Question 10
Divide an insulated homogeneous rod into three equal cells of width . Let be cell temperatures and . Approximate each internal rightward flux by , set outer fluxes to zero, and use a forward Euler time step for the cell energy balances.
Tasks
Derive the three-cell update matrix and prove that the sum of temperatures is conserved.
Find the exact range of for which every update preserves the range of every input vector. Prove necessity as well as sufficiency.
Using the orthogonal vectors , and , determine exactly when cannot increase in one step.
Apply one step with to . Check heat conservation and squared norm, and sketch the old and new cell values. Explain why energy stability need not imply preservation of nonnegative temperatures.
Show solutionHide solution
Question 10 – Solution
Strategy. Conservation, preservation of the temperature range and decay of a quadratic norm impose different conditions on a numerical scheme.
Step 1: Balance each cell. For the first cell, inflow minus outflow gives . Applying the same balance to the others yields Every column sums to one, so is conserved. Equal cell volumes and constant capacity make this proportional to total heat. Every row also sums to one, so constant profiles are fixed.
Step 2: Characterize range preservation. All entries are nonnegative exactly when . Each new value is then a convex combination of old values, preserving their range. For any , input gives a middle value , outside . Thus the exact range is .
Step 3: Test the quadratic energy on each mode. The stated mutually orthogonal vectors have update factors , and , respectively. The mean is fixed; the remaining squared mode components do not grow exactly when and . For , this is Necessity follows by choosing the offending eigenvector as input. This threshold belongs to this particular three-cell insulated matrix.
Step 4: Exhibit the distinction numerically. At , Both sums equal one. The uncentered squared norm drops from to ; since the mean is unchanged, the centered squared norm also drops. Nevertheless the middle temperature becomes negative. The scheme is quadratically stable here but fails the discrete range principle. The stem plot shows individual cell values, without suggesting a smooth interpolated temperature profile between cell centers.
See the diagram in the original worksheet below.