R: is there a way to cbind non-numeric columns

Question

library(Rmpfr)
mynumber <- new("mpfr", .Data = list(new("mpfr1", prec = 50L, exp = c(1045L, 
0L), sign = 1L, d = c(151748608L, -358118319L)), new("mpfr1", 
    prec = 50L, exp = c(20L, 0L), sign = 1L, d = c(-1114947584L, 
    -1905679017L)), new("mpfr1", prec = 50L, exp = c(-55L, -1L
), sign = 1L, d = c(-1449918464L, -906197701L)), new("mpfr1", 
    prec = 50L, exp = c(221L, 0L), sign = 1L, d = c(819707904L, 
    -1329031570L))))

mynumber is a class mpfr object with 4 numbers in it. I want to cbind mynumber with a column of 0s, i.e.

> cbind(rep(0, 4), mynumber)
       mynumber
[1,] 0 ?       
[2,] 0 ?       
[3,] 0 ?       
[4,] 0 ?  

This gives me ???? in the second column, so I tried to change mynumber to class numeric first

mydata <- cbind(rep(0, 4), sapply(mynumber, asNumeric))
> mydata
     [,1]         [,2]
[1,]    0          Inf
[2,]    0 5.833223e+05
[3,]    0 2.189941e-17
[4,]    0 2.327185e+66

However, since the first number in mynumber is really big, using asNumeric changed it into Inf instead.

Edit: My ultimate goal is to run:

mydata <- cbind(rep(0, 4), sapply(mynumber, asNumeric))
> mydata/rowSums(mydata)
     [,1] [,2]
[1,]    0  NaN
[2,]    0    1
[3,]    0    1
[4,]    0    1

and not have it print out NaN.

Answer 1

One option is to wrap with list and then create a tibble/data.frame object as cbind converts to a matrix and matrix can hold only a single class

library(tibble)
tibble(col1 = 0, col2 = list(mynumber))
# A tibble: 1 x 2
#   col1 col2  
#  <dbl> <list>
#1     0 <mpfr>

---

cbind on even a character and numeric class returns character for all the columns and this is not a good option when the vectors to bind are of different class

cbind(letters[1:4], 1:4)

---

By checking the methods for cbind after loading the package

methods('cbind')
#[1] cbind,ANY-method     cbind,Mnumber-method cbind.bigq*          cbind.bigz*         
#[5] cbind.data.frame     cbind.grouped_df*    cbind.ts*       

So, if it was using the correct cbind methods for Mnumber, it should not have given Inf

cbind(rep(0,4),  mynumber)
#'mpfrMatrix' of dim(.) =  (4, 2) of precision  50 .. 53  bits 
#     [,1] [,2]                   
#[1,]   0. 3.4556867084990952e+314
#[2,]   0.      583322.33392099757
#[3,]   0.  2.1899410233914937e-17
#[4,]   0.  2.3271850367397449e+66

Or make use of the recycling of value

cbind(0,  mynumber)
#'mpfrMatrix' of dim(.) =  (4, 2) of precision  50 .. 53  bits 
#     [,1] [,2]                   
#[1,]   0. 3.4556867084990952e+314
#[2,]   0.      583322.33392099757
#[3,]   0.  2.1899410233914937e-17
#[4,]   0.  2.3271850367397449e+66

---

Also, if we check the masked the functions, when we load the package, it says

The following objects are masked from ‘package:base’:

cbind, pmax, pmin, rbind

By using the cbind from base, the ? can be replicated. It is possible that for the OP, the cbind is from base

base::cbind(0, mynumber)
#       mynumber
#[1,] 0 ?       
#[2,] 0 ?       
#[3,] 0 ?       
#[4,] 0 ?       

If the cbind from Rmpfr got masked, then use ::

mydata <- Rmpfr::cbind(0, mynumber)
mydata
#'mpfrMatrix' of dim(.) =  (4, 2) of precision  50 .. 53  bits 
#     [,1] [,2]                   
#[1,]   0. 3.4556867084990952e+314
#[2,]   0.      583322.33392099757
#[3,]   0.  2.1899410233914937e-17
#[4,]   0.  2.3271850367397449e+66




mydata/rowSums(mydata)
#'mpfrMatrix' of dim(.) =  (4, 2) of precision  53   bits 
#     [,1] [,2]              
#[1,]   0. 1.0000000000000000
#[2,]   0. 1.0000000000000000
#[3,]   0. 1.0000000000000000
#[4,]   0. 1.0000000000000000

Answer 2

I wonder if there is something wrong with your created object. When I changed it to make sure each exp field only had a single value, I got the desired result:

library(Rmpfr)

x <- new("mpfr", .Data = list(new("mpfr1", prec = 50L, exp = 1045L, 
    sign = 1L, d = c(151748608L, -358118319L)), new("mpfr1", 
    prec = 50L, exp = 20L, sign = 1L, d = c(-1114947584L, -1905679017L
    )), new("mpfr1", prec = 50L, exp = -55L, sign = 1L, d = c(-1449918464L, 
-906197701L)), new("mpfr1", prec = 50L, exp = 221L, sign = 1L, 
    d = c(819707904L, -1329031570L))))

cbind(c(0, 0, 0, 0), x)
#> 'mpfrMatrix' of dim(.) =  (4, 2) of precision  50 .. 53  bits 
#>      [,1] [,2]                   
#> [1,]   0. 3.4556867084990952e+314
#> [2,]   0.      583322.33392099757
#> [3,]   0.  2.1899410233914937e-17
#> [4,]   0.  2.3271850367397449e+66

^{Created on 2020-02-29 by the reprex package (v0.3.0)}

Answer 3

zeros <- mpfr(c(0,0,0,0), precBits = getPrec(mynumber))
M <- cbind(mynumber, zeros)
M %*% t(M) # just to try