Berend Hasselman
2017-Mar-31 17:15 UTC
[R] Difference between R for the Mac and for Windows
I have noted a difference between R on macOS en on Kubuntu Trusty (64bits) with complex division. I don't know what would happen R on Windows. R.3.3.3: macOS (10.11.6) -----------------> (1+2i)/0[1] NaN+NaNi> (-1+2i)/0[1] NaN+NaNi> > 1i/0[1] NaN+NaNi> 1i/(0+0i)[1] NaN+NaNi KubuntuTrusty -----------------> (1+2i)/0[1] Inf+Infi> (-1+2i)/0[1] -Inf+Infi> > 1i/0[1] NaN+Infi> 1i/(0+0i)[1] NaN+Infi Interesting to see what R on Windows delivers. Berend Hasselman
On 31.03.2017 19:15, Berend Hasselman wrote:> > I have noted a difference between R on macOS en on Kubuntu Trusty (64bits) with complex division. > I don't know what would happen R on Windows. > > R.3.3.3: > > macOS (10.11.6) > ----------------- >> (1+2i)/0 > [1] NaN+NaNi >> (-1+2i)/0 > [1] NaN+NaNi >> >> 1i/0 > [1] NaN+NaNi >> 1i/(0+0i) > [1] NaN+NaNi > > > KubuntuTrusty > ----------------- >> (1+2i)/0 > [1] Inf+Infi >> (-1+2i)/0 > [1] -Inf+Infi >> >> 1i/0 > [1] NaN+Infi >> 1i/(0+0i) > [1] NaN+Infi > > Interesting to see what R on Windows delivers.Same as KubuntuTrusty and what I would expect. Best, Uwe Ligges> > Berend Hasselman > > ______________________________________________ > 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. >
On Fri, Mar 31, 2017 at 12:15 PM, Berend Hasselman <bhh at xs4all.nl> wrote:> > I have noted a difference between R on macOS en on Kubuntu Trusty (64bits) > with complex division. > I don't know what would happen R on Windows. > > R.3.3.3: > > macOS (10.11.6) > ----------------- > > (1+2i)/0 > [1] NaN+NaNi > > (-1+2i)/0 > [1] NaN+NaNi > > > > 1i/0 > [1] NaN+NaNi > > 1i/(0+0i) > [1] NaN+NaNi > > > KubuntuTrusty > ----------------- > > (1+2i)/0 > [1] Inf+Infi > > (-1+2i)/0 > [1] -Inf+Infi > > > > 1i/0 > [1] NaN+Infi > > 1i/(0+0i) > [1] NaN+Infi > > Interesting to see what R on Windows delivers. >?> (1+2i)/0 [1] Inf+Infi> (-1+2i)/0[1] -Inf+Infi> 1i/0[1] NaN+Infi> 1i/(0+0i)[1] NaN+Infi> Sys.info()sysname release "Windows" "7 x64" version nodename "build 7601, Service Pack 1" "IT-JMCKOWN" machine login "x86-64" "John.Mckown" user effective_user "John.Mckown" "John.Mckown">Same as Kubuntu. I am _guessing_ that the MacOS somehow sets up the floating point processing to work differently, since they are all on Intel machines nowadays. Or the R was customized to detect division by zero in software and not really do any floating point processing at all. ?> > Berend Hasselman > >-- "Irrigation of the land with seawater desalinated by fusion power is ancient. It's called 'rain'." -- Michael McClary, in alt.fusion Maranatha! <>< John McKown [[alternative HTML version deleted]]
Berend Hasselman
2017-Mar-31 17:37 UTC
[R] Difference between R for the Mac and for Windows
> On 31 Mar 2017, at 19:28, John McKown <john.archie.mckown at gmail.com> wrote: > > On Fri, Mar 31, 2017 at 12:15 PM, Berend Hasselman <bhh at xs4all.nl> wrote: > > I have noted a difference between R on macOS en on Kubuntu Trusty (64bits) with complex division. > I don't know what would happen R on Windows. > > R.3.3.3: > > macOS (10.11.6) > ----------------- > > (1+2i)/0 > [1] NaN+NaNi > > (-1+2i)/0 > [1] NaN+NaNi > > > > 1i/0 > [1] NaN+NaNi > > 1i/(0+0i) > [1] NaN+NaNi > > > KubuntuTrusty > ----------------- > > (1+2i)/0 > [1] Inf+Infi > > (-1+2i)/0 > [1] -Inf+Infi > > > > 1i/0 > [1] NaN+Infi > > 1i/(0+0i) > [1] NaN+Infi > > Interesting to see what R on Windows delivers. > > ?> (1+2i)/0 > [1] Inf+Infi > > (-1+2i)/0 > [1] -Inf+Infi > > 1i/0 > [1] NaN+Infi > > 1i/(0+0i) > [1] NaN+Infi > > Sys.info() > sysname release > "Windows" "7 x64" > version nodename > "build 7601, Service Pack 1" "IT-JMCKOWN" > machine login > "x86-64" "John.Mckown" > user effective_user > "John.Mckown" "John.Mckown" > > > > Same as Kubuntu. I am _guessing_ that the MacOS somehow sets up the floating point processing to work differently, since they are all on Intel machines nowadays. Or the R was customized to detect division by zero in software and not really do any floating point processing at all. > ? >I think it's the system math library that does this. I have assumed that the Kubuntu Trusty (and Windows) give the correct result. In my package geigen I have taken that into account and made a specialized complexdivision function that tries to detect a possibly wrong outcome (which appears to happen only on macOS). Berend Hasselman> Berend Hasselman > > > > -- > "Irrigation of the land with seawater desalinated by fusion power is ancient. It's called 'rain'." -- Michael McClary, in alt.fusion > > Maranatha! <>< > John McKown
[This is drifting somewhat awy from the original intention of the topic, I think]. This looks like a build dependency. I get 3.3.2 (yeah, I know, should upgrade):> (1+2i)/0[1] NaN+NaNi R-devel, march 24:> (1+2i)/0[1] Inf+Infi on the *same* machine. The difference is that one is stock CRAN, the other was built locally. So with the toolchain being updated for 3.4.0, this difference would likely go away. Or at least change... -pd> On 31 Mar 2017, at 19:15 , Berend Hasselman <bhh at xs4all.nl> wrote: > > > I have noted a difference between R on macOS en on Kubuntu Trusty (64bits) with complex division. > I don't know what would happen R on Windows. > > R.3.3.3: > > macOS (10.11.6) > ----------------- >> (1+2i)/0 > [1] NaN+NaNi >> (-1+2i)/0 > [1] NaN+NaNi >> >> 1i/0 > [1] NaN+NaNi >> 1i/(0+0i) > [1] NaN+NaNi > > > KubuntuTrusty > ----------------- >> (1+2i)/0 > [1] Inf+Infi >> (-1+2i)/0 > [1] -Inf+Infi >> >> 1i/0 > [1] NaN+Infi >> 1i/(0+0i) > [1] NaN+Infi > > Interesting to see what R on Windows delivers. > > Berend Hasselman > > ______________________________________________ > 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.-- Peter Dalgaard, Professor, Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Office: A 4.23 Email: pd.mes at cbs.dk Priv: PDalgd at gmail.com