gnuplot-py-users

[Gnuplot-py-users] Suppress visual plot, but still save hardcopy

[Conor R. <con...@gm...>]

I'm generating thousands of plots, which I'm storing as ps files as
records of different trials.  For example:

#plot my data
g2.plot(r22, r2, r21)

#save results of a trial
g2.hardcopy(trial+'/result.ps', enhanced=1, color=1)

What I would like to do is complete the plot in g2, but suppress the
visual output and just save a hardcopy to view later if need be.
Eventually, as expected, AquaTerm will crash as I keep displaying the
numerous plots.  Is there a way to implement this?

I'm running OSX (10.4.8), X11 and AquaTerm, but I'm assuming the
method or arguments, if they exist, are universal.

Thank you,
Conor

Re: [Gnuplot-py-users] Suppress visual plot, but still save hardcopy

[Michael H. <mh...@al...>]

Conor Robinson wrote:
> I'm generating thousands of plots, which I'm storing as ps files as
> records of different trials.  For example:
> 
> #plot my data
> g2.plot(r22, r2, r21)
> 
> #save results of a trial
> g2.hardcopy(trial+'/result.ps', enhanced=1, color=1)
> 
> What I would like to do is complete the plot in g2, but suppress the
> visual output and just save a hardcopy to view later if need be.
> Eventually, as expected, AquaTerm will crash as I keep displaying the
> numerous plots.  Is there a way to implement this?

To prevent the terminal output, I think you could use the 'unknown'
terminal type (which appears to send output to nirvana):

import gp, Gnuplot

# Set the term value that is restored after a hardcopy:
gp.GnuplotOpts.default_term = 'unknown'

g = Gnuplot.Gnuplot()

# Set the initial term value:
g('set term unknown')

...

If you are very worried about efficiency, though, you can't really use
the hardcopy() method at all because it causes the data to be processed
twice (once for 'unknown' then once for your desired hardcopy output
device).  If that is a problem, you probably want to set the terminal
and output file manually before plotting anything.

Michael

Re: [Gnuplot-py-users] Suppress visual plot, but still save hardcopy

[Conor R. <con...@gm...>]

Using iCy-fLaME's suggestion I threw this together.

Example:

#directory I want to save to as str
fdir = geo+'/'+str(hidden)+'/'+trial+'/%netval.eps'

#set output add quotes
g1('set output '+'"'+fdir+'"') #note there was an extra 'set' (typo)
in previous example

#set terminal
g1('set terminal postscript landscape enhanced color')

#plot
g1.plot(r12, r1, Gnuplot.Func('-x + 1', title = "ave response%"),
Gnuplot.Func('x', title = "ave cut%"))

This seems to work well.  Is this efficient, or what you refer to by
'set manually before plotting' ?  I'd like to tighten up the time it
takes for report generation albeit a negligible % of the process time.

Thanks
Conor

On 11/9/06, Michael Haggerty <mh...@al...> wrote:
> Conor Robinson wrote:
> > I'm generating thousands of plots, which I'm storing as ps files as
> > records of different trials.  For example:
> >
> > #plot my data
> > g2.plot(r22, r2, r21)
> >
> > #save results of a trial
> > g2.hardcopy(trial+'/result.ps', enhanced=1, color=1)
> >
> > What I would like to do is complete the plot in g2, but suppress the
> > visual output and just save a hardcopy to view later if need be.
> > Eventually, as expected, AquaTerm will crash as I keep displaying the
> > numerous plots.  Is there a way to implement this?
>
> To prevent the terminal output, I think you could use the 'unknown'
> terminal type (which appears to send output to nirvana):
>
> import gp, Gnuplot
>
> # Set the term value that is restored after a hardcopy:
> gp.GnuplotOpts.default_term = 'unknown'
>
> g = Gnuplot.Gnuplot()
>
> # Set the initial term value:
> g('set term unknown')
>
> ...
>
> If you are very worried about efficiency, though, you can't really use
> the hardcopy() method at all because it causes the data to be processed
> twice (once for 'unknown' then once for your desired hardcopy output
> device).  If that is a problem, you probably want to set the terminal
> and output file manually before plotting anything.
>
> Michael
>