Development of Integration by Substitution

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
3 replies · 4K views
chaotixmonjuish
Messages
284
Reaction score
0
How was Integration by Substitution and Trig Substitution developed? My calc book doesn't have much info, just a short (not really complete) proof. Could someone explain and/or lead me in the right direction?
 
Physics news on Phys.org
This is reverse of the chain rule.
(For simplicity sake assume that [tex]f(x)[/tex] and [tex]g(x)[/tex] are differentiable everywhere).

Given [tex]f(x)[/tex] let [tex]F(x)[/tex] be any anti-derivative (which we know exists by the fundamental theorem of calculus). Now consider [tex]F(g(x))[/tex]. What is the derivative of that? Using the chain rule we fine [tex]F'(g(x))g'(x)[/tex] but [tex]F'(x) = f(x)[/tex] because it is an anti-derivative. So [tex][F(g(x))]' = f(g(x))g'(x)[/tex]. This means [tex]F(g(x))[/tex] is an anti-derivative of [tex]f(g(x))g'(x)[/tex]. So [tex]\int f(g(x))g'(x) dx = F(g(x))+C[/tex].

Here is an example. Say we want to find [tex]\int (x+1)^5[/tex]. We can let [tex]f(x) = x^5[/tex] and [tex]g(x)=x+1[/tex]. Now confirm that [tex](x+1)^5 = f(g(x))g'(x)[/tex]. So the anti-derivative is [tex]F(g(x))+C[/tex] but [tex]F(x) = (1/6)x^6[/tex]. So [tex]F(g(x))+C = (1/6)(x+1)^6 + C[/tex]. That is is.

Of course, I did the most basic example. But that is how I do my integrations. I never use the calculus book approach because it does not provide any insight like you said.
 
Yes, as Kummer said, it is Integral calculus's counterpart of The Chain Rule from Differential Calculus. Courant's treatment of Calculus, Volume 1, has a nice proof wish I will post because I'm nice =] Just note from this point on, the following is not my own work.
----------------------------------------------------

We suppose that a new variable u is introduced into a function F(x) by means of the equation [itex]x=\phi (u)[/itex] so that F(x) becomes a function of u:
[tex]F(x) = F( \phi (u) ) = G(u)[/tex].

By the chain rule of differential calculus:
[tex]\frac{dG}{du} = \frac{dF}{dx} \phi ' (u)[/tex].

If we now write
[tex]F'(x) = f(x) \mbox{and} G'(u) = g(u)[/tex], or the equivalent expressions [tex]F(x) = \int f(x) dx \mbox{and} G(u) = \int g(u) du[/tex]

then on one hand the chain rule takes the form [tex]g(u) = f(x) \phi ' (u)[/tex]
and on the other hand [itex]G(u) = F(x)[/itex] by definition, that is,
[tex]\int g(u) du = \int f(x) dx[/tex], and we obtain the integral formula equivalent to the chain rule: [tex]\int f( \phi (u)) \phi '(u) du = \int f(x) dx , {x=\phi (u)}[/tex].
----------------------------------------------------------

Trig substitution is merely the case where the substitution happens to be a trigonometric function.
 
Here is how trigonometric substitution works formally using the substitution rule.

Consider [tex]\int \sqrt{1-x^2} \ dx[/tex] (on [tex](-1,1)[/tex].)

We can write, [tex]\int \frac{1-x^2}{\sqrt{1-x^2}} \ dx[/tex].
Let [tex]g(x) = \sin^{-1} x[/tex] then [tex]g'(x) = \left( \sqrt{1-x^2} \right)^{-1}[/tex]. And [tex]f(x) = \cos^2 x[/tex] then [tex]f(g(x)) = 1-x^2[/tex]. This means,
[tex]\int f(g(x))g'(x) dx = F(g(x))+C[/tex].
Where [tex]F(x)[/tex] is an anti-derivative of [tex]\cos^2 x[/tex]. Which is [tex]F(x) = \frac{1}{2}x + \frac{1}{2}\sin x\cos x[/tex] now evaluate this at [tex]g(x) = \sin^{-1} x[/tex] which gives [tex]\frac{1}{2}\sin^{-1} x + \frac{1}{2} x \sqrt{1-x^2}[/tex].

I might have a mistake because I was not careful. But you get the idea.