Review : Taylor Series — Question 5

PDF ↗

Question 5

Consider the nonelementary integral I=∫01/2e−t2dt,Sm=∑k=0m(−1)k22k+1k!(2k+1).I=\int_0^{1/2}e^{-t^2}\,dt,\qquad S_m=\sum_{k=0}^m\frac{(-1)^k}{2^{2k+1}k!(2k+1)}. An accuracy certificate is required; a decimal from a calculator is insufficient.

Tasks

  1. Derive the series for II and justify integrating the Maclaurin series term by term on the whole integration interval.

  2. Find the smallest mm for which the first-omitted-term bound certifies |I−Sm|≤10−6|I-S_m|\le 10^{-6}, and state the number of terms used.

  3. Give an exact rational enclosure for II using that partial sum and the next one. Report an approximation with its certified absolute error.

  4. Explain the sign of the error geometrically by comparing e−t2e^{-t^2} with the degree-2m2m polynomial under the integral on [0,1/2][0,1/2].

Original worksheet page 1: question and worked solution for 6-2-005
Show solutionHide solution

Question 5 – Solution

Strategy. Integrate a uniformly convergent series, then use the integrated alternating tail as an error certificate.

Step 1: Justify the interchange. On [0,1/2][0,1/2], |(−t2)k/k!|≤(1/4)k/k!|(-t^2)^k/k!|\le(1/4)^k/k!, whose sum is finite. The Weierstrass test gives uniform absolute convergence. Integrating yields I=∑k=0∞(−1)k22k+1k!(2k+1).I=\sum_{k=0}^{\infty}\frac{(-1)^k}{2^{2k+1}k!(2k+1)}. The positive term magnitudes decrease strictly to zero.

Step 2: Choose a certified truncation. The first omitted magnitude is bm=1/[22m+3(m+1)!(2m+3)]b_m=1/[2^{2m+3}(m+1)!(2m+3)]. It decreases with mm. For m=3m=3, b3=1/110592>10−6b_3=1/110592>10^{-6}, whereas b4=12703360<3.70×10−7<10−6.b_4=\frac 1{2703360}<3.70\times 10^{-7}<10^{-6}. Thus m=4\boxed{m=4} is the smallest index certified by this bound, using five terms.

Step 3: Supply an exact enclosure. The even partial sum is an upper bound and the next odd sum a lower bound: S4−12703360=S5<I<S4=12−124+1320−15376+1110592.\boxed{S_4-\frac 1{2703360}=S_5<I<S_4 =\frac 12-\frac 1{24}+\frac 1{320}-\frac 1{5376}+\frac 1{110592}.} Numerically, S4≈0.4612813637S_4\approx 0.4612813637. Reporting I≈0.46128136I\approx 0.46128136 introduces less than 4×10−94\times 10^{-9} additional rounding error, so its total absolute error is less than 3.74×10−73.74\times 10^{-7}.

Step 4: Interpret the signed area. Set p8(t)=1−t2+t4/2−t6/6+t8/24p_8(t)=1-t^2+t^4/2-t^6/6+t^8/24. The alternating tail is strictly negative for t>0t>0 in this interval, so p8(t)>e−t2p_8(t)>e^{-t^2}. Hence S4−IS_4-I is the positive area between the curves. The scaled gap below makes the very small difference visible; it vanishes at t=0t=0.

See the diagram in the original worksheet below.

Original worksheet page 2: question and worked solution for 6-2-005

Original worksheet layout. Use Enlarge or open the PDF for a closer view.