The pitch and heave of a simulated aircraft, the bending of a wing spar and the vibration of a multi-storey building do not fit into a second-order equation. A linearized airplane has four longitudinal state variables, and eliminating all but one leaves a fourth-order equation, (from memory: the two oscillations it contains are the slow, lightly damped phugoid and the fast, well-damped short-period mode). The good news from 5.2 Constant Coefficients & the Characteristic Equation carries over unchanged: try , and calculus becomes algebra. What changes is that a degree-4 polynomial can have repeated roots and two conjugate pairs, and finding its roots is no longer a formula. This lesson sets up the th-order theory, ties it to first-order systems through the companion matrix, and ends with a numerical warning about high-degree root finding.
Transcript
- A linearized aircraft has four longitudinal states; eliminating all but one leaves a fourth-order equation. From memory, it hides a slow phugoid and a fast short-period oscillation. With constant coefficients, try e to the r t, and calculus becomes algebra.
- Write D for d by d t, so the equation is p of D applied to y equals zero. Since D to the k sends e to the r t to r to the k times itself, p of D gives p of r times e to the r t. A solution exactly when p of r is zero.
- Why exactly n? Send each solution to its initial data. The map is linear. It is one-to-one, since zero data gives the zero function, by uniqueness, and onto, since every data vector has a solution, by existence. A linear bijection preserves dimension: n.
- Example one: the polynomial factors as r minus 2, r plus 1, r plus 3. The initial data give a Vandermonde system with determinant minus 30, and c is 1, 3, minus 2. The root 2 grows; the other modes die.
- Abel's formula. Differentiate the Wronskian one row at a time: only the last row survives, and the equation turns it into minus 2 times the row y double prime. So W prime is minus 2 W. At t equals 0, one half and 1 the ratio is exactly 1.
- D minus r acting on u times e to the r t just differentiates u. Applied m times, it gives the m-th derivative. For m equals 3, t to the k e to the r t is killed for k from 0 to 2, and k equals 3 leaves 6 e to the r t.
- Example three: r squared plus 2r plus 5, squared, expands to r to the fourth plus 4r cubed plus 14r squared plus 20r plus 25, with roots minus 1 plus or minus 2 i, each twice. Applying p of D to these three functions gives zero; with t squared it does not.
- Make it a first-order system. An eigenvector must be one, lambda, lambda squared, lambda cubed, and the last row then says p of lambda is zero. But C minus lambda I has rank 3 of 4: one eigenvector for a double root. The missing direction is the factor t.
- Stability is the signs of the real parts. Positive coefficients are not enough: r cubed plus r squared plus r plus 2 has a pair at 0.1766 plus or minus 1.2028 i. In the k family, 5 keeps all roots left, 9 puts a pair on the axis, 10 has crossed. Routh-Hurwitz, from memory, tests this.
- Wilkinson's polynomial has roots 1 through 20. At 20 the derivative is 19 factorial, so the sensitivity is about 4.31 times 10 to the 7; a change of 2 to the minus 23 predicts a shift near 5. In floating point, 10 of 20 roots go complex.
- A repeated root is worse. Move the constant term of x minus 1 to the fourth by 10 to the minus 12, and the roots move by 0.001: nine orders of magnitude more. Even exact coefficients scatter them by about 0.0002. For multiplicity m, the error is about epsilon to the 1 over m.
- From memory, a beam obeys E I y four prime equals q. Clamp one end, free the other, integrate four times: the tip deflection is q L to the fourth over 8 E I, which is 3.3333. A boundary-value solver agrees, and gives 1.1806 at midspan.
- To recap: dimension n; multiplicity gives t to the k e to the r t; the companion matrix has the roots as eigenvalues; stability reads the real parts; repeated roots are ill-conditioned. Next: series solutions.
The equation and its polynomial
Write for the derivative operator. Then and the equation
reads with the characteristic polynomial . The coefficients are constants, which is what makes and its powers commute and lets us treat like ordinary algebra. The key computation is the same as before: , so by linearity
The function solves the equation exactly when . If has distinct roots we get solutions at once; the rest of the lesson is about whether that count is right, and what to do when roots coincide.
Why exactly n solutions
5.1 Second-Order Linear Equations: Structure showed that for the solutions form a 2-dimensional space. The argument extends word for word. Set . Then , where the last component of is and the other components just shift (the matrix is written out below). The right-hand side is linear in , so it is continuous and Lipschitz everywhere with constant , and the existence-and-uniqueness theorem for first-order systems (the vector version of 3.7 Existence & Uniqueness, used here without re-proof) gives, for every initial vector, exactly one solution, defined for all .
Now look at the map that sends a solution to its initial data in . It is linear. It is injective because the solution with zero data is the zero function (uniqueness). It is surjective because every vector of data has a solution (existence). A linear bijection preserves dimension, so the solution space has dimension exactly . Any independent solutions form a basis, and they are independent exactly when their initial-data vectors are, i.e. when the determinant of the matrix with columns is nonzero. That determinant is the Wronskian at , taken up below.
Distinct real roots
If has distinct real roots , then . The initial-data matrix is (row is the th derivative at ), a Vandermonde matrix, whose determinant is (from memory). It is nonzero exactly when the roots are distinct, which settles independence in this case.
Example 1. Solve with , , . The polynomial equals , so . The data give the system
Its determinant is , and solving gives , so (check: and ). The root makes this solution grow while the other two modes die.
Repeated roots: the shift identity
Suppose is a root of multiplicity , meaning with (call the root for the moment). Then , and the factors commute, so any with solves the equation. How do we find all such ? Compute how acts on a function written as :
This shift identity says that acting on simply differentiates and leaves the exponential alone. Applying it times,
Now the multiplicity rule is immediate: exactly when is a polynomial of degree less than . So all solve the equation, and does not, since . (The lab cell below checks this with sympy for , and prints for .) One root of multiplicity contributes solutions, and multiplicities add up to , so the count is complete.
Independence across different roots: suppose with . Applying kills the first group and maps each other group to , a polynomial of the same degree, nonzero whenever is, since . Repeating removes the groups one by one until a single term is left, which vanishes only if . So every is zero.
Example 2. Solve . Here , so . To see that really works, differentiate: , , , and the combination is
Complex roots and conjugate multiplicity
Real coefficients make complex roots come in conjugate pairs with equal multiplicity. The reason: , and the same holds for every derivative, . A root of multiplicity is a number where vanish and does not, and conjugation preserves that pattern. The shift-identity derivation never assumed is real, so solves the equation for . The operator has real coefficients, so the real and imaginary parts of a complex solution are separately solutions:
A pair of multiplicity accounts for roots and gives real functions: in total.
Example 3. For the polynomial is (expanding gives ), with roots , each twice. The general real solution is
The envelope is , so it can grow briefly before decay wins. With repeated pure imaginary roots the same factor makes the amplitude grow without bound, which is resonance (5.6 Forced Oscillations & Resonance).
In the lab, the double pair preset reads out the polynomial and a basis containing , as in Example 3. Drag one root of the pair away from the other and the factors disappear with the multiplicity. The crosses are eigenvalues of the companion matrix, found by a QR iteration in the sim with no polynomial root finder; they sit on the dots to about for distinct roots, but only to about for the double pair, for the triple root and for the quadruple root. That degradation is explained in the numerics section.
The Wronskian of n functions, and Abel’s formula
For functions the Wronskian is the determinant of the matrix whose rows are the functions and their first derivatives:
For solutions of our equation, means the initial-data vectors are independent, hence the solutions are a basis, and the next computation shows is then never zero.
Abel’s formula says , hence . The derivation for carries over to any . Differentiating a determinant differentiates one row at a time and adds the results. For the rows , differentiating row 1 or row 2 produces a repeated row and a zero determinant. Only the term where the last row becomes survives. Each column satisfies , so in that last row we can subtract times row 2 and times row 1 (adding multiples of other rows does not change a determinant) and are left with times the row . Therefore .
The coefficient has a second meaning. The matrix in is a fundamental matrix of (its columns are solutions) and is the trace of , so Abel’s formula is (Liouville, from memory). It also matches the roots, whose sum is . For Example 1, and the roots sum to (also ), so ; block E of the lab cell evaluates the determinant at and the ratio to is each time.
The companion matrix: from one equation to a system
With and , the equation becomes with
This is how every ODE solver handles a higher-order equation: 4.3 Runge–Kutta Methods & Butcher Tableaux methods and RK45 only know first-order systems, so you hand them (or the nonlinear analogue).
Its eigenvalues are exactly the roots. Here is the concrete reason, without any determinant. An eigenvector with satisfies, row by row for the first rows, . So , and the last row demands , which rearranges to , i.e. . Conversely every root gives such an eigenvector.
The determinant agrees. For ,
In general, expand along the first column, whose only entries are (top) and (bottom). The minor under has the same shape with coefficients ; the minor under is triangular with diagonal , so with its sign it contributes . By induction .
The eigenvector formula also shows why repeated roots are special: for each the eigenvector is forced onto the single line through , however often is a root. A root of multiplicity has one eigenvector, so is not diagonalizable (1.6 Change of Basis & Diagonalization needs a full set). For Example 3’s matrix and , the rank of is of (block F below): one eigenvector, not two. The missing directions are generalized eigenvectors, and in they appear as the factors , the system-language reason for (see 1.8 The Matrix Exponential: From Hops to Flows).
Stability from the roots
Every basis function has the form or (possibly ). A power of never beats an exponential, so such a term tends to if and is unbounded if . Since every initial vector corresponds to a solution (the bijection above), we get the criterion:
- All roots have if and only if every solution decays to (and then decays exponentially).
- Some root has : the initial data of that basis function give a solution that grows without bound.
- Roots on the imaginary axis give bounded solutions only if each is simple; multiplicity or more brings back a factor (the lab’s “double undamped pair” lists ).
Checking signs of real parts without computing roots is the job of the Routh-Hurwitz criterion, a set of inequalities on the coefficients; I use only its name and one consequence I can prove. If all roots have negative real part, factors over the reals into with and with , and a product of polynomials with positive coefficients has positive coefficients. So stable implies all coefficients have the same sign. The converse fails already at degree : has all coefficients positive but roots and , with a positive real part. (For a monic cubic with positive coefficients the exact condition is , from memory; here .)
Numerical honesty: roots of high-degree polynomials
Computing roots is delicate. If the coefficient changes by , a simple root moves, to first order, by (differentiate with respect to ).
Wilkinson’s example takes . At the root the derivative is , but the numerator for the coefficient is , so the first-order sensitivity is : a change of in the coefficient () predicts a shift of about . That is too big for a linearization, so we ask the computer. The lab cell below uses np.roots, which takes eigenvalues of the companion matrix, with exact integer coefficients converted to double precision. Unperturbed, the largest distance from a root to the nearest integer is (coefficients up to do not fit in 53 bits). After subtracting from the coefficient, of the roots become complex, with imaginary parts up to . The roots through stay at the integers to six decimals ( is small against there); the large ones scatter.
Repeated roots split. Take and change the constant term by , so that and . The four roots are and : a perturbation of moves the roots by , nine orders of magnitude larger. The cell reports from np.roots (exactly in exact arithmetic; the last digits are rounding). The unperturbed polynomial gives , because rounding alone acts like a perturbation of machine epsilon, , and . The same mechanism explains the lab’s eigenvalue errors: for multiplicity the error is of order , namely , and for , within a factor of about five of the , , the sim shows.
The beam equation: a fourth-order preview
A slender beam with bending stiffness (Young’s modulus times the second moment of area) carrying a distributed load per unit length obeys the Euler-Bernoulli equation , with the deflection at position along the beam (from memory; it assumes small deflections and ignores shear deformation). Physically is the bending moment and the shear force. The homogeneous polynomial is , the root with multiplicity , so the homogeneous solutions are : what four integrations produce. A constant load is times a constant, which resonates with the root four times over, so undetermined coefficients (5.4 Nonhomogeneous Equations: Undetermined Coefficients) multiplies the guess by : gives , so .
A beam is held at both ends, so the four conditions are split between them. Take a cantilever of length : clamped at () and free at (no moment, no shear: ). Integrate from the free end:
where the constant in makes . One more integration, with , gives
For , , this is ; block C of the lab cell solves the same four conditions as a linear system and gets , and scipy.integrate.solve_bvp (same block) independently returns the same tip value and a midspan deflection , which agrees with the formula above at , namely . A wing spar is a crude cantilever and real lift is not uniform, so this shows the shape of the model only.
On a spring foundation the equation is with . For , has roots (the lab’s foundation preset).
Where the method stops
Everything here rests on constant coefficients, which make an eigenfunction of . With variable coefficients there is no characteristic polynomial; the solution space still has dimension (the existence-and-uniqueness argument did not use constancy), but its basis must come from series (the next topic) or numerical integration. Euler-type equations, with a constant times , have a polynomial of their own (try ), mentioned only in passing. Nonlinear equations have no solution space; the form still simulates them, and linearization (2.2 Linearization, the Chain Rule & the Jacobian) brings the characteristic polynomial back near an equilibrium.
Problems
Solve with , , . Which term dominates for large ?
Show worked solution
, so the roots are and and is a convenient real basis. Derivatives at : , , , . The data give , , so ; and , , so . Then
Check: , , . The term dominates; the others are bounded.
Consider . (a) Find the general real solution. (b) For which solutions is bounded as ? (c) Solve the initial-value problem with , , , , .
Show worked solution
(a) : the root has multiplicity and the root multiplicity , so
(Shortcut: solves ; integrating twice adds .)
(b) The terms decay no matter how large the polynomial in front is, so . This is bounded exactly when . The double root at is a case of the “imaginary axis roots must be simple” rule.
(c) With : , , . Writing : ; , so ; , so . Thus and . Then gives and gives :
Check: , , , , and . Every nonzero root is stable, yet grows like because of the double root at zero.
Let with a real number. For which do all solutions tend to ? Justify the endpoints, and check one interior value.
Show worked solution
Let . If then while , so there is a positive real root. If then is a root (a constant solution). Neither decays.
For , the roots depend continuously on , so a root can leave the left half-plane only by crossing the imaginary axis, i.e. for some real . Compute
The imaginary part vanishes for or . Then the real part vanishes only if (for ) or (for ). So for no root crosses. At the polynomial is , all roots at , so they stay in the left half-plane for all . At the roots are and (indeed ), and the imaginary pair gives the non-decaying solutions and . For a pair has crossed into the right half-plane (at the roots are and ). Answer: all solutions decay exactly for .
Interior check, : the roots are and , as predicted. (This is the cubic condition , here .)
Key Takeaways
- An th-order constant-coefficient equation is , and turns solving into finding roots of the degree- characteristic polynomial.
- The solution space has dimension exactly , because the map from a solution to its initial data is a linear bijection (existence plus uniqueness for the system ).
- A root of multiplicity contributes for , by the shift identity ; conjugate pairs of multiplicity give and .
- The Wronskian obeys Abel’s formula : , the sum of the roots in the exponent.
- The companion matrix has eigenvector for each root and characteristic polynomial ; a repeated root has only one eigenvector, which is why appears.
- All solutions decay exactly when every root has negative real part; Routh-Hurwitz tests this from the coefficients, and same-sign coefficients are necessary but not sufficient.
- High-degree roots are ill-conditioned (Wilkinson’s degree 20), and a root of multiplicity is only known to about ; integrate the first-order system when it matters.
- The beam equation has polynomial and cantilever tip deflection ; the method fails for variable-coefficient and nonlinear equations.