MathGroup Archive 2012

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

Search the Archive

Re: compile a numerical integral

  • To: mathgroup at smc.vnet.net
  • Subject: [mg124581] Re: compile a numerical integral
  • From: Bill Rowe <readnews at sbcglobal.net>
  • Date: Wed, 25 Jan 2012 07:07:21 -0500 (EST)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com

On 1/24/12 at 5:07 AM, ruth.lazkoz at ehu.es (Ruth Lazkoz S=C3=A1ez) wrote:

>I am working to make a code faster using compile and failed at the
>very beginning. Is there a way to make a version that works?

>f=Compile[{u},NIntegrate[x*u,{x,0.,#}]&/@{1,2,3}]

Try,

f = Compile[{{u, _Real}, {a, _Real}}, NIntegrate[u x, {x, 0, a}]]

But, while this compiles, it would be very surprising if there
were any measurable improvement in execution speed. Compile
cannot optimize built-in functions. Calls to built-in functions
already call compiled code. What Compile can do is Compile an
Mathematica expression. That is:

u*x

could be compiled by Compile. But, it is very unlikely you will
see any significant improvement for such a simple expression.
You would almost certainly get better performance by doing

In[2]:= int = Integrate[u x, {x, 0, a}]

Out[2]= (a^2*u)/2

and then supplying numeric values for u,a.

or by doing:

g = Compile[{{u, _Real}, {a, _Real}},
   Evaluate[Integrate[u x, {x, 0, a}]]]

That is telling Mathematica to evaluate the integral
symbolically then compile the result.




  • Prev by Date: Using same pure function for curve and surface mappings
  • Next by Date: Mapping Bezier curve to surface. Locator problem, still.
  • Previous by thread: Re: compile a numerical integral
  • Next by thread: Re: compile a numerical integral