best,
Dimitris


Quoting Adrian Dusa <dusa.adrian at gmail.com>:
> Hello Marc and Dimitris,
>
> There was an error in my first example (therefore not reproducible), so
> mat1 <- expand.grid(0:2, 0:2, 0:2)
> mat2 <- mat1[c(19, 16, 13, 24, 8), ]
>
> Your solution works if and only if the elements in both matrices are
unique.
> Unfortunately, it does not apply for my matrices where elements do repeat
> (only the rows are unique).
>
>> which(apply(matrix(mat1 %in% mat2, dim(mat1)), 1, all))
> integer(0)
>
>> which((mat1 %in% mat2)[1:nrow(mat1)])
> integer(0)
>
>
> Another solution would be using base 3 operations:
> mat1 <- expand.grid(0:2, 0:2, 0:2)[, 3:1]
> mat2 <- mat1[c(19, 16, 13, 24, 8), ]
>
> mylines <- mat2[, 1]
> for (i in 2:ncol(mat2)) {mylines <- 3*mylines + mat2[, i]}
> mylines + 1
> [1] 19 16 13 24  8
>
>
> I was still hoping for a direct matrix function to avoid the for() loop.
> Thanks,
> Adrian
>
>
> On Sunday 21 January 2007 01:06, Marc Schwartz wrote:
>> On Sun, 2007-01-21 at 00:14 +0200, Adrian Dusa wrote:
>> > Dear helpeRs,
>> >
>> > I have two matrices:
>> > mat1 <- expand.grid(0:2, 0:2, 0:2)
>> > mat2 <- aa[c(19, 16, 13, 24, 8), ]
>> >
>> > where mat2 is always a subset of mat1
>> >
>> > I need to find the corersponding row numbers in mat1 for each row
in
>> > mat2. For this I have the following code:
>> >
>> > apply(mat2, 1, function(x) {
>> >     which(apply(mat1, 1, function(y) {
>> >         sum(x == y)
>> >         }) == ncol(mat1))
>> >     })
>> >
>> > The code is vectorized, but I wonder if there is a simpler (hence
faster)
>> > matrix computation that I miss.
>> >
>> > Thank you,
>> > Adrian
>>
>> I have not fully tested this, but how about:
>>
>> mat1 <- matrix(1:20, ncol = 4, byrow = TRUE)
>> mat2 <- matrix(1:60, ncol = 4, byrow = TRUE)
>> mat2 <- mat2[sample(15), ]
>>
>> > mat1
>>
>>      [,1] [,2] [,3] [,4]
>> [1,]    1    2    3    4
>> [2,]    5    6    7    8
>> [3,]    9   10   11   12
>> [4,]   13   14   15   16
>> [5,]   17   18   19   20
>>
>> > mat2
>>
>>       [,1] [,2] [,3] [,4]
>>  [1,]   13   14   15   16
>>  [2,]    5    6    7    8
>>  [3,]   41   42   43   44
>>  [4,]   17   18   19   20
>>  [5,]   21   22   23   24
>>  [6,]   25   26   27   28
>>  [7,]   53   54   55   56
>>  [8,]    9   10   11   12
>>  [9,]   57   58   59   60
>> [10,]   33   34   35   36
>> [11,]   49   50   51   52
>> [12,]   45   46   47   48
>> [13,]    1    2    3    4
>> [14,]   29   30   31   32
>> [15,]   37   38   39   40
>>
>> > which(apply(matrix(mat2 %in% mat1, dim(mat2)), 1, all))
>>
>> [1]  1  2  4  8 13
>>
>>
>> HTH,
>>
>> Marc Schwartz
>
> --
> Adrian Dusa
> Romanian Social Data Archive
> 1, Schitu Magureanu Bd
> 050025 Bucharest sector 5
> Romania
> Tel./Fax: +40 21 3126618 \
>           +40 21 3120210 / int.101
>
>


Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm

Adrian Dusa

2007-Jan-21 10:17 UTC

head link

[R] comparing two matrices

On Sunday 21 January 2007 12:04, Dimitris Rizopoulos
wrote:> I think the following should work in your case:
>
> mat1 <- data.matrix(expand.grid(0:2, 0:2, 0:2))
> mat2 <- mat1[c(19, 16, 13, 24, 8), ]
> ############
> ind1 <- apply(mat1, 1, paste, collapse = "/")
> ind2 <- apply(mat2, 1, paste, collapse = "/")
> match(ind2, ind1)

Oh yes, I thought about that too.
It works fast enough for small matrices, but I deal with very large ones. 
Using paste() on such matrices decreases the speed dramatically.

Thanks again,
Adrian

-- 
Adrian Dusa
Romanian Social Data Archive
1, Schitu Magureanu Bd
050025 Bucharest sector 5
Romania
Tel./Fax: +40 21 3126618 \
          +40 21 3120210 / int.101

Bill.Venables at csiro.au

2007-Jan-21 22:44 UTC

head link

[R] comparing two matrices

But this is using paste() the wrong way round. A better way would be
> join <- function(x) do.call("paste", c(as.data.frame(x), sep =
"\r"))
> which(join(mat1) %in% join(mat2))[1]  8 13 16 19 24

This is essentially the technique used by duplicated.data.frame

Bill Venables 

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Adrian Dusa
Sent: Sunday, 21 January 2007 8:17 PM
To: Dimitris Rizopoulos
Cc: marc_schwartz at comcast.net; r-help at stat.math.ethz.ch
Subject: Re: [R] comparing two matrices

On Sunday 21 January 2007 12:04, Dimitris Rizopoulos
wrote:> I think the following should work in your case:
>
> mat1 <- data.matrix(expand.grid(0:2, 0:2, 0:2))
> mat2 <- mat1[c(19, 16, 13, 24, 8), ]
> ############
> ind1 <- apply(mat1, 1, paste, collapse = "/")
> ind2 <- apply(mat2, 1, paste, collapse = "/")
> match(ind2, ind1)

Oh yes, I thought about that too.
It works fast enough for small matrices, but I deal with very large
ones. 
Using paste() on such matrices decreases the speed dramatically.

Thanks again,
Adrian

-- 
Adrian Dusa
Romanian Social Data Archive
1, Schitu Magureanu Bd
050025 Bucharest sector 5
Romania
Tel./Fax: +40 21 3126618 \
          +40 21 3120210 / int.101

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

R help - Jan 2007 - comparing two matrices

[R] comparing two matrices

[R] comparing two matrices

[R] comparing two matrices

[R] comparing two matrices

[R] comparing two matrices

[R] comparing two matrices

[R] comparing two matrices