[Rd] Floating point issue

Taras Zakharko t@r@@@z@kh@rko @end|ng |rom uzh@ch
Tue Jul 19 16:33:21 CEST 2022


> So in this case, the different results are not due to the architecture (M1 aka ARM64) but to the system/implementation.

I would say that your results demonstrate that R number parsing code relies on higher-than double precision to perform correct number parsing. Extended precision is not guaranteed by the C standard, so I would classify this as a bug in R. Unless of course R doesn’t want to make any  number parsing guarantees, but that would be an odd choice to make :)

Cheers, 

Taras 

> On 19 Jul 2022, at 14:59, Olivier Benz <olivier.benz using b-data.ch> wrote:
> 
> I found that quite puzzling indeed. Thanks for the references.
> 
> So in this case, the different results are not due to the architecture (M1 aka ARM64) but to the system/implementation.
> 
>> On 19 Jul 2022, at 14:22, Taras Zakharko <taras.zakharko using uzh.ch <mailto:taras.zakharko using uzh.ch>> wrote:
>> 
>> Just a quick note of this in case someone finds Olivier's results puzzling. Standard ARM64 ABI (which Linux follows) defines long double as a 16-byte extended precision type. LLVM’s compiler-rt has a software implementation for doing operations on such types. Apple instead defines long double as identical to double in their ABIs. 
>> 
>> References
>> 
>> https://developer.arm.com/documentation/ka004751/1-0
>> 
>> https://developer.apple.com/documentation/xcode/writing-arm64-code-for-apple-platforms
>> 
>> 
>> 
>> 
>>> On 19 Jul 2022, at 13:04, Olivier Benz <olivier.benz using b-data.ch> wrote:
>>> 
>>> The difference between macOS 12.4 and Debian 11 (Docker, virtualization framework) running on a MacBook Pro (M1 Max).
>>> -> `.Machine$sizeof.longdouble` on macOS returns 8 whereas on Debian 11 it returns 16.
>>> 
>>> macOS 12.4 on MacBook Pro (M1 Max):
>>> ``` r
>>> .Machine
>>> #> $double.eps
>>> #> [1] 2.220446e-16
>>> #> 
>>> #> $double.neg.eps
>>> #> [1] 1.110223e-16
>>> #> 
>>> #> $double.xmin
>>> #> [1] 2.225074e-308
>>> #> 
>>> #> $double.xmax
>>> #> [1] 1.797693e+308
>>> #> 
>>> #> $double.base
>>> #> [1] 2
>>> #> 
>>> #> $double.digits
>>> #> [1] 53
>>> #> 
>>> #> $double.rounding
>>> #> [1] 5
>>> #> 
>>> #> $double.guard
>>> #> [1] 0
>>> #> 
>>> #> $double.ulp.digits
>>> #> [1] -52
>>> #> 
>>> #> $double.neg.ulp.digits
>>> #> [1] -53
>>> #> 
>>> #> $double.exponent
>>> #> [1] 11
>>> #> 
>>> #> $double.min.exp
>>> #> [1] -1022
>>> #> 
>>> #> $double.max.exp
>>> #> [1] 1024
>>> #> 
>>> #> $integer.max
>>> #> [1] 2147483647
>>> #> 
>>> #> $sizeof.long
>>> #> [1] 8
>>> #> 
>>> #> $sizeof.longlong
>>> #> [1] 8
>>> #> 
>>> #> $sizeof.longdouble
>>> #> [1] 8
>>> #> 
>>> #> $sizeof.pointer
>>> #> [1] 8
>>> 
>>> bitC <- function(x) noquote(vapply(as.double(x), function(x) { # split one
>>> double
>>> b <- substr(as.character(rev(numToBits(x))), 2L, 2L)
>>>    paste0(c(b[1L], " ", b[2:12], " | ", b[13:64]), collapse = "")
>>>  }, ""))
>>> bitC(10^25)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>> bitC(10000000000000000905969664)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010010
>>> bitC(10000000000000000905969664 - 10^25)
>>> #> [1] 0 10000011110 | 0000000000000000000000000000000000000000000000000000
>>> bitC(1e25)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>> ```
>>> 
>>> <sup>Created on 2022-07-19 by the [reprex package](https://reprex.tidyverse.org) (v2.0.1)</sup>
>>> 
>>> <details style="margin-bottom:10px;">
>>> <summary>
>>> Session info
>>> </summary>
>>> 
>>> ``` r
>>> sessioninfo::session_info()
>>> #> ─ Session info ───────────────────────────────────────────────────────────────
>>> #>  setting  value
>>> #>  version  R version 4.2.1 (2022-06-23)
>>> #>  os       macOS Monterey 12.4
>>> #>  system   aarch64, darwin20
>>> #>  ui       X11
>>> #>  language (EN)
>>> #>  collate  en_US.UTF-8
>>> #>  ctype    UTF-8
>>> #>  tz       Europe/Copenhagen
>>> #>  date     2022-07-19
>>> #>  pandoc   2.18 @ /opt/local/bin/ (via rmarkdown)
>>> #> 
>>> #> ─ Packages ───────────────────────────────────────────────────────────────────
>>> #>  package     * version date (UTC) lib source
>>> #>  cli           3.3.0   2022-04-25 [1] CRAN (R 4.2.0)
>>> #>  digest        0.6.29  2021-12-01 [1] CRAN (R 4.2.0)
>>> #>  ellipsis      0.3.2   2021-04-29 [1] CRAN (R 4.2.0)
>>> #>  evaluate      0.15    2022-02-18 [1] CRAN (R 4.2.0)
>>> #>  fansi         1.0.3   2022-03-24 [1] CRAN (R 4.2.0)
>>> #>  fastmap       1.1.0   2021-01-25 [1] CRAN (R 4.2.0)
>>> #>  fs            1.5.2   2021-12-08 [1] CRAN (R 4.2.0)
>>> #>  glue          1.6.2   2022-02-24 [1] CRAN (R 4.2.0)
>>> #>  highr         0.9     2021-04-16 [1] CRAN (R 4.2.0)
>>> #>  htmltools     0.5.3   2022-07-18 [1] CRAN (R 4.2.1)
>>> #>  knitr         1.39    2022-04-26 [1] CRAN (R 4.2.0)
>>> #>  lifecycle     1.0.1   2021-09-24 [1] CRAN (R 4.2.0)
>>> #>  magrittr      2.0.3   2022-03-30 [1] CRAN (R 4.2.0)
>>> #>  pillar        1.8.0   2022-07-18 [1] CRAN (R 4.2.1)
>>> #>  pkgconfig     2.0.3   2019-09-22 [1] CRAN (R 4.2.0)
>>> #>  purrr         0.3.4   2020-04-17 [1] CRAN (R 4.2.0)
>>> #>  R.cache       0.15.0  2021-04-30 [1] CRAN (R 4.2.0)
>>> #>  R.methodsS3   1.8.2   2022-06-13 [1] CRAN (R 4.2.0)
>>> #>  R.oo          1.25.0  2022-06-12 [1] CRAN (R 4.2.0)
>>> #>  R.utils       2.12.0  2022-06-28 [1] CRAN (R 4.2.0)
>>> #>  reprex        2.0.1   2021-08-05 [1] CRAN (R 4.2.0)
>>> #>  rlang         1.0.4   2022-07-12 [1] CRAN (R 4.2.1)
>>> #>  rmarkdown     2.14    2022-04-25 [1] CRAN (R 4.2.0)
>>> #>  sessioninfo   1.2.2   2021-12-06 [1] CRAN (R 4.2.0)
>>> #>  stringi       1.7.8   2022-07-11 [1] CRAN (R 4.2.1)
>>> #>  stringr       1.4.0   2019-02-10 [1] CRAN (R 4.2.0)
>>> #>  styler        1.7.0   2022-03-13 [1] CRAN (R 4.2.0)
>>> #>  tibble        3.1.7   2022-05-03 [1] CRAN (R 4.2.0)
>>> #>  utf8          1.2.2   2021-07-24 [1] CRAN (R 4.2.0)
>>> #>  vctrs         0.4.1   2022-04-13 [1] CRAN (R 4.2.0)
>>> #>  withr         2.5.0   2022-03-03 [1] CRAN (R 4.2.0)
>>> #>  xfun          0.31    2022-05-10 [1] CRAN (R 4.2.0)
>>> #>  yaml          2.3.5   2022-02-21 [1] CRAN (R 4.2.0)
>>> #> 
>>> #>  [1] /Users/benz0li/Library/R/arm64/4.2/library
>>> #>  [2] /Library/Frameworks/R.framework/Versions/4.2-arm64/Resources/library
>>> #> 
>>> #> ──────────────────────────────────────────────────────────────────────────────
>>> ```
>>> 
>>> </details>
>>> 
>>> Debian 11 (Docker, virtualization framework) on MacBook Pro (M1 Max)
>>> ``` r
>>> .Machine
>>> #> $double.eps
>>> #> [1] 2.220446e-16
>>> #> 
>>> #> $double.neg.eps
>>> #> [1] 1.110223e-16
>>> #> 
>>> #> $double.xmin
>>> #> [1] 2.225074e-308
>>> #> 
>>> #> $double.xmax
>>> #> [1] 1.797693e+308
>>> #> 
>>> #> $double.base
>>> #> [1] 2
>>> #> 
>>> #> $double.digits
>>> #> [1] 53
>>> #> 
>>> #> $double.rounding
>>> #> [1] 5
>>> #> 
>>> #> $double.guard
>>> #> [1] 0
>>> #> 
>>> #> $double.ulp.digits
>>> #> [1] -52
>>> #> 
>>> #> $double.neg.ulp.digits
>>> #> [1] -53
>>> #> 
>>> #> $double.exponent
>>> #> [1] 11
>>> #> 
>>> #> $double.min.exp
>>> #> [1] -1022
>>> #> 
>>> #> $double.max.exp
>>> #> [1] 1024
>>> #> 
>>> #> $integer.max
>>> #> [1] 2147483647
>>> #> 
>>> #> $sizeof.long
>>> #> [1] 8
>>> #> 
>>> #> $sizeof.longlong
>>> #> [1] 8
>>> #> 
>>> #> $sizeof.longdouble
>>> #> [1] 16
>>> #> 
>>> #> $sizeof.pointer
>>> #> [1] 8
>>> #> 
>>> #> $longdouble.eps
>>> #> [1] 1.92593e-34
>>> #> 
>>> #> $longdouble.neg.eps
>>> #> [1] 9.62965e-35
>>> #> 
>>> #> $longdouble.digits
>>> #> [1] 113
>>> #> 
>>> #> $longdouble.rounding
>>> #> [1] 5
>>> #> 
>>> #> $longdouble.guard
>>> #> [1] 0
>>> #> 
>>> #> $longdouble.ulp.digits
>>> #> [1] -112
>>> #> 
>>> #> $longdouble.neg.ulp.digits
>>> #> [1] -113
>>> #> 
>>> #> $longdouble.exponent
>>> #> [1] 15
>>> #> 
>>> #> $longdouble.min.exp
>>> #> [1] -16382
>>> #> 
>>> #> $longdouble.max.exp
>>> #> [1] 16384
>>> 
>>> bitC <- function(x) noquote(vapply(as.double(x), function(x) { # split one
>>> double
>>> b <- substr(as.character(rev(numToBits(x))), 2L, 2L)
>>>    paste0(c(b[1L], " ", b[2:12], " | ", b[13:64]), collapse = "")
>>>  }, ""))
>>> bitC(10^25)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>> bitC(10000000000000000905969664)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>> bitC(10000000000000000905969664 - 10^25)
>>> #> [1] 0 00000000000 | 0000000000000000000000000000000000000000000000000000
>>> bitC(1e25)
>>> #> [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>> ```
>>> 
>>> <sup>Created on 2022-07-19 by the [reprex package](https://reprex.tidyverse.org) (v2.0.1)</sup>
>>> 
>>> <details style="margin-bottom:10px;">
>>> <summary>
>>> Session info
>>> </summary>
>>> 
>>> ``` r
>>> sessioninfo::session_info()
>>> #> ─ Session info ───────────────────────────────────────────────────────────────
>>> #>  setting  value
>>> #>  version  R version 4.2.1 (2022-06-23)
>>> #>  os       Debian GNU/Linux 11 (bullseye)
>>> #>  system   aarch64, linux-gnu
>>> #>  ui       X11
>>> #>  language (EN)
>>> #>  collate  en_US.UTF-8
>>> #>  ctype    en_US.UTF-8
>>> #>  tz       Etc/UTC
>>> #>  date     2022-07-19
>>> #>  pandoc   2.18 @ /usr/bin/ (via rmarkdown)
>>> #> 
>>> #> ─ Packages ───────────────────────────────────────────────────────────────────
>>> #>  package     * version date (UTC) lib source
>>> #>  cli           3.3.0   2022-04-25 [2] CRAN (R 4.2.1)
>>> #>  crayon        1.5.1   2022-03-26 [2] CRAN (R 4.2.1)
>>> #>  digest        0.6.29  2021-12-01 [2] CRAN (R 4.2.1)
>>> #>  ellipsis      0.3.2   2021-04-29 [2] CRAN (R 4.2.1)
>>> #>  evaluate      0.15    2022-02-18 [2] CRAN (R 4.2.1)
>>> #>  fansi         1.0.3   2022-03-24 [2] CRAN (R 4.2.1)
>>> #>  fastmap       1.1.0   2021-01-25 [2] CRAN (R 4.2.1)
>>> #>  fs            1.5.2   2021-12-08 [2] CRAN (R 4.2.1)
>>> #>  glue          1.6.2   2022-02-24 [2] CRAN (R 4.2.1)
>>> #>  highr         0.9     2021-04-16 [2] CRAN (R 4.2.1)
>>> #>  htmltools     0.5.2   2021-08-25 [2] CRAN (R 4.2.1)
>>> #>  knitr         1.39    2022-04-26 [2] CRAN (R 4.2.1)
>>> #>  lifecycle     1.0.1   2021-09-24 [2] CRAN (R 4.2.1)
>>> #>  magrittr      2.0.3   2022-03-30 [2] CRAN (R 4.2.1)
>>> #>  pillar        1.7.0   2022-02-01 [2] CRAN (R 4.2.1)
>>> #>  pkgconfig     2.0.3   2019-09-22 [2] CRAN (R 4.2.1)
>>> #>  purrr         0.3.4   2020-04-17 [2] CRAN (R 4.2.1)
>>> #>  R.cache       0.15.0  2021-04-30 [2] CRAN (R 4.2.1)
>>> #>  R.methodsS3   1.8.2   2022-06-13 [2] CRAN (R 4.2.1)
>>> #>  R.oo          1.25.0  2022-06-12 [2] CRAN (R 4.2.1)
>>> #>  R.utils       2.12.0  2022-06-28 [2] CRAN (R 4.2.1)
>>> #>  reprex        2.0.1   2021-08-05 [2] CRAN (R 4.2.1)
>>> #>  rlang         1.0.4   2022-07-12 [2] CRAN (R 4.2.1)
>>> #>  rmarkdown     2.14    2022-04-25 [2] CRAN (R 4.2.1)
>>> #>  sessioninfo   1.2.2   2021-12-06 [2] CRAN (R 4.2.1)
>>> #>  stringi       1.7.8   2022-07-11 [2] CRAN (R 4.2.1)
>>> #>  stringr       1.4.0   2019-02-10 [2] CRAN (R 4.2.1)
>>> #>  styler        1.7.0   2022-03-13 [2] CRAN (R 4.2.1)
>>> #>  tibble        3.1.7   2022-05-03 [2] CRAN (R 4.2.1)
>>> #>  utf8          1.2.2   2021-07-24 [2] CRAN (R 4.2.1)
>>> #>  vctrs         0.4.1   2022-04-13 [2] CRAN (R 4.2.1)
>>> #>  withr         2.5.0   2022-03-03 [2] CRAN (R 4.2.1)
>>> #>  xfun          0.31    2022-05-10 [2] CRAN (R 4.2.1)
>>> #>  yaml          2.3.5   2022-02-21 [2] CRAN (R 4.2.1)
>>> #> 
>>> #>  [1] /home/benz0li/R/aarch64-unknown-linux-gnu-library/4.2
>>> #>  [2] /usr/local/lib/R/site-library
>>> #>  [3] /usr/local/lib/R/library
>>> #> 
>>> #> ──────────────────────────────────────────────────────────────────────────────
>>> ```
>>> 
>>> </details>
>>> 
>>>> On 10 Jul 2022, at 22:23, Bill Dunlap <williamwdunlap using gmail.com> wrote:
>>>> 
>>>> The following function, 'bitC' from ?numToBits, displays the bits in a
>>>> double precision number, separated into the sign bit, the 11 exponent bits,
>>>> and the 52 bits in the mantissa.  I've shown the results with your numbers
>>>> from R-2.4.0 on my Windows 11 Lenovo laptop: what do you get?
>>>> 
>>>>> bitC <- function(x) noquote(vapply(as.double(x), function(x) { # split
>>>> one double
>>>> +     b <- substr(as.character(rev(numToBits(x))), 2L, 2L)
>>>> +     paste0(c(b[1L], " ", b[2:12], " | ", b[13:64]), collapse = "")
>>>> +   }, ""))
>>>>> bitC(10^25)
>>>> # [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>>>> bitC(10000000000000000905969664)
>>>> # [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>>>> bitC(10000000000000000905969664 - 10^25)
>>>> # [1] 0 00000000000 | 0000000000000000000000000000000000000000000000000000
>>>>> bitC(1e25)
>>>> # [1] 0 10001010010 | 0000100010110010101000101100001010000000001010010001
>>>> 
>>>> -Bill
>>>> 
>>>> On Sun, Jul 10, 2022 at 7:00 AM Antoine Fabri <antoine.fabri using gmail.com>
>>>> wrote:
>>>> 
>>>>> Dear r-devel,
>>>>> 
>>>>> For some numbers, the printed value is not equivalent to the input :
>>>>> 
>>>>> options(scipen = 999)
>>>>> ## GOOD
>>>>> 1e24
>>>>> #> [1]  999999999999999983222784
>>>>> 1e24 == 999999999999999983222784
>>>>> #> [1] TRUE
>>>>> 
>>>>> ## BAD
>>>>> 1e25
>>>>> #> [1] 10000000000000000905969664
>>>>> 1e25 == 10000000000000000905969664
>>>>> #> [1] FALSE
>>>>> 
>>>>> ## STILL BAD
>>>>> 10000000000000000905969664
>>>>> #> [1] 10000000000000003053453312
>>>>> 
>>>>> ## GOOD AGAIN
>>>>> 10000000000000003053453312
>>>>> #> [1] 10000000000000003053453312
>>>>> 
>>>>> # Additionally
>>>>> 10000000000000000000000000 == 1e25
>>>>> #> [1] FALSE
>>>>> 
>>>>> Are these bugs ?
>>>>> 
>>>>>     [[alternative HTML version deleted]]
>>>>> 
>>>>> ______________________________________________
>>>>> R-devel using r-project.org mailing list
>>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>>> 
>>>> 
>>>> 	[[alternative HTML version deleted]]
>>>> 
>>>> ______________________________________________
>>>> R-devel using r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>> 
>>> --
>>> benz0li.b-data.io | @benz0li | olivier.benz using b-data.ch
>>> 
>>> ______________________________________________
>>> R-devel using r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>> 
>> 
>> 	[[alternative HTML version deleted]]
>> 
>> ______________________________________________
>> R-devel using r-project.org <mailto:R-devel using r-project.org> mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
> --
> benz0li.b-data.io <http://benz0li.b-data.io/> | @benz0li | olivier.benz using b-data.ch <mailto:olivier.benz using b-data.ch>

	[[alternative HTML version deleted]]



More information about the R-devel mailing list