[no subject]

• From: Steven M. Christensen <stevec at yoda.physics.unc.edu>
• Date: Wed, 7 Apr 93 10:24:50 +0200
• Apparently-to: mathgroup-send at yoda.physics.unc.edu

```Hi mathgroup, dear Wolfram Researcher

A trivial Integral like Int(z^-(3/2) dz) takes almost a minute if the
function is within a product of some constant parameters, which could
be pulled out very easily automatically (for some reasons I want such
trivial integrals to be integrated automatically). Look at the
example below where I ask Mma to perform just the above Integral, which takes
51 Seconds, then I introduce a function FastInt that pulls out the
constants first and then starts integrating - this takes only 0.07 Seconds!!
Crazy - isn't it?

I think Mma should do such simple transformation on its own, or at least
supply a function like FastInt (perhaps a bit smarter than this) performing
these transformations.

I also found that the same behaviour slows down Solve substantially,  when
there are too many Symbols (constants) involved.

Mathematica 2.1 for SPARC
-- OPEN LOOK graphics initialized --
-- Laserprinter -Psp(arc) installed. Use PSPrint[ -Graphics- ] --

In[1]:= ff:=(-(1.2154068053283*10^-14)*2^(7/8)*c^(7/4)*cm^(3/4)*Mdot^(7/4)*qm^(7/8)* q0^(7/8)*Second*xi^(7/4)*Csc[phi]^(3/2))/ (D^2*g^(3/4)*Lambda*nu^(1/2)*(1 + xi)^(7/8)*z^(1/2)*Csc[alpha]^(3/2))

In[2]:= Timing[Integrate[ff,z]]

-14  7/8  7/4   3/4     7/4   7/8   7/8
Out[2]= {51.7167 Second, (-2.43081 10    2    c    cm    Mdot    qm    q0

7/4                 3/2
>       Second xi    Sqrt[z] Csc[phi]   ) /

2  3/4                         7/8           3/2
>     (D  g    Lambda Sqrt[nu] (1 + xi)    Csc[alpha]   )}

In[3]:= FastInt[f_,z_Symbol]:=((Int[f,z] //. Int[\$a_ \$f_,\$z_]:>\$a Int[\$f,\$z] /; D[\$a,\$z]==0) /. Int->Integrate)

In[4]:= Timing[FastInt[ff,z]]

-14  7/8  7/4   3/4     7/4   7/8
Out[4]= {0.0666667 Second, (-2.43081 10    2    c    cm    Mdot    qm

7/8          7/4                 3/2
>       q0    Second xi    Sqrt[z] Csc[phi]   ) /

2  3/4                         7/8           3/2
>     (D  g    Lambda Sqrt[nu] (1 + xi)    Csc[alpha]   )}

Heino

```
(o o)
--------------oOO--(_)--OOo------------------------------------------------
___      ___   _____   ___   _____  |Heino Falcke
/__/\    /__/| /____/\ /__/| /____/\ |Max-Planck-Institut f. Radioastronomie
|  \ \  /|  || |  o \/ |  || |  o \/ |Auf dem Huegel 69
|  |\ \//|  || |  __/  |  ||f|  __/  |D-W5300 Bonn, Germany
|  ||\_/ |  || |  ||   |  || |  |\ \ |E-mail: HFALCKE at mpifr-bonn.mpg.de
|__|/    |__|/ |__|/   |__|/ |__|/\_||Fax: 49/(0)228/525-229
|Tel.: 49/(0)228/525-285
----------------------------------------------------------------------------

```

• Prev by Date: Some help/advice
• Next by Date: Re: sequential joining of 3D points