Reduction to Riemann Integrals

3.5. Reduction to Riemann Integrals#

The next theorem tells us that we may replace the symbol dα with α(x)dx under some conditions.

Theorem 3.6

Suppose fR(α) on [a,b], and α has a continuous derivative on [a,b]. Then fαR on [a,b], and

abf(x)dα(x)=abf(x)α(x)dx

Proof. First, suppose f is bounded by M>0, i.e.,

(3.9)#|f(x)|Mx[a,b]

Let ε>0 be arbitrary.

Because α is continuous on [a,b], it is continuous uniformly there. There exists δ>0 such that

(3.10)#|st||α(s)α(t)|<ε2M(ba)

Since f is integrable w.r.t. α on [a,b], there exists a partition Pε of [a,b] such that for any refinement P of Pε, and any list of representatives T of P, we have

(3.11)#|S(P,T,f,α)abfdα|<ε/2

Then, we can find a finer partition PεPε such that Pε<δ.

Let PPε be a refinement such that and T be a list of representatives of P. Note that P is of course also a refinement of Pε. Applying the mean value theorem, we have

S(P,T,f,α)=k=1nf(tk)[α(tk)α(tk1)]=k=1nf(tk)α(sk)Δxk

where each sk(xk1,xk).

Taking the difference of S(P,T,fα,x) and S(P,T,f,α), we have

|S(P,T,fα,x)S(P,T,f,α)|=|k=1nf(tk)[α(tk)α(sk)]Δxk|k=1n|f(tk)[α(tk)α(sk)]Δxk|=k=1n|f(tk)||α(tk)α(sk)|Δxk

Then applying (3.9) and (3.10), the above difference is further bounded by

(3.12)#|S(P,T,fα,x)S(P,T,f,α)|<Mε2M(ba)k=1nΔxk=ε/2

Recall PPε. Then we may conclude this proof by comparing (3.11) and (3.12).