V6: ListPlot memory requeriments
- To: mathgroup at smc.vnet.net
- Subject: [mg78567] V6: ListPlot memory requeriments
- From: Nacho <ncc1701zzz at gmail.com>
- Date: Wed, 4 Jul 2007 05:39:59 -0400 (EDT)
Hello Group.
I would like to express some concerns about the memory usage in V6
when dealing with graphics.
I have noticed from the first time I used V6 that it takes a lot of
RAM to display the same graphics than in V5. Well, of course, not the
same, V6 is much nicer, but with the same commands.
For example, here is a extracted code from the Logistic Map notebook
from MathWorld:
LogisticMap = Compile[{{ mu, _Real}}, ({mu, #} &) /@
Union[Drop[NestList[
mu # (1 - #) &, .2, 300], 100]]];
f = Table[LogisticMap[mu], {mu, 0, 4, .002}]; // Timing
ListPlot[Flatten[f, 1],
PlotStyle -> {Red, AbsolutePointSize[.001]}, Frame -> True,
FrameLabel -> TraditionalForm /@ {r, x},
PlotLabel ->
TraditionalForm[
Subscript[x, n] ==
r Subscript[x, n - 1] (1 - Subscript[x, n - 1])]
] // Timing
This plots the logistic map.
After running it in V5.2, I get the draw in less than 0.5s, and memory
consumption is low, (of course, starting with a fresh kernel):
MemoryInUse[]
Out[4]=
23 541 104
MaxMemoryUsed[]
Out[5]=
24 953 432
In V6, instead, I get the plot in 5 seconds (10 times more) and with
the following memory consumption:
MemoryInUse[]
33 325 232
MaxMemoryUsed[]
318 339 384
Memory used is a little more than in V5, but maximum memory used is
more than 12 times bigger! to get the same plot.
I suppose that it means that if I want to plot 10 times more points,
it will be possible in V5.2 but not with V6 (at least in my computer
with "only" 1Gb of RAM).
I don't know if I'm doing something wrong to plot the result. Is there
is an alternative in V6 to ListPlot for lot of points? Maybe calling
Point directly, but it sounds like a backward step to me...
Other example... the "Neat Example" in the Help system for
ContourPlot3D, how much memory does it need? I have to abort the
process when more than 700Mb of RAM are taken and the computer starts
to swap memory.
I'm sorry for the complains, I'm just frustrated. The graphs look
great, really, but I have found that in more than one occasion, I have
to switch to V5.2 to show some results, because V5 plots are better
than no plot at all.
I can imagine the same problems happening in thousands of 2-3 years
old computers in the labs of most universities. I think that it
should really re-thought for V6.1.
Thanks for listening to us, Wolfram people.
Best regards.