Right, I think I know what is causing the problem. The new peaks are
coming out with infinite volume (you can check in the seleced peaks
dialog). And for some obscure reason which I can't fathom (especially
since I wrote that code) the point peak search algorithm is using the
volume to scale the search region. I'm going to change that but the real
point is that the volume should not be coming out infinite. And I don't
know yet why it is in 1.0.3 on the Mac (I cannot reproduce this on Linux).
It's quite possibly a stupid missing initialisation (or similar) in the C
world.
In the meantime, we need to eliminate the infinite volumes. As of
1.0.3 there is a new option for volumes. Under Crosspeaks -> Peak Find
Parameters there is "Volume method" in the "Other parameters". The
"parabolic fit" is the current method. There is also a "box sum" and a
"trucated box sum". The former is just what you think it is and the
latter truncates the box by what it tries to figure out is the peak shape.
The box size is determined (for now) by the Boxwidth column in the
"Spectrum Widths" table in the same dialog.
Unfortunately, I've just noticed that that information is not being
propagated down to the C world. So this is still not a perfect
solution. But it seems to make things work (0 being a better volume
than infinite for the purpose at hand). So for now I would recommend (at
least on OSX) using "box sum" or "truncated box sum", after removing all
the peaks with "inf". (If you edit the ccp/Nmr.xml file that might be the
fastest way, as long as you haven't done anything with those peaks yet.)
I'll fix this mess in the next release.
Wayne
On Wed, 1 Jun 2005, Magnus Helgstrand wrote:
> No, that was not the problem. The active peaklist in the experiment
> is number 1, and the only peaklist available for that experiment is
> number 1.
>
> I tried another window with another peaklist and spectrum, I can then
> select peaks in the usual way. But if I pick any peaks in that window
> I get the same probelm as in the other window.
>
> Making a new window identical to the one not working does not help.
> It is still not possible to select peaks. Seems the problem is in the
> peaklist or the peaks. Problem is that whatever is wrong is stored to
> the next session.
>
> Magnus
>
> 31 maj 2005 kl. 17.11 skrev Wayne Boucher:
>
> > I'm wondering if the peak list is not active. We had that problem
> > this
> > morning here with a project. So if you go into Experiment -> Edit
> > Spectra
> > then there is a column for active peak list. It is only that peaklist
> > which you can select peaks in. And it's possible (I'm not sure why
> > off
> > the top of my head, Tim might know more) that for some reason that
> > shift-ctrl-leftbutton created a new peak list into which to insert the
> > peak.
> >
> > Wayne
> >
> > On Tue, 31 May 2005, Magnus Helgstrand wrote:
> >
> >
> >> Hi again,
> >>
> >> Now I am getting a bit confused. I restarted Analysis and I can still
> >> not select the CPs. Something saved in the project?
> >>
> >> Magnus
> >>
> >> 31 maj 2005 kl. 16.58 skrev Magnus Helgstrand:
> >>
> >>
> >>> Hi Analysis users,
> >>>
> >>> I used the shift-ctrl-leftbutton function to pick a new peak in a
> >>> 3D-spectrum. After that I can no longer use the left key to select
> >>> a peak in the spectrum, except the peak I just created. Other peaks
> >>> can be selected using shift-leftbutton.
> >>>
> >>> Macros such as "a" to assign or rightbutton commands such as peak-
> >>>
> >>>> center z-plane can not be used on other peaks than the newly
> >>>> created.
> >>>>
> >>>
> >>> I does not help to close the window and then opening it again...
> >>>
> >>> I universal problem or just in my compilation?
> >>>
> >>> I use Analysis 1.0.3 on a Mac.
> >>>
> >>> Magnus
> >>>
> >>>
> >>
> >
>
|