Mathematica 9 is now available
Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2008

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

Search the Archive

Re: Problems on definit integratiion of gaussian profiles

  • To: mathgroup at smc.vnet.net
  • Subject: [mg86830] Re: [mg86785] Problems on definit integratiion of gaussian profiles
  • From: Bob Hanlon <hanlonr at cox.net>
  • Date: Sat, 22 Mar 2008 00:54:54 -0500 (EST)
  • Reply-to: hanlonr at cox.net

Your first expression

expr1 = Power[E, 
  Plus[Times[-1, Power[a, 2], Power[c, -2]], 
   Times[-1, Power[d, -2], Power[Plus[b, Times[-1, x]], 2]]]]

E^(-(a^2/c^2)-(b-x)^2/d^2)

Your second expression

expr21 = Power[E, 
  Plus[Times[-1, Power[a, 2], Power[c, -2]], 
   Times[-1, Power[b, 2],  Power[d, -2]], Times[2, b, Power[d, -1], x], 
   Times[-1, Power[d, -2], Power[x,  2]]]]

E^(-(a^2/c^2) - x^2/d^2 + (2*b*x)/d - 
      b^2/d^2)

What your second expression should be

expr22 = Power[E, 
  Plus[Times[-1, Power[a, 2], Power[c, -2]], 
   Times[-1, Power[b, 2],  Power[d, -2]], Times[2, b, Power[d, -2], x], 
   Times[-1, Power[d, -2], Power[x,  2]]]]

E^(-(a^2/c^2) - x^2/d^2 + (2*b*x)/d^2 - 
      b^2/d^2)

expr1 == expr22 // Simplify

True

int1 = Integrate[expr1, {x, -Infinity, Infinity}, Assumptions -> {d > 0}]

(d*Sqrt[Pi])/E^(a^2/c^2)

int2 = Integrate[expr22, {x, -Infinity, Infinity}, Assumptions -> {d > 0}]

(d*Sqrt[Pi])/E^(a^2/c^2)

As expected, the results are identical.


Bob Hanlon

---- Regaly Zsolt <regaly at konkoly.hu> wrote: 
> Dear All Mathgroup Fellows!
> 
> I dont understand the behavoir of Mathematica integrating  
> exponential functions. I try to integrate the same functions in  
> different forms. The integrands are
> 
> Power[E, Plus[Times[-1, Power[a, 2], Power[c, -2]], Times[-1, Power 
> [d, -2], \
> Power[Plus[b, Times[-1, x]], 2]]]]
> 
> or
> 
> Power[E, Plus[Times[-1, Power[a, 2], Power[c, -2]], Times[-1, Power 
> [b, 2], \
> Power[d, -2]], Times[2, b, Power[d, -1], x], Times[-1, Power[d, -2],  
> Power[x, \
> 2]]]]
> 
> which are the same, but the results will differ. Indeed calculating  
> definit integral in a -Infinity to Infinity domain, assuming that all  
> variables are larger than 0, the result will be completly different!
> 
> 1'st case the result will be:
> 
> Times[d, Power[E, Times[-1, Power[a, 2], Power[c, -2]]], Power[Pi, \
> Rational[1, 2]]]
> 
> while in the 2'nd case
> 
> Times[d, Power[E, Plus[Power[b, 2], Times[-1, Power[a, 2], Power[c,  
> -2]], \
> Times[-1, Power[b, 2], Power[d, -2]]]], Power[Pi, Rational[1, 2]]]
> 
> As You can see the 1'st result is completly independent of b! I dont  
> understand. How can it be fixed?
> 
> Cheers,
> Zsolt Regaly
> 



  • Prev by Date: Re: Saving Packages
  • Next by Date: Re: Mathematica hangs...
  • Previous by thread: Re: Problems on definit integratiion of gaussian profiles
  • Next by thread: Variable Order Markov Models in Mathematica