Review : Systems of Equations — Question 5

PDF ↗

Question 5

Work over the real numbers. Use substitution or elimination, keeping track of conditions under which an operation preserves all solutions. Check candidates in the original equations.

Two nearly parallel measurement equations are x+y=2,x+(1+ε)y=2+δ,x+y=2,\qquad x+(1+\varepsilon)y=2+\delta, where ε>0\varepsilon>0 is known and δ\delta is an unknown measurement error satisfying |δ|≤η|\delta|\le\eta, with η>0\eta>0.

Tasks

  1. Find the exact solution for every admissible δ\delta and explain what would happen if ε=0\varepsilon=0 were allowed.

  2. Find the smallest uniform bound on the largest absolute coordinate error relative to the nominal solution at δ=0\delta=0. Prove sharpness.

  3. For ε=η=10−3\varepsilon=\eta=10^{-3}, find the full range of possible solutions. Compare the small equation residual of the nominal answer with its possible coordinate error.

  4. Find a necessary and sufficient condition on ε\varepsilon ensuring both coordinate errors are at most 0.10.1 for every |δ|≤η|\delta|\le\eta. Interpret the condition geometrically.

Original worksheet page 1: question and worked solution for 5-1-005
Show solutionHide solution

Question 5 – Solution

Strategy. Subtracting almost identical equations exposes the small coefficient that amplifies uncertainty.

Step 1: Solve and retain the excluded case. Subtract the first equation from the second to get εy=δ\varepsilon y=\delta. For ε>0\varepsilon>0, (x,y)=(2−δε,δε).\boxed{(x,y)=\left(2-\frac\delta\varepsilon, \frac\delta\varepsilon\right).} If ε=0\varepsilon=0 and δ≠0\delta\ne 0, there is no solution; if both are zero, the two equations coincide and every point on x+y=2x+y=2 is a solution.

Step 2: Derive the exact uncertainty bound. The nominal solution is (2,0)(2,0). Each coordinate changes in magnitude by |δ|/ε|\delta|/\varepsilon. Therefore max⁡(|x−2|,|y|)≤η/ε.\boxed{\max(|x-2|,|y|)\le\eta/\varepsilon.} Equality occurs when δ=η\delta=\eta or δ=−η\delta=-\eta, so no smaller universal bound is possible.

Step 3: Quantify the nearly parallel example. For ε=η=10−3\varepsilon=\eta=10^{-3}, let u=δ/ε∈[−1,1]u=\delta/\varepsilon\in[-1,1]. The full possible solution set is the segment (x,y)=(2−u,u),−1≤u≤1\boxed{(x,y)=(2-u,u),\ -1\le u\le 1}, from (3,−1)(3,-1) to (1,1)(1,1). The nominal answer satisfies the first equation exactly and has residual −δ-\delta in the second, of magnitude at most 0.0010.001. Nevertheless its coordinate error can be 11. Small residuals do not always certify small solution errors. The graph uses true scales; the two measured lines are nearly indistinguishable, while the possible solution segment is visible.

Step 4: Give the exact design condition. The requested guarantee holds exactly when η/ε≤0.1\eta/\varepsilon\le 0.1, or ε≥10η\boxed{\varepsilon\ge 10\eta}. Necessity follows by taking an endpoint error; sufficiency follows from the bound. Increasing ε\varepsilon separates the line directions and reduces the effect of shifting one measurement’s right-hand side.

See the diagram in the original worksheet below.

Original worksheet page 2: question and worked solution for 5-1-005

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