Repeated Roots — Question 10

PDF ↗

Question 10

A dimensionless family of repeated-root responses satisfies y″+2λy′+λ2y=0,y(0)=1,y′(0)=0,λ>0.y''+2\lambda y'+\lambda^2y=0,\qquad y(0)=1,\qquad y'(0)=0,\qquad \lambda>0. The design requirement is y(t)≤0.05y(t)\le 0.05 for every t≥2t\ge 2, with the additional constraint |y″(0)|≤4|y''(0)|\le 4.

Tasks

  1. Solve the IVP and prove that the response is positive and strictly decreasing for t>0t>0.

  2. Show that the first time at level 0.050.05 is z*/λz_*/\lambda, where z*>0z_*>0 is the unique root of (1+z)e−z=0.05(1+z)e^{-z}=0.05. Locate z*z_* between two consecutive four-decimal-place numbers by a justified numerical calculation.

  3. Translate both the settling requirement and the initial-acceleration constraint into restrictions on λ\lambda. Determine whether they can be satisfied simultaneously.

  4. Find the earliest achievable settling time under the acceleration constraint alone. Sketch the responses for λ=1,2\lambda=1,2 with the tolerance and required time marked, and explain why simply increasing λ\lambda is not an admissible unlimited improvement.

Original worksheet page 1: question and worked solution for 3-4-010
Show solutionHide solution

Question 10 – Solution

Strategy. Scale time by the repeated decay rate, then compare the performance requirement with the independent constraint on the initial acceleration.

Step 1: Solve and establish monotonicity. The repeated root is −λ-\lambda. The data give y=(1+λt)e−λt,y′=−λ2te−λt.\boxed{y=(1+\lambda t)e^{-\lambda t}},\qquad y'=-\lambda^2t e^{-\lambda t}. The response is positive, strictly decreasing for t>0t>0, starts at 1 and tends to zero. Therefore crossing the tolerance once is enough to guarantee it thereafter.

Step 2: Reduce the crossing calculation to one dimensionless number. Let f(z)=(1+z)e−zf(z)=(1+z)e^{-z} for z≥0z\ge 0. Then f(0)=1f(0)=1, f(z)→0f(z)\to 0 and f′(z)=−ze−z<0f'(z)=-ze^{-z}<0 for z>0z>0. The intermediate value theorem and strict decrease give exactly one positive root f(z*)=0.05f(z_*)=0.05.

Direct evaluation gives f(4.7438)>0.05f(4.7438)>0.05 and f(4.7439)<0.05f(4.7439)<0.05. Thus 4.7438<z*<4.7439,tsettle=z*/λ.\boxed{4.7438<z_*<4.7439},\qquad t_{\mathrm{settle}}=z_*/\lambda. Numerically z*≈4.743865z_*\approx 4.743865; the bracket, rather than a rounded equality, supports the feasibility comparisons below.

Step 3: Test simultaneous feasibility. Settling by time 2 requires z*/λ≤2z_*/\lambda\le 2, or λ≥z*/2>2.3719\lambda\ge z_*/2>2.3719. But the equation at zero gives y″(0)=−λ2y''(0)=-\lambda^2, so the acceleration constraint requires λ≤2\lambda\le 2. These ranges do not intersect: the two requirements are incompatible\boxed{\text{the two requirements are incompatible}}.

At the largest admissible value λ=2\lambda=2, the response at the deadline is y(2)=5e−4≈0.091578>0.05y(2)=5e^{-4}\approx 0.091578>0.05, confirming the failure directly.

See the diagram in the original worksheet below.

Step 4: Optimize within the allowed family. The settling time z*/λz_*/\lambda strictly decreases as λ\lambda increases. Under 0<λ≤20<\lambda\le 2, it is minimized at λ=2\lambda=2, giving tmin=z*/2≈2.371932.\boxed{t_{\min}=z_*/2\approx 2.371932}. This is later than the required time 2. Larger λ\lambda would improve the settling time but violate the initial-acceleration bound. The conclusion optimizes this explicitly specified family and constraint, not every possible differential-equation model or control law.

Original worksheet page 2: question and worked solution for 3-4-010

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