[R] Force evaluation of a symbol when a function is created

[Bert Gunter]

Thanks Duncan. That clarifies it!

-- Bert

On Tue, Aug 7, 2012 at 12:28 PM, Duncan Murdoch
<murdoch.duncan at gmail.com> wrote:
> On 12-08-07 10:46 AM, Bert Gunter wrote:
>>
>> Duncan, et.al:
>>
>> Many thanks: let the closure do the work automatically rather than
>> manually manipulating it.
>>
>> However, in the spirit of the OP's original request, I believe the
>> call would be:
>>
>>> Y <- 3  ## That is, Y gets a value at some prior point, perhaps
>>> programmatically.
>>>    F <- multiply_by_Y(Y) # ... F picks up this value "implicitly" -- no
>>> need for explicit assignment.
>>
>>
>> But then there is no need for force(), is there?
>
>
> You still need force:
>
> fy <- function(Y)function(x) x*Y
> Y <- 2
> F <- fy(Y)
> Y <- 3
> F(5)
>
> This will print 15, because F only contains a promise to evaluate Y, it
> hasn't been evaluated until the very last line, and by that time Y has been
> changed to 3.
>
> If you are going to construct functions in functions, and their results
> depend on the arguments to the constructor, then it's almost always a good
> idea to force the arguments.  Sometimes it isn't necessary (the value will
> be forced implicitly), and in some rare circumstances you might want to
> capture the promise instead of its value, but it's generally a good idea.
> It is a fairly cheap operation.
>
> Duncan Murdoch
>
>>
>>> fy <- function(Y)function(x) x*Y
>>> Y <- 2
>>> F <- fy(Y)
>>> F(5)
>>
>> [1] 10
>>>
>>> Y <- 3
>>> F(5)
>>
>> [1] 10
>>>
>>> G <- fy(Y)
>>> G(5)
>>
>> [1] 15
>>
>> That is, one simply relies on lexical scoping/closures to "retain" the
>> value of  Y used as a free variable in  function(x)x*Y when it is
>> defined. No need to explicitly force() it. If wrong, I would be
>> grateful for correction. This appears to me to duplicate the Matlab
>> behavior rather closely.
>>
>>
>> -- Bert
>>
>> On Tue, Aug 7, 2012 at 3:48 AM, Duncan Murdoch <murdoch.duncan at gmail.com>
>> wrote:
>>>
>>> Here's one more way.  It seems to me this is the most R-like way to do
>>> what
>>> you want:
>>>
>>>    multiply_by_Y <- function(Y) {
>>>      force(Y)
>>>      function(x) x*Y
>>>    }
>>>
>>>    F <- multiply_by_Y(3)
>>>
>>> The "force" call forces Y to be evaluated at that point, so its value is
>>> fixed from that point forward.
>>>
>>> Duncan Murdoch
>>>
>>> On 12-08-06 5:07 PM, Schoenfeld, David Alan,Ph.D.,Biostatistics wrote:
>>>>
>>>>
>>>>
>>>> I am porting a program in matlab to R,
>>>> The problem is that Matlab has a feature where symbols that aren't
>>>> arguments are evaluated immediately.
>>>> That is:
>>>> Y=3
>>>> F=@(x) x*Y
>>>>
>>>> Will yield a function such that F(2)=6.
>>>> If later say. Y=4 then F(2) will still equal 6.
>>>>
>>>> R on the other hand has lazy evaluation.
>>>> F<-function(x){x*Y}
>>>> Will do the following
>>>> Y=3
>>>> F(2)=6
>>>> Y=4
>>>> F(2)=8.
>>>> Does anyone know of away to defeat lazy evaluation in R so that I can
>>>> easily simulate the Matlab behavior.  I know that I can live without
>>>> this in
>>>> ordinary programming but it would make my port much easier.
>>>>
>>>> Thanks.
>>>>
>>>>
>>>>
>>>>
>>>> The information in this e-mail is intended only for the
>>>> ...{{dropped:14}}
>>>>
>>>> ______________________________________________
>>>> 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.
>>>>
>>>
>>> ______________________________________________
>>> 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.
>>
>>
>>
>>
>



-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm