       change of variables

• To: mathgroup at smc.vnet.net
• Subject: [mg81963] change of variables
• From: "Chris Chiasson" <chris at chiasson.name>
• Date: Sun, 7 Oct 2007 23:58:07 -0400 (EDT)

I don't understand the method for change of variables in the old MG
whatever standard method there is for a change of integration
variables in Mathematica? I only need the single variable case, but it
would be helpful to see the multidimensional extension.

Here is the integral I am working on, from Debye and Huckel's first
paper of the theory of electrolytes:

Subscript[\[CapitalPhi], e] == \[Integral]Subscript[U, e]/
T^2 \[DifferentialD]T

\Phi_e is the electrostatic contribution to the Gibbs free entropy
(yea, entropy) for an electrolyte solution.

The electric contribution to the internal energy is
eqn@Ue = Subscript[U, e] == -\!$$\*UnderoverscriptBox[\(\[Sum]$$, $$i = 1$$, $$s$$]$$FractionBox[\( SubscriptBox[\(N$$, $$i$$]
\*SubsuperscriptBox[$$z$$, $$i$$, $$2$$]\), $$2$$]
FractionBox[$$SuperscriptBox[\(q$$, $$2$$] \[Kappa]\), $$4 SubscriptBox[\(\[Pi]\[CurlyEpsilon]$$, $$r$$]
\*SubscriptBox[$$\[CurlyEpsilon]$$, $$0$$]\)]
\*FractionBox[$$1$$, $$1 + \[Kappa] \*SubscriptBox[\(a$$, $$i$$]\)]\)\)

where

eqn@\[Kappa] = \[Kappa]^2 == \!$$\*UnderoverscriptBox[\(\[Sum]$$, $$i = 1$$, $$s$$]
\*FractionBox[$$SubsuperscriptBox[\(z$$, $$i$$, $$2$$]
SuperscriptBox[$$q$$, $$2$$]
\*SubsuperscriptBox[$$n$$, $$i$$, $$o$$]\), $$SubscriptBox[\(\[CurlyEpsilon]$$, $$r$$]
SubscriptBox[$$\[CurlyEpsilon]$$, $$0$$]
SubscriptBox[$$k$$, $$B$$] T\)]\)

\kappa^{-1} is the Debye screening length, a measure of the size of
the ionic atmosphere around a particular ion in an electrolyte. As can
be seen by the definition, \kappa is a function of 1/T. However,
integrating with respect to T (which is constant for any given
solution, as is \kappa) isn't particularly informative. Instead, Debye
and Huckel integrate from \kappa = 0 to its value for the solution
under consideration. \kappa = 0 corresponds to an infinite screening
length and infinite dilution of the solution, thus \Phi_e is zero at

http://en.wikipedia.org/wiki/Debye-H%C3%BCckel_equation

So, taking the differential of the defining equation for \kappa gives

In:=
factorOut =
HoldPattern[Sum][expr_. factor_, its__] /;
FreeQ[factor, Alternatives[its][[All, 1]]] :>
factor Sum[expr, its];
deqn@d\[Kappa] =
D[eqn@\[Kappa], blah, NonConstants -> {\[Kappa], T}] /. factorOut /.
HoldPattern[D][xpr_, ___] :> Dt[xpr]

Out=
2 \[Kappa] Dt[\[Kappa]] == -((q^2 Dt[T] \!$$\*UnderoverscriptBox[\(\[Sum]$$, $$i = 1$$, $$s$$]$$\*SubsuperscriptBox[\(n$$, $$i$$, $$o$$]\
\*SubsuperscriptBox[$$z$$, $$i$$, $$2$$]\)\))/(
T^2 Subscript[k, B] Subscript[\[CurlyEpsilon], 0]
Subscript[\[CurlyEpsilon], r]))

(factorOut is a slightly modified version of a function in an old

By hand, I think the next step would be to solve the equation for the
differential Dt[T] and replace that in the integral. Things aren't so
simple in Mathematica. I am going to try to program my own function
for this, but I hope someone already has a good solution ready to go.