R help - Dec 2009 - Newbie: colSums() compared with Matlab's sum()

Hi all,

I'm trying to learn R after years of Matlab's experience. Here is an
issue I couldn't solve today.

Consider the following piece of code (written by memory):

for(i in 1:n){
    submat <- data[1:i,]
    C <- colSums(submat)
    }

The problem is that at the first iteration, data[1:1,] reduces to a
vector and colSums returns an error. This sounds really strange to me
because a sum-over-columns operation should have no problem working with
columns of length 1. Matlab's "sum()" works just fine in such
case.

The error says that I need an at least 2D array. So I try using
matrix(data[1:i,]). Unfortunately this returns a column instead of a
row. So I could do t(matrix(data[1:i,])), but this would transpose the
also the sub-matrices from the 2nd iteration on. I could fix everything
with some "if" but it would be really terrible code :-/.

I'm sure I'm missing something because my neurons stick with Matlab
behaviour. Can you help me to understand what's wrong with my reasoning?

Thank you very much,
Francesco.

replace data[1:i,] by data[1:i,drop=FALSE].

b

On Dec 24, 2009, at 12:46 AM, Francesco Napolitano wrote:
> Hi all,
> 
> I'm trying to learn R after years of Matlab's experience. Here is
an
> issue I couldn't solve today.
> 
> Consider the following piece of code (written by memory):
> 
> for(i in 1:n){
>    submat <- data[1:i,]
>    C <- colSums(submat)
>    }
> 
> The problem is that at the first iteration, data[1:1,] reduces to a
> vector and colSums returns an error. This sounds really strange to me
> because a sum-over-columns operation should have no problem working with
> columns of length 1. Matlab's "sum()" works just fine in such
case.
> 
> The error says that I need an at least 2D array. So I try using
> matrix(data[1:i,]). Unfortunately this returns a column instead of a
> row. So I could do t(matrix(data[1:i,])), but this would transpose the
> also the sub-matrices from the 2nd iteration on. I could fix everything
> with some "if" but it would be really terrible code :-/.
> 
> I'm sure I'm missing something because my neurons stick with Matlab
> behaviour. Can you help me to understand what's wrong with my
reasoning?
> 
> Thank you very much,
> Francesco.
> 
> ______________________________________________
> 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.

Use

   submat <- data[1:i,, drop=FALSE]

/H

On Wed, Dec 23, 2009 at 6:46 PM, Francesco Napolitano
<franapoli at gmail.com> wrote:
> Hi all,
>
> I'm trying to learn R after years of Matlab's experience. Here is
an
> issue I couldn't solve today.
>
> Consider the following piece of code (written by memory):
>
> for(i in 1:n){
> ? ?submat <- data[1:i,]
> ? ?C <- colSums(submat)
> ? ?}
>
> The problem is that at the first iteration, data[1:1,] reduces to a
> vector and colSums returns an error. This sounds really strange to me
> because a sum-over-columns operation should have no problem working with
> columns of length 1. Matlab's "sum()" works just fine in such
case.
>
> The error says that I need an at least 2D array. So I try using
> matrix(data[1:i,]). Unfortunately this returns a column instead of a
> row. So I could do t(matrix(data[1:i,])), but this would transpose the
> also the sub-matrices from the 2nd iteration on. I could fix everything
> with some "if" but it would be really terrible code :-/.
>
> I'm sure I'm missing something because my neurons stick with Matlab
> behaviour. Can you help me to understand what's wrong with my
reasoning?
>
> Thank you very much,
> Francesco.
>
> ______________________________________________
> 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 Dec 23, 2009, at 9:46 PM, Francesco Napolitano wrote:
> Hi all,
>
> I'm trying to learn R after years of Matlab's experience. Here is
an
> issue I couldn't solve today.
>
> Consider the following piece of code (written by memory):
>
> for(i in 1:n){
>    submat <- data[1:i,]
>    C <- colSums(submat)
>    }
>
In R the loop is not necessary, even confusing as you are demonstrating:

 > mat <- matrix(rnorm(100), nrow=10)  # 10 x 10
 > colSums(mat[1:5, ])                 # sums of 1st 5 rows
  [1]  3.33331735  0.86672248 -3.10971483  1.23620455 -0.31887421  
-0.50544837
  [7]  3.13636155  0.02175862 -2.18816961 -1.31760196

> The problem is that at the first iteration, data[1:1,] reduces to a
> vector and colSums returns an error. This sounds really strange to me
> because a sum-over-columns operation should have no problem working  
> with
> columns of length 1. Matlab's "sum()" works just fine in such
case.
>
> The error says that I need an at least 2D array. So I try using
> matrix(data[1:i,]). Unfortunately this returns a column instead of a
> row. So I could do t(matrix(data[1:i,])), but this would transpose the
> also the sub-matrices from the 2nd iteration on. I could fix  
> everything
> with some "if" but it would be really terrible code :-/.
>
> I'm sure I'm missing something because my neurons stick with Matlab
> behaviour. Can you help me to understand what's wrong with my  
> reasoning?
>
> Thank you very much,
> Francesco.
>
> ______________________________________________
> 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.
David Winsemius, MD
Heritage Laboratories
West Hartford, CT