Brobdingnagian numbers
brob.Rd
Create, coerce to or test for a Brobdingnagian object
Usage
brob(x = double(), positive)
as.brob(x)
is.brob(x)
Details
Function as.brob()
is the user's workhorse: use this to coerce
numeric vectors to brob
s.
Function is.brob()
tests for its arguments being of class
brob
.
Function brob()
takes argument x
and returns a brob
formally equal to \(e^x\); set argument positive
to
FALSE
to return \(-e^x\). Thus calling function
exp(x)
simply returns brob(x)
. This function is not
really intended for the end user: it is confusing and includes no
argument checking. In general numerical work, use function
as.brob()
instead, although be aware that if you really really
want \(e^{10^7}\), you should use brob(1e7)
;
this would be an exact representation.
Note
Real numbers are represented by two objects: a real, holding the logarithm of their absolute values; and a logical, indicating the sign. Multiplication and exponentiation are easy: the challenge is addition. This is achieved using the (trivial) identity \(\log(e^x+e^y)=x+\log(1+e^{y-x})\) where, WLOG, \(y<x\).
Complex numbers are stored as a pair of brob
s: objects of class
glub
.
The package is a simple example of S4 methods. However, it could be viewed as a cautionary tale: the underlying R concepts are easy yet the S4 implementation is long and difficult. I would not recommend using S4 methods for a package as simple as this; S3 methods would have been perfectly adequate. I would suggest that S4 methods should only be used when S3 methods are demonstrably inadequate.
The package has poor handling of NA
and NaN
. Currently,
as.brob(1) + as.brob(c(1,NA))
returns an error.