MathGroup Archive 2008

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

Search the Archive

Re: Request for Collective Wisdom...

  • To: mathgroup at smc.vnet.net
  • Subject: [mg88703] Re: Request for Collective Wisdom...
  • From: David Bailey <dave at Remove_Thisdbailey.co.uk>
  • Date: Tue, 13 May 2008 07:09:46 -0400 (EDT)
  • References: <fvpcoc$met$1@smc.vnet.net>

Here is a second tip.

If a calculation is going very slowly, don't jump to the conclusion that 
you need to use a more functional programming style - at least without a 
little investigation. A spectacular way to consume machine time is to 
create an algorithm - say an iterative approximation to a square root - 
that is meant for real data, and then give it an argument such as Sqrt[3]/2.

If the algorithm does not use explicit real numbers, and does not use N 
for example, it will build up a huge expression involving roots and 
fractions. This is particularly subtle if you simply give the output to 
Plot - which will apply N - your calculation will run extremely slowly 
and yet appear to work (if it ever terminates).

David Bailey
http://www.dbaileyconsultancy.co.uk


  • Prev by Date: Re: FullForm OK but StandardForm fails
  • Next by Date: Re: Request for Collective Wisdom...
  • Previous by thread: Re: Request for Collective Wisdom...
  • Next by thread: Re: Request for Collective Wisdom...