Specific Functions
Staircase Function
Definition:
If g ( x ) g(x) g(x) is defined at [ a , b ] [a,b] [a,b], which is the union of finite intervals having no intersection with each other and in each intersection it’s a constant function, then it’s a staircase function.
Properties:
1, Staircase function is definitely Riemann integrable.
Gamma Function
Definition:
Γ
(
p
+
1
)
=
∫
0
+
∞
t
p
e
−
t
d
t
.
\Gamma(p+1)=\int_0^{+\infty}t^pe^{-t}\text dt.
Γ(p+1)=∫0+∞tpe−tdt.
Properties:
Γ ( 1 ) = 1 , Γ ( s + 1 ) = s Γ ( s ) , ln Γ is convex . \Gamma(1)=1,\Gamma(s+1)=s\Gamma(s),\ln\Gamma\text{ is convex}. Γ(1)=1,Γ(s+1)=sΓ(s),lnΓ is convex.
These 3 conditions uniquely determines this function.
And we know that Γ ( n ) = ( n − 1 ) ! \Gamma(n)=(n-1)! Γ(n)=(n−1)!
Γ ( 1 2 ) = π . \Gamma(\frac 12)=\sqrt{\pi}. Γ(21)=π.
Beta Function
B ( a , b ) = Γ ( a ) Γ ( b ) Γ ( a + b ) . B(a,b)=\frac{\Gamma(a)\Gamma(b)}{\Gamma(a+b)}. B(a,b)=Γ(a+b)Γ(a)Γ(b).
Check Function
Definition:
δ
i
,
j
=
{
0
,
i
≠
j
.
1
,
i
=
j
.
\delta_{i,j}=\begin{cases} 0,i\neq j.\\ 1,i=j. \end{cases}
δi,j={0,i=j.1,i=j.
Class of Functions
Functions of Bounded Variation
Definition:
∀ partition p at [ a , b ] , V p ( f ) : = ∑ ∣ f ( x i ) − f ( x i − 1 ) ∣ . \forall\text{ partition }p\text{ at }[a,b],V_p(f):=\sum|f(x_i)-f(x_{i-1})|. ∀ partition p at [a,b],Vp(f):=∑∣f(xi)−f(xi−1)∣.
f f f is of bounded variation at [ a , b ] [a,b] [a,b], if sup V p ( f ) ∃ \sup V_p(f)\exist supVp(f)∃, set down as total variation V ( f ) V(f) V(f).
Properties:
Monotonic functions are of bounded variation.
f f f is of bounded variation, if f ′ f' f′ is bounded.
f f f is bounded, if it is of bounded variation.
V ( f ) = 0. ⇔ f ≡ C V(f)=0.\Leftrightarrow f\equiv C V(f)=0.⇔f≡C.
V ( f ± g ) ⩽ V ( f ) + V ( g ) . V(f\pm g)\leqslant V(f)+V(g). V(f±g)⩽V(f)+V(g).
V ( f g ) ⩽ sup ∣ f ∣ V ( g ) + sup ∣ g ∣ V ( f ) . V(fg)\leqslant\sup|f|V(g)+\sup|g| V(f). V(fg)⩽sup∣f∣V(g)+sup∣g∣V(f).
V ( 1 / f ) ⩽ V ( f ) / m 2 , ∣ f ∣ ⩾ m > 0. V(1/f)\leqslant V(f)/m^2,|f|\geqslant m>0. V(1/f)⩽V(f)/m2,∣f∣⩾m>0.
V x : = V f at [ a , x ] V_x:=V_f\text{ at }[a,x] Vx:=Vf at [a,x].
Obviously, V x V_x Vx is increasing.
And consider a refinement p ∗ p^* p∗ of partition p , V p ( f ) ⩽ V p ∗ ( f ) p,V_p(f)\leqslant V_{p^*}(f) p,Vp(f)⩽Vp∗(f).
Consider interval U 0 − ( x 0 , δ ) U_0^-(x_0,\delta) U0−(x0,δ).
∀ partition p at ( x 0 − δ , x 0 ) , V f ( p ) ⩾ ∣ f ( x 0 ) − f ( x 0 − δ ) ∣ \forall\text{ partition }p\text{ at }(x_0-\delta,x_0),V_f(p)\geqslant|f(x_0)-f(x_0-\delta)| ∀ partition p at (x0−δ,x0),Vf(p)⩾∣f(x0)−f(x0−δ)∣, because the right part is produced by a partition having only endpoints.
Thus V p ( f ) δ − ∣ f ( x 0 ) − f ( x 0 − δ ) ∣ δ \frac{V_p(f)}{\delta}-\frac{|f(x_0)-f(x_0-\delta)|}{\delta} δVp(f)−δ∣f(x0)−f(x0−δ)∣ .
Consider δ → 0 \delta\rightarrow0 δ→0, you know D ( x ) = V x − f ( x ) D(x)=V_x-f(x) D(x)=Vx−f(x) is increasing.
Then any function of bounded variation can be decomposed as the difference of two increasing functions.
Thus it’s definitely Riemann integrable.