MathGroup Archive 2012

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Kernel exits

  • To: mathgroup at smc.vnet.net
  • Subject: [mg128091] Re: Kernel exits
  • From: Roland Franzius <roland.franzius at uos.de>
  • Date: Sat, 15 Sep 2012 03:36:59 -0400 (EDT)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com
  • Delivered-to: l-mathgroup@wolfram.com
  • Delivered-to: mathgroup-newout@smc.vnet.net
  • Delivered-to: mathgroup-newsend@smc.vnet.net
  • References: <k2s2re$dq5$1@smc.vnet.net>

Am 13.09.2012 09:43, schrieb davidgolber:
> Running Mathematica 4 on a new PC (lots of memory)under Windows 7 Professional and XP Virtual Machine.
>
> When doing long calculations, the kernal exits with no error messages of any kind, after about two hours.  The front end is there, but the kernel is gone.  (So if I type a command into the front end, it is as though the notebook has just been loaded and this is the first command given.)
>
> The calculation is a lot of FindMinimum, with some error messages ("zero gradient" and "did not converge") which are suppressed as usual after four appearances.  I wonder if the error process uses stack which is then not released correctly.
>
> Any suggestions?

Start Windows Task Manager (Ctrl+Alt+Del) and look at the memory. The 
kernel shuts down at 2 GB. Of course, looking at the graph for two hours 
is a bit tedious.

Alternatives:

Include a (....  ;  Print[n, ": ", MemoryInUse[]] ;  ...) somewhere in 
an outer loop.

Write the memory values togehter with valuue of characteristic variables 
into a text file using
(OpenWrite ... StreamWrite .. Close  ;)

or a Run[ "cmd", ... ]   statement. Information about the command shell 
is aquired by starting a cmd box in the start menu and typing "help cmd"

Break up your search problem into serial hunting pieces, write the end 
values of an episode into a file, Quit[] the kernel and resume the 
hunting with the results of the last day as starting values.

That's the way the brain is working discarding all the memory garbage of 
the last day....


-- 

Roland Franzius



  • Prev by Date: Re: Landau letter, Re: Mathematica as a New Approach...
  • Next by Date: Re: Substitute variable name for value (reverse of simplification)
  • Previous by thread: Re: Kernel exits
  • Next by thread: Re: Kernel exits