R help - Jun 2021 - Unexplained behaviour of .Last.value

Hello:

Here is a bit of R-code:
> mvr
[1] 2.2658
> (sin(.Last.value)+sin(2*.Last.value))
[1] -0.2156507
> acos((-sqrt(17)-1)/8)
[1] 2.2658
> (sin(.Last.value)+sin(2*.Last.value))
[1] 1.76745
> (sin(mvr)+sin(2*mvr))
[1] -0.2156507
>
Note that my variable mvr is the value of acos((-sqrt(17)-1)/8).   However, if I
use .Last.value after invoking the acos function in a further calculation, I get
a nonsensical result.   What is going on?

I am running R version 4.0.0 (2020-04-24) on an iMac, operating system MacOS
11.4 ?Big Sur?.

Mik Bickis

Behaviour is confirmed on 4.1.0 on Ubuntu 20.04 and Windows 10.

What's happening is

(sin(.Last.value) + sin(2 * .Last.value))

is evaluating the first sin(.Last.value) which resolves to 0.7680514
then using that in the second expression, sin(2*.Last.value) := sin(2
* 0.7680514) which resolves to 0.9993982. Add them together to get the
1.76745 value you're showing.

Interestingly, if you're using R Studio, if you go to Tools > Global
Options > General > Advanced > Other and check "Show .Last.Value
in
environment listing," then

acos((-sqrt(17) - 1) / 8)
# [1] 2.2658
(sin(.Last.value)+sin(2*.Last.value))
# [1] -0.2156507

Sadly, .Last.value isn't very well documented. The only thing I see in
the manuals is in R-ints,
https://cran.r-project.org/doc/manuals/r-release/R-ints.html#Base-environment
> Similarly, the evaluator creates a symbol .Last.value which appears as a
variable in the base environment.
Playing around though, it seems if the last value is the result of a
symbol lookup or assignment, .Last.value seems to do something like
"re-evaluate" the symbol, resetting the .Last.value value to the
stored value each time, whereas if the last expression evaluated
returns a pure value, calling .Last.value in an expression can update
the .Last.value value.

    1
    #> [1] 1
    # This is evaluated as 1 + (1) + (2) + (4) + (8) as .Last.Value
changes each time.
    .Last.value + .Last.value + .Last.value + .Last.value + .Last.value
    #> [1] 16

    x <- 1
    # This is evaluated as 1 + 1 + 1 + 1 + 1
    .Last.value + .Last.value + .Last.value + .Last.value + .Last.value
    #> [1] 5

    x <- 1
    x
    # This is still evaluated as 1 + 1 + 1 + 1 + 1
    .Last.value + .Last.value + .Last.value + .Last.value + .Last.value
    #> [1] 5

    1
    # This is also evaluated as 1 + 1 + 1 + 1 + 1
    (x <- .Last.value) + .Last.value + .Last.value + .Last.value +
.Last.value

Created on 2021-06-19 by the [reprex
package](https://reprex.tidyverse.org) (v2.0.0.9000)

The lesson here is .Last.value should (almost) never be used
programmatically, and in an interactive context the only approved use
of it is to store a result before losing it.

On Fri, Jun 18, 2021 at 11:03 PM Bickis, Mikelis <bickis at math.usask.ca>
wrote:
>
> Hello:
>
> Here is a bit of R-code:
>
> > mvr
> [1] 2.2658
> > (sin(.Last.value)+sin(2*.Last.value))
> [1] -0.2156507
> > acos((-sqrt(17)-1)/8)
> [1] 2.2658
> > (sin(.Last.value)+sin(2*.Last.value))
> [1] 1.76745
> > (sin(mvr)+sin(2*mvr))
> [1] -0.2156507
> >
>
> Note that my variable mvr is the value of acos((-sqrt(17)-1)/8).   However,
if I use .Last.value after invoking the acos function in a further calculation,
I get a nonsensical result.   What is going on?
>
> I am running R version 4.0.0 (2020-04-24) on an iMac, operating system
MacOS 11.4 ?Big Sur?.
>
> Mik Bickis
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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 for the report. Fixed in R-devel in r80530; will port to
R-patched soon.

Best,

luke

On Fri, 18 Jun 2021, Bickis, Mikelis wrote:
> Hello:
>
> Here is a bit of R-code:
>
>> mvr
> [1] 2.2658
>> (sin(.Last.value)+sin(2*.Last.value))
> [1] -0.2156507
>> acos((-sqrt(17)-1)/8)
> [1] 2.2658
>> (sin(.Last.value)+sin(2*.Last.value))
> [1] 1.76745
>> (sin(mvr)+sin(2*mvr))
> [1] -0.2156507
>>
>
> Note that my variable mvr is the value of acos((-sqrt(17)-1)/8).   However,
if I use .Last.value after invoking the acos function in a further calculation,
I get a nonsensical result.   What is going on?
>
> I am running R version 4.0.0 (2020-04-24) on an iMac, operating system
MacOS 11.4 ?Big Sur?.
>
> Mik Bickis
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.
-- 
Luke Tierney
Ralph E. Wareham Professor of Mathematical Sciences
University of Iowa                  Phone:             319-335-3386
Department of Statistics and        Fax:               319-335-3017
    Actuarial Science
241 Schaeffer Hall                  email:   luke-tierney at uiowa.edu
Iowa City, IA 52242                 WWW:  http://www.stat.uiowa.edu