R help - Feb 2009 - learning R

I was wondering why the following doesn't work:

 > a=c(1,2)
 > names(a)=c("one","two")
 > a
one two
   1   2
 >
 > names(a[2])
[1] "two"
 >
 > names(a[2])="too"
 > names(a)
[1] "one" "two"
 > a
one two
   1   2

I must not be understanding some basic concept here.
Why doesn't the 2nd name change to "too"?

also unrelated:  if I have two vectors and I want to combine them to  
form a matrix ,is cbind (or rbind) the most direct way to do this?

e.g.

x=c(1,2,3)
y=c(3,4,5)
z=rbind(x,y)

alternatively: is there a way to make a matrix with dim=2,3 and then  
to replace the 2nd row with y

something like this (which doesn't work but perhaps there is another  
way to do the equivalent?)

attr(x,"dim")=c(2,3)
x[2,]=y

On Feb 24, 2009, at 11:36 PM, Fuchs Ira wrote:
> I was wondering why the following doesn't work:
>
> > a=c(1,2)
> > names(a)=c("one","two")
> > a
> one two
>  1   2
> >
> > names(a[2])
> [1] "two"
> >
> > names(a[2])="too"
> > names(a)
> [1] "one" "two"
> > a
> one two
>  1   2
>
> I must not be understanding some basic concept here.
> Why doesn't the 2nd name change to "too"?
I cannot tell you why, perhaps you are not actually working with the  
names of a, but I can show you that:

 > names(a)[2] <- "too"
 > a[2]
too
   2
 > a
one too
   1   2

And this is seen as well in the help page examples. The help page also  
says the following, which I cannot understand:

It is possible to update just part of the names attribute via the  
general rules: see the examples. This works because the expression  
there is evaluated as z <- "names<-"(z,
"[<-"(names(z), 3, "c2")).

>
> also unrelated:  if I have two vectors and I want to combine them to  
> form a matrix ,is cbind (or rbind) the most direct way to do this?
>
> e.g.
>
> x=c(1,2,3)
> y=c(3,4,5)
> z=rbind(x,y)
That is ok. Also effective would be:

z <- matrix( c(x,y), ncol=length(x) )
>
>
> alternatively: is there a way to make a matrix with dim=2,3 and then  
> to replace the 2nd row with y
>
> something like this (which doesn't work but perhaps there is another  
> way to do the equivalent?)
>
> attr(x,"dim")=c(2,3)
> x[2,]=y
Not sure why you are trying to do that to x since it is a vector but  
it can be done if you make it a matrix first. Take a look at these and  
see if you can figure out what is happening:

 > x <- matrix(x,2,3)  # second and third arguments of matrix function  
are nrow and ncol.
 > x
      [,1] [,2] [,3]
[1,]    1    3    2
[2,]    2    1    3

And then try:
 > x = c(1,2,3)
 > x <- matrix(x,2,3,byrow=TRUE)
 > x
      [,1] [,2] [,3]
[1,]    1    2    3
[2,]    1    2    3

And now that x is a matrix, this will work:

 > x[2,] <- y
 > x
      [,1] [,2] [,3]
[1,]    1    2    3
[2,]    3    4    5

-- David Winsemius
>
>
> ______________________________________________
> R-help at r-project.org 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.

On Feb 25, 2009, at 12:12 AM, David Winsemius wrote:
>
> On Feb 24, 2009, at 11:36 PM, Fuchs Ira wrote:
>
>>
>
>
>>
>> also unrelated:  if I have two vectors and I want to combine them  
>> to form a matrix ,is cbind (or rbind) the most direct way to do this?
>>
>> e.g.
>>
>> x=c(1,2,3)
>> y=c(3,4,5)
>> z=rbind(x,y)
>
> That is ok. Also effective would be:
>
> z <- matrix( c(x,y), ncol=length(x) )
>>
Would need byrow=TRUE for that to behave as desired.
 > z <- matrix(c(x,y), 2.3, byrow=TRUE)
 > z
      [,1] [,2] [,3]
[1,]    1    2    3
[2,]    3    4    5


-- 
David Winsemius

Hi Ira:

For your first question, under the hood of R, names<- is actually a 
function so , when you do that, you need to say names(a)[2] rather
than names(a[2]). why this is is tricky and I wouldn't do it justice if 
i tried to explain it. it's best if you do ?"names<-" at an R
prompt and
read that.

For the second one, you can rbind x with anything that of length one and 
the recycling concept in R will add the extra now but maybe there's a 
better way that someone else will hopefully send.

 
#=================================================================================
a=c(1,2)
names(a)=c("one","two")
names(a[2])
names(a)[2]<-"too"
names(a)

 
#==================================================================================
x=c(1,2,3)
y=c(3,4,5)

x <- matrix(x,nrow=1)
print(x)
x <- rbind(x,NA)
x[2,] <- y
print(x)



On Tue, Feb 24, 2009 at 11:36 PM, Fuchs Ira wrote:
> I was wondering why the following doesn't work:
>
>> a=c(1,2)
>> names(a)=c("one","two")
>> a
> one two
>   1   2
>>
>> names(a[2])
> [1] "two"
>>
>> names(a[2])="too"
>> names(a)
> [1] "one" "two"
>> a
> one two
>   1   2
>
> I must not be understanding some basic concept here.
> Why doesn't the 2nd name change to "too"?
>
> also unrelated:  if I have two vectors and I want to combine them to 
> form a matrix ,is cbind (or rbind) the most direct way to do this?
>
> e.g.
>
> x=c(1,2,3)
> y=c(3,4,5)
> z=rbind(x,y)
>
> alternatively: is there a way to make a matrix with dim=2,3 and then 
> to replace the 2nd row with y
>
> something like this (which doesn't work but perhaps there is another 
> way to do the equivalent?)
>
> attr(x,"dim")=c(2,3)
> x[2,]=y
>
> ______________________________________________
> R-help at r-project.org 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.

Fuchs Ira wrote:
> I was wondering why the following doesn't work:
>
> > a=c(1,2)
> > names(a)=c("one","two")
> > a
> one two
>   1   2
> >
> > names(a[2])
> [1] "two"
> >
> > names(a[2])="too"
> > names(a)
> [1] "one" "two"
> > a
> one two
>   1   2
>
> I must not be understanding some basic concept here.
> Why doesn't the 2nd name change to "too"?
because a[2] becomes a newly allocated vector once you make the
assignment, and so the assignment does not affect a.  however:

    names(a)[2] = 'too'
   
will affect a the way you seem to wish.
>
> also unrelated:  if I have two vectors and I want to combine them to
> form a matrix ,is cbind (or rbind) the most direct way to do this?
>
> e.g.
>
> x=c(1,2,3)
> y=c(3,4,5)
> z=rbind(x,y)
>
> alternatively: is there a way to make a matrix with dim=2,3 and then
> to replace the 2nd row with y
>
> something like this (which doesn't work but perhaps there is another
> way to do the equivalent?)
>
> attr(x,"dim")=c(2,3)
> x[2,]=y
you can do this:

    z = matrix(c(x, y), nrow=2, ncol=3, byrow=TRUE)

but rbind seems much simpler.

vQ

David, Wacek: Just so everyone knows, I just looked and this is 
explained quite clearly in the R Language Reference manual, very 
similarly to what Wacek did below.

On Wed, Feb 25, 2009 at  3:42 AM, Wacek Kusnierczyk wrote:
> David Winsemius wrote:
>>
>> On Feb 24, 2009, at 11:36 PM, Fuchs Ira wrote:
>>
>>> I was wondering why the following doesn't work:
>>>
>>>> a=c(1,2)
>>>> names(a)=c("one","two")
>>>> a
>>> one two
>>>  1   2
>>>>
>>>> names(a[2])
>>> [1] "two"
>>>>
>>>> names(a[2])="too"
>>>> names(a)
>>> [1] "one" "two"
>>>> a
>>> one two
>>>  1   2
>>>
>>> I must not be understanding some basic concept here.
>>> Why doesn't the 2nd name change to "too"?
>>
>> I cannot tell you why, perhaps you are not actually working with the
>> names of a, but I can show you that:
>>
>>> names(a)[2] <- "too"
>>> a[2]
>> too
>>   2
>>> a
>> one too
>>   1   2
>>
>> And this is seen as well in the help page examples. The help page 
>> also
>> says the following, which I cannot understand:
>>
>> It is possible to update just part of the names attribute via the
>> general rules: see the examples. This works because the expression
>> there is evaluated as z <- "names<-"(z,
"[<-"(names(z), 3, "c2")).
>>
>
> the following:
>
>     names(a[2]) = 'foo'
>
> has (partially) a functional flavour, in that you assign to the names 
> of
> a *copy* of a part of a, while
>
>     names(a)[2] = 'foo'
>
> does not have the flavour, in that you assign to the names of a;  it
> seems, according to the man page you quote, to be equivalent to:
>
>     a = 'names<-'(a, '[<-.'(names(a), 2,
'foo'))
>
> which proceeds as follows:
>
>     tmp1 = names(a)
>     # get a copy of the names of a, no effect on a
>
>     tmp2 = '[<-'(tmp1, 2, 'foo')
>     # get a copy of tmp1 with the second element replaced with
'foo'
>     # no effect on either a or tmp1
>
>     tmp3 = 'names<-'(a, tmp2)
>     # get a copy of a with its names replaced with tmp2
>     # no effect on either a, tmp1, or tmp2
>
>     a = tmp3
>     # backassign the result to a
>
> vQ
>
> ______________________________________________
> R-help at r-project.org 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.

Hi Wacek: Somewhere I remember reading that environments have 
functionality like lists EXCEPT for the names part. IIRC, I think that I 
read this in the R Language Reference manual also.



On Wed, Feb 25, 2009 at  4:32 AM, Wacek Kusnierczyk wrote:
> a quick follow-up:
>
>     e = new.env()
>     e$a = 1
>     names(e)
>     # NULL
>     names(e) = 'a'
>     # error in names(e) = "foo" : names() applied to a non-vector
>
> this is surprising.  names(e) 'works', there is no complaint, but
when
> names<- is used, the error is about the use of names, not names<-.
>
> btw. ?names says:
>
> "Description:
>
>      Functions to get or set the names of an object.
>
> Usage:
>
>      names(x)
>      names(x) <- value
>
> Arguments:
>
>        x: an R object.
> "
>
> and there is no clarification in the rest of the page that x cannot be
> an environment, or that it has to be a vector.  furthermore:
>
>     p = pairlist(a=1)
>     names(p)
>     # "a"
>     names(p) = 'b'
>     # fine
>     is.vector(p)
>     # FALSE
>
> which is incoherent with the above error message, in that p is *not* a
> vector.
>
> vQ
>
>
>
> Wacek Kusnierczyk wrote:
>>
>> the following:
>>
>>     names(a[2]) = 'foo'
>>
>> has (partially) a functional flavour, in that you assign to the names 
>> of
>> a *copy* of a part of a, while
>>
>>     names(a)[2] = 'foo'
>>
>> does not have the flavour, in that you assign to the names of a;  it
>> seems, according to the man page you quote, to be equivalent to:
>>
>>     a = 'names<-'(a, '[<-.'(names(a), 2,
'foo'))
>>
>> which proceeds as follows:
>>
>>     tmp1 = names(a)
>>     # get a copy of the names of a, no effect on a
>>
>>     tmp2 = '[<-'(tmp1, 2, 'foo')
>>     # get a copy of tmp1 with the second element replaced with
'foo'
>>     # no effect on either a or tmp1
>>
>>     tmp3 = 'names<-'(a, tmp2)
>>     # get a copy of a with its names replaced with tmp2
>>     # no effect on either a, tmp1, or tmp2
>>
>>     a = tmp3
>>     # backassign the result to a
>>
>
> ______________________________________________
> R-help at r-project.org 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.

Thanks Berwin. You're correct in that I meant the R Language 
Definition. Well, it may be a draft but I read it for the first time a 
few months ago and
it was very enlightening so, whether it's a draft or not, I highly 
recommend it. (  but not for total beginners. The  R-intro is better for 
a more gentle introduction ).


On Wed, Feb 25, 2009 at  4:34 AM, Berwin A Turlach wrote:
> On Wed, 25 Feb 2009 09:57:36 +0100
> Wacek Kusnierczyk <Waclaw.Marcin.Kusnierczyk at idi.ntnu.no> wrote:
>
>> markleeds at verizon.net wrote:
>>>  David, Wacek: Just so everyone knows, I just looked and this is
>>> explained quite clearly in the R Language Reference manual, very
>>> similarly to what Wacek did below.
>>>
>> thanks, that's good, because i made it up following the page quoted
>> by
>> david, and if i'm flamed for nonsense, the langref has to be flamed
>> too.
>
> Actually, no.  :)
>
> I do not remember the details, but I once tried to use the "R Language
> Definition" (I presume this is what we are talking about) to support 
> my
> argument about some issue I had at that time.  It was pointed out to 
> me
> that the "R Language Definition" is a draft, hence not
authoritative;
> and it seems its status of being a draft has not changed since.
>
> Cheers,
>
> 	Berwin

markleeds at verizon.net wrote:
>  Hi Wacek: Somewhere I remember reading that environments have
> functionality like lists EXCEPT for the names part. IIRC, I think that
> I read this in the R Language Reference manual also.
>
>
this would be a confused and confusing statement, unless 'functionality'
has some rather vague sense here.  (which would not be very
surprising.)  if you find it, i humbly suggest that you politely report
it as a statement to be fixed.

there are important differences between environments and lists, in
particular wrt. to assignment:

    l = list(a=1)
    ll = l
    l$a = 0
    l$a == ll$a
    # FALSE

    e = new.env()
    e$a = 1
    ee = e
    e$a = 0
    e$a == ee$a
    # TRUE

this is a fundamental difference, one that could/should be more
carefully acknowledged in functions that return objects represented with
environments as opposed to lists (e.g., srcfile).

vQ