[seqfan] Re: Chewing an old bone
David Applegate
david at research.att.com
Tue Oct 25 02:21:35 CEST 2011
> I really hope that the qlog plot will remain *supplementary* to the
> log plot (for large signed sequences), and will not replace it. Each
> of these plots can be the more useful one depending on applications.
Do you have an example of when plotting log(|a(n)|+1) or
log(a(n)+1) is more informative than plotting qlog(a(n))?
David Applegate
> From seqfan-bounces at list.seqfan.eu Mon Oct 24 20:08:24 2011
> Date: Mon, 24 Oct 2011 17:08:13 -0700
> From: =?ISO-8859-1?Q?Beno=EEt_Jubin?= <benoit.jubin at gmail.com>
> To: Sequence Fanatics Discussion list <seqfan at list.seqfan.eu>
> Subject: [seqfan] Re: Chewing an old bone
> > Currently, the qlog plot is supplementary to the log plot, if I understand Russ
> > Cox correctly, this situation is provisional, and if the qlog plot works out, it
> > will replace the log plot.
> I really hope that the qlog plot will remain *supplementary* to the
> log plot (for large signed sequences), and will not replace it. Each
> of these plots can be the more useful one depending on applications.
> On Mon, Oct 24, 2011 at 6:48 AM, David Wilson <davidwwilson at comcast.net> wrote:
> > From looking at the old code, the scatter plot scale was as follows:
> >
> > For small graphs, a(n)
> > For large nonnegative graphs, log(a(n)+1)
> > For large negative graphs, log(|a(n)|+1)
> >
> > Except for graph title, this is equivalent to
> >
> > For small graphs, a(n)
> > For large graphs, log(|a(n)|+1)
> >
> > The criterion for deciding what is a large graph is complicated, based
> > on a sample of five elements. Personally, I would have chosen a log plot
> > based on some simpler criterion, say if |a(n)| or max(a)-min(a) exceeds
> > some constant N.
> >
> > Currently, the qlog plot is supplementary to the log plot, if I understand
> > Russ
> > Cox correctly, this situation is provisional, and if the qlog plot works
> > out, it
> > will replace the log plot. By this I mean that a(n) scatterplot will appear
> > when
> > it now does, the qlog scatterplot will appear when the log plot now does.
> >
> >
> > I don't think it is much of a decision: look at, say, A000055, apart from
> > titles,
> > the log and qlog plots are visually identical for nonnegative sequences. If
> > you
> > then look at, say, A055615, you see the advantage of the qlog plot over the
> > log plot re showing negative values.
> >
> >
> > On 10/23/2011 11:28 PM, Benoît Jubin wrote:
> >>
> >> Is it normal that there is a scatter plot of qlog for every signed
> >> sequence, even when there is no scatter plot of log (because the
> >> sequence assumes only small values in the displayed interval; see
> >> https://oeis.org/A002321/graph)?
> >>
> >> Also, this is nitpicking, but there is some discrepancy in the titles
> >> of all the graphs: "Pin plot of Axxxxx" and "Scatterplot of Axxxxx(n)"
> >> (one or two words, and with or without argument).
> >>
> >> Thanks for the addition of the qlog graphs, and I agree with previous
> >> posts that it should stay "in addition to" log, and not "instead of".
> >>
> >> Benoit
> >>
> >> On Sun, Oct 23, 2011 at 5:00 PM, David Wilson<davidwwilson at comcast.net>
> >> wrote:
> >>>
> >>> On 10/23/2011 5:23 PM, Russ Cox wrote:
> >>>>
> >>>> Sounds good. I have installed the graph:
> >>>>
> >>>> https://oeis.org/A103718/graph
> >>>>
> >>>> Russ
> >>>>
> >>> Thank you all for putting up with me regarding this.
> >>>
> >>>
> >>> _______________________________________________
> >>>
> >>> Seqfan Mailing list - http://list.seqfan.eu/
> >>>
> >> _______________________________________________
> >>
> >> Seqfan Mailing list - http://list.seqfan.eu/
> >>
> >
> >
> > _______________________________________________
> >
> > Seqfan Mailing list - http://list.seqfan.eu/
> >
> _______________________________________________
> Seqfan Mailing list - http://list.seqfan.eu/
More information about the SeqFan
mailing list