10.3. LAPLACE TRANSFORMS 235

Now this converges to 0 as k → ∞. In fact, for 0 < a < 1,

limk→∞

kk+1

k!(e−aa

)k= 0

To see this, take the ratio of the k+1 term to the kth term.

(k+1)k+2

(k+1)! (e−aa)k+1

kk+1

k! (e−aa)k= e−aa

1k+1

(k+1)k+2

kk+1 = e−aa(

1+1k

)k(1+

1k

)which converges to e1−aa, a positive number less than 1. Verify this. You can see it istrue by graphing xe1−x on [0,1] for example. Therefore, denoting as Ak the expressionkk+1

k! (e−aa)k, and letting e1−aa < r < 1, it follows that for all k large enough,

Ak+1

Ak< r < 1

and so, iterating this,

Ak+m

Ak=

Ak+m

Ak+m−1

Ak+m−1

Ak+m−2

Ak+m−2

Ak+m−3· · · Ak+1

Ak≤ rm−1

Since |r|< 1, limm→∞ Ak+m ≤ limm→∞ Akrm−1 = 0. Here a = 1−δ .Next consider the last integral. This obviously converges to 0 because of the exponential

growth of φ . In fact,∣∣∣∣∫ ∞

1+δ

kk+1

k!(e−uu

)k(φ (u)−φ (1))du

∣∣∣∣≤ ∫ ∞

1+δ

kk+1

k!(e−uu

)k(

a+beλu)

du

Now changing the variable letting uk = t, and doing everything on finite intervals followedby passing to a limit, the absolute value of the above is dominated by∫

k(1+δ )

kk+1

k!e−t( t

k

)k 1k

(a+beλ (t/k)

)dt

=∫

k(1+δ )

1k!

e−ttk(

a+beλ (t/k))

dt for some a,b ≥ 0

=∫

0

1k!

e−ttk(

a+beλ (t/k))

dt −∫ k(1+δ )

0

1k!

e−ttk(

a+beλ (t/k))

dt

However, the limit as k → ∞ of the integral on the right equals the improper integral onthe left. Thus this converges to 0 as k → ∞. Thus all that is left to consider is the middleintegral in which δ was chosen such that |φ (u)−φ (1)| < ε. Then from what was shownearlier, ∣∣∣∣∫ 1+δ

1−δ

kk+1

k!(e−uu

)k(φ (u)−φ (1))du

∣∣∣∣≤ ε

∫∞

0

kk+1

k!(e−uu

)k du = ε

It follows that if φ is continuous at 1,

limk→∞

∫∞

0

kk+1

k!(e−uu

)k(φ (u)−φ (1))du = 0

10.3. LAPLACE TRANSFORMS 235Now this converges to 0 as k > ©. In fact, forO <a < 1,k+l~ ® (,-ajim (ea)=0To see this, take the ratio of the k+ 1 term to the k” term.(k--1)+? ( ~ag)et1A k+2 kGarr a 1 (k+1) a 1 1= = 1+-—) (14-i (e~4a)é e ELT kk+1 e@ + k + kwhich converges to e!~“a, a positive number less than 1. Verify this. You can see it istrue by graphing xe!~* on [0,1] for example. Therefore, denoting as A, the expressionxe (e4a)*, and letting e!~“a < r <1, it follows that for all k large enough,Ay— <r<lA rkand so, iterating this,Aitm — Aktm Aktm=1 Ak+m-2 | Akt © m-1Ax Ak+m—1 Ak+m—2 Ak+m—3 AqSince |r| <1, lim so Agim < limps. Agr” | =0. Herea = 1-6.Next consider the last integral. This obviously converges to 0 because of the exponentialgrowth of @. In fact,<[- eta (a+ be") du~ Ji+6 ik!co Ek+l[ey ow -oaau+5 k!Now changing the variable letting uk = t, and doing everything on finite intervals followedby passing to a limit, the absolute value of the above is dominated byDosw ee (a) lervent™ aee 1= | —e tt (at beMl®)) at for some a,b > 0k(1+6) k!— [Pl re A(t/k) pent 1k A(t/k)=| ne tK (a-+be )at F ue (a-+be )atHowever, the limit as k — © of the integral on the right equals the improper integral onthe left. Thus this converges to 0 as k — co. Thus all that is left to consider is the middleintegral in which 6 was chosen such that |@ (w) — @ (1)| < €. Then from what was shownearlier,co pk+1<e | k (e“u) du =e0146 xk+1 “ak|, ale to @W-o aul se [a_5 k!It follows that if @ is continuous at 1,; co k+l “akjim P ar fe u) ($ (u) —@ (1))du=0