MathGroup Archive 2012

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

Search the Archive

NDSolve slow when using generic function definition

  • To: mathgroup at smc.vnet.net
  • Subject: [mg124998] NDSolve slow when using generic function definition
  • From: Panayiotis Georgiou <ps.georgiou at gmail.com>
  • Date: Fri, 17 Feb 2012 06:23:33 -0500 (EST)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com

Hello,

I am using NDSolve to solve a simple 1st order ODE. The problem I have
is that when I solve the function with predefined parameters it runs
extremely fast. However, if I try to do the same but by defining a
generic function it is extremely slow. I have tried placing the
evaluate at various places but still no luck.

To be more specific, when solving this for example:

sol = NDSolve[{z'[t] == \[Alpha]*fmod2[z[t], 2, 0.5, k, 1, d]*
      I1[t], z[0] == 0.1}, z, {t, 0, 1/f0}, MaxStepSize -> 0.0001,
   AccuracyGoal -> 6, PrecisionGoal -> 5];

pp4 = Plot[{Evaluate[z[t] /. sol]}, {t, 0, 1/f0}, PlotRange -> All,
  AxesOrigin -> {0, 0},
  GridLines -> {{{0.5, Dashed}, {max[n1], Dashed}}, None}]

It runs almost instantly. But then if I do the following:

Zt[t_, n_, m_, j_, d_] :=
  z[t] /. NDSolve[{z'[t3] == \[Alpha]*fmod2[z[t3], n, m, K[n], j, d]*
       I1[t3], z[0] == z0}, z, {t3, 0, 1/f0}, MaxStepSize -> 0.0001,
    AccuracyGoal -> 6, PrecisionGoal -> 5];

pp4 = Plot[{Evaluate[Zt[t, 2, 0.5, 1, d]]}, {t, 0, 1/f0},
  PlotRange -> All, AxesOrigin -> {0, 0},
  GridLines -> {{{0.5, Dashed}, {max[n1], Dashed}}, None}]

It takes about 5 minutes to evaluate and plot. I have tried enclosing
in Evaluate[] various parts of the function definition but it is still
extremely slow.

Do you have any idea why there is this speed difference between the
two approaches?

Thanks for your help.







  • Prev by Date: Reading Single and Double reals in Little-endian and Big-endian
  • Next by Date: Re: simple question on DSolve
  • Previous by thread: Re: Reading Single and Double reals in Little-endian and Big-endian
  • Next by thread: Re: more questions about NIntegrate