Fact-checked by Grok 2 weeks ago

Quartic function

A quartic function is a polynomial function of degree four, expressed in the general form f(x) = ax^4 + bx^3 + cx^2 + dx + e, where a \neq 0 and the coefficients b, c, d, e are real numbers. This form distinguishes it from lower-degree polynomials like quadratics (degree 2) or cubics (degree 3), and it represents the highest degree for which general algebraic solutions to the corresponding f(x) = 0 are feasible using radicals. Quartic functions exhibit characteristic end behavior determined by the leading coefficient a: if a > 0, the approaches positive as x tends to both positive and negative , forming a U-shaped or W-shaped ; if a < 0, it approaches negative on both ends. The can have up to three turning points, including local maxima and minima, and up to four real roots, with the sum of the roots given by -b/a via Vieta's formulas. These properties make quartic functions useful in physics and engineering. The roots of a quartic equation x^4 + a_3 x^3 + a_2 x^2 + a_1 x + a_0 = 0 can be found algebraically by reducing it to a depressed form via substitution and solving a resolvent cubic equation, a method developed by Lodovico Ferrari around 1540 and first published by Gerolamo Cardano in 1545. Unlike quintic or higher-degree polynomials, which are generally unsolvable by radicals per the Abel-Ruffini theorem, quartics remain solvable in closed form, though the expressions are often complex and involve nested radicals. This solvability has historical significance in the development of algebra, bridging the quadratic formula and the challenges of higher-degree equations.

Definition and Forms

General Polynomial Form

A quartic function is a polynomial function of degree four, expressed in the general form f(x) = ax^4 + bx^3 + cx^2 + dx + e, where a \neq 0 is the leading coefficient and b, c, d, e are constant coefficients. This form represents the most general quartic polynomial, encompassing all possible terms up to the fourth degree. The leading coefficient a primarily determines the end behavior of the function: if a > 0, both ends of the approach positive as x tends to positive or negative , resulting in an upward-opening ; if a < 0, both ends approach negative , creating a downward-opening . The remaining coefficients b, c, d, and e shape the function's overall form and position: b introduces asymmetry via the cubic term, c affects the quadratic curvature, d influences the linear slope near the origin, and e sets the y-intercept at (0, e). Unlike quadratic functions (degree 2), which have at most one turning point, or cubic functions (degree 3), which have at most two, a quartic function's first derivative is a cubic polynomial that can have up to three real roots, allowing for up to three turning points and more intricate variations in the graph's direction. This higher degree enables behaviors such as a local maximum, local minimum, and another local maximum (or similar configurations), distinguishing quartics in modeling scenarios requiring greater flexibility. For illustration, consider the monic quartic equation x^4 - 1 = 0, a simplified case with a = 1, b = c = d = 0, and e = -1; its roots are x = \pm 1 (real) and x = \pm i (complex), demonstrating how quartics can balance real and non-real solutions.

Depressed Quartic Form

The depressed quartic form simplifies the general quartic equation by eliminating the cubic term, which introduces asymmetry and complicates algebraic manipulations for finding roots. This transformation is a standard preliminary step in solving quartic equations, as it reduces the equation to a more symmetric structure amenable to further resolution techniques, such as . Consider the general monic quartic equation x^4 + ax^3 + bx^2 + cx + d = 0. To depress it, perform the substitution x = y - \frac{a}{4}, which shifts the variable to remove the y^3 term. This choice of shift is derived by expanding the substituted polynomial and setting the coefficient of y^3 to zero, yielding the specific offset -\frac{a}{4}. Substituting and expanding leads to the depressed quartic y^4 + py^2 + qy + r = 0, where the coefficients are given by: p = b - 6\left(\frac{a}{4}\right)^2 = b - \frac{3a^2}{8}, q = c - 2b\left(\frac{a}{4}\right) + 8\left(\frac{a}{4}\right)^3 = c - \frac{ab}{2} + \frac{a^3}{8}, r = d - c\left(\frac{a}{4}\right) + b\left(\frac{a}{4}\right)^2 - 3\left(\frac{a}{4}\right)^4 = d - \frac{ca}{4} + \frac{ba^2}{16} - \frac{3a^4}{256}. These expressions ensure the absence of the linear cubic term while preserving the roots up to the translation. For a non-monic general quartic Ax^4 + Bx^3 + Cx^2 + Dx + E = 0, first divide through by A to make it monic, then apply the above process with a = B/A, b = C/A, c = D/A, d = E/A. As an example, consider the equation x^4 + 2x^3 - x + 1 = 0, where a = 2, b = 0, c = -1, d = 1. The substitution is x = y - \frac{2}{4} = y - \frac{1}{2}. The coefficients are: p = 0 - \frac{3(2)^2}{8} = -\frac{12}{8} = -1.5, q = -1 - \frac{2 \cdot 0}{2} + \frac{8}{8} = -1 + 1 = 0, r = 1 - \frac{-1 \cdot 2}{4} + \frac{0 \cdot 4}{16} - \frac{3 \cdot 16}{256} = 1 + 0.5 - \frac{48}{256} = 1.5 - 0.1875 = 1.3125. Thus, the depressed form is y^4 - 1.5 y^2 + 1.3125 = 0. The roots in x are obtained by subtracting \frac{1}{2} from each root in y.

Properties

Graphical Behavior and Extrema

The graphical behavior of a quartic function f(x) = ax^4 + bx^3 + cx^2 + dx + e, where a \neq 0, is characterized by its even degree, leading to symmetric end behavior on both sides of the graph. As x \to \pm \infty, f(x) approaches +\infty if a > 0, resulting in an overall upward-opening shape, or -\infty if a < 0, yielding a downward-opening graph. This end behavior mirrors that of but allows for more complex intermediate features due to the higher degree. The overall shape of the graph can vary significantly based on the coefficients, commonly forming U-shaped (similar to a parabola with no turning points in between), M-shaped (one local maximum flanked by two local minima), or W-shaped (two local minima separated by a local maximum) configurations. These shapes arise from the interaction of up to three critical points and the function's roots, influencing the number of "hills" and "valleys." A quartic function exhibits even symmetry about the y-axis if b = 0 and d = 0, meaning f(-x) = f(x) and the graph is mirror-symmetric; such functions consist solely of even-powered terms. Quartics cannot be odd functions, as even-degree polynomials do not satisfy f(-x) = -f(x) except in the trivial zero case. To identify local extrema, compute the first derivative f'(x) = 4ax^3 + 3bx^2 + 2cx + d, set it equal to zero, and solve the resulting cubic equation for critical points; this equation can yield up to three real roots, each representing a potential turning point. The nature of each critical point—local maximum or minimum—is determined using the second derivative test: compute f''(x) = 12ax^2 + 6bx + 2c; if f''(x_0) > 0 at a critical point x_0, it is a local minimum, while f''(x_0) < 0 indicates a local maximum; if f''(x_0) = 0, further analysis is needed. If the cubic has fewer than three real roots, the graph has fewer turning points, simplifying to fewer extrema. For example, consider the quartic function f(x) = x^4 - 5x^2 + 4. Its first derivative is f'(x) = 4x^3 - 10x = 2x(2x^2 - 5), with critical points at x = 0 and x = \pm \sqrt{5/2} \approx \pm 1.58. The second derivative f''(x) = 12x^2 - 10 evaluates to f''(0) = -10 < 0 (local maximum at (0, 4)) and f''(\pm \sqrt{5/2}) = 10 > 0 (local minima at approximately (\pm 1.58, -2.25)). This configuration produces a W-shaped with two symmetric minima and a central maximum, illustrating the potential for three extrema in a positive-leading-coefficient quartic.

Inflection Points and Golden Ratio

For a general quartic function f(x) = a x^4 + b x^3 + c x^2 + d x + e, the inflection points occur where the second derivative f''(x) = 12 a x^2 + 6 b x + 2 c = 0. This is a quadratic equation in x, with solutions given by the quadratic formula: x = \frac{ -6b \pm \sqrt{ (6b)^2 - 4 \cdot 12a \cdot 2c } }{ 2 \cdot 12a } = \frac{ -6b \pm \sqrt{ 36 b^2 - 96 a c } }{ 24 a }. The discriminant $36 b^2 - 96 a c determines the number of real roots: positive for two distinct real inflection points, zero for one (a point of inflection at a horizontal tangent), and negative for none. A striking property links these inflection points to the \phi = \frac{1 + \sqrt{5}}{2} \approx 1.618. Consider the passing through the two distinct inflection points at x = \alpha < \beta. This line intersects the quartic curve at two additional points, say at x = \gamma < \alpha and x = \delta > \beta. The x-coordinates satisfy \gamma = \alpha - \frac{\beta - \alpha}{\phi} and \delta = \beta + \frac{\beta - \alpha}{\phi}, or equivalently, the distances form segments in the ratio $1 : \phi : 1, where the outer segments |\gamma - \alpha| = |\delta - \beta| and the middle segment |\alpha - \beta| = \phi \cdot |\gamma - \alpha|. Moreover, the areas between the curve and this —specifically, the two outer regions and the central region—have areas in the ratio $1 : 2 : 1. This connection extends to the solution of intersection points. Subtracting the equation of the from f(x) yields a monic quartic with roots \gamma, \alpha, \beta, \delta and no x^3 term (due to the choice of line), which factors as (x^2 + p x + q)(x^2 - p x + r) = 0. Solving leads to a whose roots incorporate \phi, confirming the proportional spacing. For example, consider f(x) = x^4 - 8x^3 + 18x^2 - 12x + 24, with inflection points at x = [1](/page/1) and x = 3 (where f''(x) = 12(x-[1](/page/1))(x-3) = 0). The y = -4x + 27 intersects the curve again at x = 2 \pm \sqrt{5} \approx -0.236, 4.236, yielding segments of lengths \sqrt{5}-[1](/page/1) \approx 1.236, $2, and $1.236, in the ratio [1](/page/1) : \phi : [1](/page/1). In the case of depressed quartics y^4 + p y^2 + q y + r = 0 with specific coefficients (e.g., chosen such that the resolvent cubic has roots tied to \phi), the inflection points' positions manifest ratios involving \phi through the auxiliary equations in the solution process. For palindromic quartics (where coefficients are symmetric, like x^4 + s x^3 + t x^2 + s x + 1), the inflection points divide the x-projections of the curve into golden proportions under the reciprocal substitution z = x + 1/x, emphasizing the geometric harmony. An illustrative case is the quartic x^4 - x^3 - \frac{3}{2} x^2 + \frac{1}{2} x, whose inflection points at x = \frac{1 \pm \sqrt{5}}{4} directly incorporate \sqrt{5}, linking to \phi via the segment ratios along the inflection line.

Discriminant and Root Analysis

The discriminant of a quartic polynomial p(x) = a x^4 + b x^3 + c x^2 + d x + e, with a \neq 0, is a of degree 6 in the coefficients that provides information about the nature and multiplicity of its without requiring their explicit . It is defined as D = a^{6} \prod_{i < j} (r_i - r_j)^2, where r_1, r_2, r_3, r_4 are the (counted with multiplicity) of p(x) = 0. Equivalently, D can be expressed directly in terms of the coefficients as the 16-term formula: \begin{align*} D &= 256 a^3 e^3 - 192 a^2 b d e^2 - 128 a^2 c^2 e^2 + 144 a^2 c d^2 e - 27 a^2 d^4 \\ &\quad + 144 a b^2 c e^2 - 6 a b^2 d^2 e - 80 a b c^2 d e + 18 a b c d^3 + 16 a c^4 e \\ &\quad - 4 a c^3 d^2 - 27 b^4 e^2 + 18 b^3 c d e - 4 b^3 d^3 - 4 b^2 c^3 e + b^2 c^2 d^2. \end{align*} The sign of the discriminant classifies the root configuration for a quartic with real coefficients: if D > 0, there are either four distinct real roots or no real roots (two pairs of complex conjugate roots); if D < 0, there are two distinct real roots and one pair of complex conjugate roots; if D = 0, there is at least one multiple root. These conditions arise because complex roots occur in conjugate pairs, and the discriminant's value reflects the parity of the number of non-real roots through the resultant structure. For example, consider the quartic equation x^4 - 1 = 0, where a = 1, b = c = d = 0, e = -1. Substituting into the formula yields D = 256(1)^3(-1)^3 = -256 < 0, confirming the presence of two distinct real roots (x = \pm 1) and two roots (x = \pm i). To further analyze root pairings, sub-discriminants can be defined for potential quadratic factors of the quartic. If the quartic factors as (x^2 + p x + q)(x^2 + r x + s), the sub-discriminants are \Delta_1 = p^2 - 4q and \Delta_2 = r^2 - 4s, which determine whether each quadratic has two real roots (\Delta > 0), a repeated real root (\Delta = 0), or two roots (\Delta < 0). The existence of such a factorization over the reals is equivalent to the resolvent cubic having a real root, and the signs of these sub-discriminants help distinguish between the cases of four real roots (both \Delta_1, \Delta_2 > 0) and zero real roots (both \Delta_1, \Delta_2 < 0) when D > 0. The also relates to the of the quartic over (assuming rational coefficients and irreducibility). The transitive subgroups of S_4 are S_4, A_4, D_4 ( of order 8), V_4 (), and C_4 (cyclic of order 4). Whether D is a square in \mathbb{Q} determines if the lies in the A_4 (if square, possible groups A_4 or V_4; if not, possible S_4, D_4, or C_4). This follows from the fact that the is the square of the product of differences of , and its square-root generates a quadratic extension whose splitting behavior indicates the even/odd permutations in the Galois action. All such groups indicate solvability by radicals, consistent with the degree being 4.

Historical Development

Early Contributions

The earliest encounters with problems resembling quartic equations trace back to ancient around 2000 BCE, where scholars developed algorithmic procedures for calculations involving areas and volumes that occasionally led to higher-degree relations, though solved through approximations rather than general algebraic methods. These practical computations, such as determining field dimensions or irrigation volumes, implicitly required handling expressions akin to quartics in geometric contexts, but the Babylonians lacked a formal concept and focused primarily on positive quantities using table-based approximations. In , mathematicians like (c. 287–212 BCE) advanced the study of curves beyond linear and quadratic forms, employing the to compute areas and volumes under parabolas and spirals, which anticipated techniques for higher-degree curves but did not yield a general solution for quartic equations. ' geometric investigations, detailed in works such as On Spirals and , emphasized rigorous proofs for curved figures, setting a foundation for analyzing behaviors without algebraic symbolism. During the medieval , (c. 780–850 ) systematized solutions for quadratic equations in his Al-Kitab al-mukhtasar fi hisab al-jabr wal-muqabala, using geometric completion of squares to handle cases like x^2 + 10x = 39, but his work did not extend to general higher degrees. This algebraic framework was built upon by (1048–1131 ), who in his Treatise on Demonstration of Problems of Algebra employed intersections of conic sections—such as circles and hyperbolas—to geometrically resolve cubic equations and certain biquadratic forms, marking a significant shift toward visualizing higher-degree solutions. In parallel, saw Bhaskara II (1114–1185 CE) address biquadratic equations in Lilavati and Bijaganita, reducing forms like x^4 + ax^2 + b = 0 to quadratics by substitution y = x^2 and incorporating conic-based geometric interpretations for verification and special cases. Bhaskara's methods emphasized practical resolution, including positive roots and approximations, within a broader treatment of indeterminate equations and progressions. As these geometric and proto-algebraic approaches matured by the late medieval period, the transition to the highlighted a toward symbolic , exemplified by Gerolamo Cardano's (1501–1576 CE) focus on cubic equations in Ars Magna (1545), which inadvertently underscored the unresolved challenge of general quartics and paved the way for subsequent innovations.

Renaissance and Modern Solutions

In 1540, Italian mathematician developed the first general algebraic solution to the , reducing it to the resolution of a through a method that involved on a cubic resolvent . This breakthrough, achieved while Ferrari served as a lecturer in , relied on prior advances in solving cubics and marked a significant step in algebraic theory, though it remained unpublished during his lifetime. Ferrari's mentor, , included the solution in his seminal work Ars Magna in 1545, ensuring its dissemination and crediting Ferrari posthumously after his death in 1565. The approach demonstrated that quartics could be solved by radicals, building on the depressed form of the equation where the cubic term is eliminated. By the mid-17th century, geometric interpretations of algebraic solutions gained prominence, with René Descartes outlining methods in his 1637 treatise La Géométrie to construct roots of quartic equations through intersections of conic sections and circles. Descartes' analytic geometry framework allowed for the visualization and resolution of higher-degree equations, including quartics, by translating algebraic problems into geometric constructions that could be performed with ruler and compass. Contemporaneously, Girard Desargues contributed early resolvent techniques in his work on conic sections around 1639, publishing solutions to quartic equations derived from Ferrari and Tartaglia, and integrating projective methods to handle the symmetries of root configurations. These efforts shifted focus toward both geometric and algebraic resolvents, laying groundwork for later simplifications. In the , Leonhard Euler refined the solution process in works such as his Elements of Algebra (1770), introducing trigonometric and hyperbolic identities to express the roots of the depressed quartic more elegantly than Ferrari's radical-heavy approach. Euler emphasized the resolvent cubic's role, showing how its roots could be paired using sums of square roots, and developed formulas that avoided excessive nesting of radicals by leveraging identities like those for cosine of multiple angles. His methods, detailed in correspondence and treatises from the onward, highlighted the quartic's solvability while anticipating connections to elliptic functions, influencing subsequent algebraic developments. The 19th and early 20th centuries saw deeper theoretical insights through ' (1831), which classified the Galois groups of irreducible quartics as transitive subgroups of the S_4, including S_4, A_4, the D_4, the V_4, and the C_4. This framework proved all quartics solvable by radicals, as their Galois groups are solvable, contrasting with the general quintic's S_5 group, which is not. Refinements to the , originally proposed in 1683, emerged in the through extensions by mathematicians like Charles Hermite and , who used quadratic substitutions to further depress quartics and analyze their resolvents under Galois actions, facilitating computations and root separation. These advances, building on Galois' memoir, solidified the algebraic structure of quartic solutions by the early 1900s. In modern mathematics, computer algebra systems such as Mathematica and implement explicit radical solutions for quartics using Ferrari's or Euler's methods, enabling symbolic computation of roots with high precision for practical applications, while numerical algorithms like those based on eigenvalue handle ill-conditioned cases efficiently. Unlike quintics, which lack general radical solutions per Abel-Ruffini, quartics remain fully hand-solvable, though computational tools streamline the often cumbersome expressions involving nested . This timeline—from Ferrari's 1540 innovation to 20th-century theoretical closure—underscores the quartic's pivotal role in the evolution of solvable polynomials.

Applications

In Physics and Mechanics

Quartic functions play a significant role in modeling in anharmonic oscillators, where deviations from are captured by including higher-order terms in the potential. A common form is the quartic potential V(x) = \frac{1}{2} k x^2 + \lambda x^4, with k > 0 representing the stiffness and \lambda > 0 introducing nonlinearity that leads to asymmetric or bounded motion depending on levels. This model is widely used for nonlinear springs in systems, where large displacements cause stiffening or softening effects. In beam deflection analysis under the Euler-Bernoulli theory, the governing equation for transverse displacement w(x) of a under distributed load q(x) is the fourth-order EI \frac{d^4 w}{dx^4} = q(x), where E is the modulus of elasticity and I is the . For curved beams subjected to loads, integrating this equation yields solutions where the deflection profile involves quartic polynomials, particularly for uniform or linearly varying loads, allowing precise prediction of bending behavior in arched or circular structures. In , quartic perturbations arise in the for the perturbed , modifying the standard inverse-square to account for multipolar expansions or additional forces in multi-body systems. The can include quartic terms in the radial coordinate expansion, influencing orbital stability and in systems like binary stars or planetary rings, where small deviations lead to bounded or trajectories. Quartic is essential in for applied to molecular vibrations, where the expands as V(q) = \frac{1}{2} \omega^2 q^2 + \frac{1}{3!} \phi_3 q^3 + \frac{1}{4!} \phi_4 q^4 + \cdots, with the quartic term \phi_4 q^4 correcting harmonic approximations for frequencies and Fermi resonances in diatomic or polyatomic molecules. This approach computes anharmonic corrections to vibrational spectra by treating the quartic contribution as a to the unperturbed . A prominent example is the Duffing equation, \ddot{x} + \delta \dot{x} + \alpha x + \beta x^3 = \gamma \cos(\omega t), which stems from a quartic potential V(x) = \frac{1}{2} \alpha x^2 + \frac{1}{4} \beta x^4 and models nonlinear oscillations in driven systems like mechanical resonators. In phase space, trajectories approximate closed curves for low energies but exhibit bistability and chaos for higher drives, with quartic terms enabling analysis of amplitude-dependent frequency shifts via perturbation methods.

In Geometry and Other Fields

In , plane curves, defined by homogeneous polynomials of four in the \mathbb{P}^2, serve as canonical models for , non-hyperelliptic curves of three. The g of a plane curve of d=4 is computed via the formula g = \frac{(d-1)(d-2)}{2} = 3, reflecting their role in studying linear series, Weierstrass points (with 24 such points on a ), and moduli spaces. These curves also arise as complete intersections of two surfaces in \mathbb{P}^3, yielding space quartics of four, often elliptic (genus one) when , which embed them in higher-dimensional projective spaces for enumerative purposes. The Cayley-Bacharach theorem and its generalizations apply to quartic curves in , asserting that for two plane curves of degrees d_1 and d_2 intersecting transversely at d_1 d_2 points, any curve of degree d_1 + d_2 - 3 passing through all but one of these points must pass through the remaining point. For instance, when d_1 + d_2 - 3 = 4, such as the intersection of a line and a sextic (yielding six points) or a conic and a quintic (yielding ten points), the theorem constrains quartic curves through all but one of them to include the last point, aiding in counting configurations like bitangents on quartics. Beyond geometry, quartic functions appear in optimization problems, such as quartic Bézier curves in computer graphics, where a degree-four parametric polynomial \mathbf{B}(t) = \sum_{i=0}^{4} \binom{4}{i} (1-t)^{4-i} t^i \mathbf{P}_i (with control points \mathbf{P}_i) models complex smooth paths for rendering and animation, offering greater flexibility than cubics while maintaining C^1 continuity at joins. In economics, quartic objective functions model nonlinear cost or utility surfaces in multivariate optimization, capturing higher-order interactions in resource allocation beyond quadratic approximations. In statistics, the quartic (fourth) moment \mu_4 = \mathbb{E}[(X - \mu)^4] enters higher-order cumulants, such as the fourth cumulant \kappa_4 = \mu_4 - 3 \mu_2^2, which quantify non-Gaussian deviations in distributions for tail analysis and independence testing. A representative example is the , a -like quartic defined by (x^2 + y^2)^2 - 2a^2(x^2 - y^2) = b^4 - a^4, where the product of distances from any point to two fixed foci (at (\pm a, 0)) equals b^2; for b > a\sqrt{2}, it forms a single oval, at b = a\sqrt{2} a figure-eight , for a < b < a\sqrt{2} a dog-bone shape, and for b < a two separate ovals, illustrating quartic symmetry in polar coordinates.

Solving Methods

Reduction Techniques

Reduction techniques for quartic equations involve algebraic substitutions and factorizations that simplify the general form ax^4 + bx^3 + cx^2 + dx + e = 0 before applying more advanced solution methods. These approaches aim to eliminate specific terms or decompose the polynomial into lower-degree factors solvable by radicals. The Tschirnhaus transformation is a key substitution method that maps the variable x to a rational function y = g(x)/h(x), where g and h are polynomials, to simplify the equation by removing higher-degree coefficients. For quartics, after the initial depression to eliminate the cubic term (via x = y - b/(4a)), further Tschirnhaus transformations can target the quadratic term, reducing the equation to a form like y^4 + py^2 + qy + r = 0 or even a biquadratic. This involves solving for transformation parameters that satisfy conditions derived from coefficient matching, often leading to a cubic auxiliary equation. Such reductions facilitate subsequent factoring or resolution. A common reduction exploits the possibility of factoring the quartic into two quadratics over , assuming ax^4 + bx^3 + cx^2 + dx + e = (x^2 + px + q)(x^2 + rx + [s](/page/%s)) after scaling to monic form and depressing. Expanding yields the system: p + r = b/a, q + [s](/page/%s) + pr = c/a, ps + qr = d/a, qs = e/a. Solving this nonlinear system determines if rational p, q, r, [s](/page/%s) exist, often via a resolvent cubic in a variable like z = (p - r)^2. Reducibility over is checked first by the , which tests possible linear s \pm factors of e/a divided by factors of a; if a rational \rho is found, synthetic division yields a cubic factor. For irreducible linear factors but possible ones, the resolvent cubic must have a positive rational that is a , confirming the . If the resolvent has no such and no linear factors exist, the quartic is irreducible over . For illustration, consider the reducible quartic x^4 + 5x^2 + 4 = 0, which factors as (x^2 + 1)(x^2 + 4) = 0 by assuming factors with zero linear terms and solving the resulting system for the constants. A special case is the biquadratic equation x^4 + ax^2 + b = 0, where odd-powered terms vanish. This reduces to a via the y = x^2, yielding y^2 + ay + b = 0, solved by the y = \frac{-a \pm \sqrt{a^2 - 4b}}{2}; the roots are then x = \pm \sqrt{y} for each positive real y. This form arises naturally after depressing a general quartic with zero linear and cubic coefficients.

Ferrari's Solution

Lodovico Ferrari developed a in the 1540s to solve the general by first reducing it to the depressed form y^4 + p y^2 + q y + r = 0, where the linear term is absent, and then introducing a parameter that leads to a resolvent whose solution allows into quadratics. This approach, published posthumously in Gerolamo Cardano's Ars Magna (1545), expresses the roots using nested radicals, building on Cardano's cubic solution. The resolvent cubic arises from and setting the expression as a difference of squares involving an auxiliary variable. The resolvent cubic for the depressed quartic y^4 + p y^2 + q y + r = 0 is z^3 + \frac{p}{2} z^2 + \frac{p^2 - 4r}{16} z - \frac{q^2}{64} = 0. This equation is derived by adding and subtracting terms to write the quartic as (y^2 + \frac{p}{2} + z)^2 - (2 z y^2 + q y + 2 z \frac{p}{2} + z^2 + r - (\frac{p}{2})^2) = 0, then choosing z such that the subtracted term is a perfect square. Since the resolvent cubic always has at least one real root for real coefficients (due to the intermediate value theorem and the behavior of cubics), select that real root z. Then define m = \sqrt{2z + \frac{p}{2}}, assuming the principal square root for real positive argument when possible. The roots of the original quartic are obtained by solving the resulting pair of quadratics from the difference of squares factorization. These expressions may involve intermediates even for quartics with all real roots, particularly in the where the resolvent cubic has three real roots but the quartic has two real and two roots. Real roots are extracted by evaluating all combinations and selecting those that are real, or by pairing conjugate terms to ensure real outputs.

Alternative Approaches

Descartes developed a geometric construction for solving the depressed x^4 + p x^2 + q x + r = 0 by finding the intersection points of a parabola y = x^2 and a suitably chosen . The is rewritten by splitting the p x^2 term and substituting y = x^2, yielding y^2 + (p-1)y + x^2 + q x + r = 0; in both y and the linear terms in x transforms the second part into a , such as (y - h)^2 + (x - k)^2 = \rho for appropriate h, k, \rho. The real roots correspond to the x-coordinates of these intersection points, providing a ruler-and-compass constructible when intersections exist in the real plane. Euler offered an alternative algebraic resolution for the depressed quartic x^4 + p x^2 + q x + r = 0, expressing each as a of square roots: \pm \sqrt{r_1} \pm \sqrt{r_2} \pm \sqrt{r_3}, where the r_i are positive of the resolvent cubic t^3 + 2 p t^2 + (p^2 - 4 r) t - q^2 = 0. The correct sign choices are selected to satisfy \sqrt{r_1 r_2 r_3} = -q/2, ensuring the roots sum appropriately to match the coefficients; this method highlights the quartic's solvability by radicals without introducing auxiliary cubics beyond the resolvent. For cases with all real roots or specific coefficient relations, can express the roots explicitly, replacing square roots with forms like \sqrt{A} \cosh \phi + \sqrt{B} \sinh \psi, to avoid intermediates. Trigonometric identities apply in restricted scenarios, such as when the quartic admits roots expressible via multiple . A concrete illustration of Euler's trigonometric adaptation arises for the equation y^4 + a y^2 + b y + c = 0. By scaling y = k \tan \theta and aligning coefficients with the quadruple-angle formula \tan 4\theta = \frac{4 \tan \theta - 4 \tan^3 \theta}{1 - 6 \tan^2 \theta + \tan^4 \theta}, the equation reduces to solving for \theta such that \tan 4\theta equals a constant derived from a, b, c, k; the four roots then follow as k \tan(\theta + m \pi/4) for m = 0,1,2,3, leveraging known identities for angle addition. This approach simplifies computation when the discriminant indicates four real roots aligned with angular separations. Lagrange advanced the resolution of quartics through resolvents, treating the roots \alpha_1, \alpha_2, \alpha_3, \alpha_4 of x^4 + a x^3 + b x^2 + c x + d = 0 (first depressed if needed) and forming symmetric functions like f = (\alpha_1 + \alpha_2)(\alpha_3 + \alpha_4). The orbit of f under the action of the S_4 generates a cubic resolvent in the elementary symmetric polynomials (the coefficients a,b,c,d), solvable to yield the f_i; pairing these then allows quadratic factorizations to recover the \alpha_i. This framework, rooted in permutations of roots, prefigures by revealing the quartic's as a of S_4 with a isomorphic to S_3, thus explaining the resolvent cubic's role in the . From an perspective, the roots of a general quartic can be interpreted geometrically as the points (in suitable coordinates) of two hypersurfaces in projective 3-space \mathbb{P}^3. Embedding the affine into transforms it into a of two equations defining the quadrics; parameterizing their (a genus-1 ) via or birational maps to a plane conic yields explicit radical expressions for the points, aligning with classical solutions while facilitating computational verification through computations or eigenvalue methods on the quadric pencils. These alternative methods often simplify under symmetry conditions; notably, Euler's resolvent approach streamlines for palindromic quartics like x^4 + e x^3 + f x^2 + e x + 1 = 0, where appear in pairs r, 1/r. The w = x + 1/x reduces the equation to a biquadratic w^2 + (f-2) w + (e^2 - 2 e w + 1) = 0 wait, actually to a in w, solvable directly, with then found via quadratics x^2 - w x + 1 = 0, avoiding the full cubic resolvent.

Special Solvable Cases

Certain subclasses of quartic equations admit simplified solution procedures that reduce them to quadratics or lower-degree equations, avoiding the full complexity of the general case. The biquadratic equation x^4 + a x^2 + b = 0 is solved by the z = x^2, which transforms it into the z^2 + a z + b = 0. The solutions are z = \frac{-a \pm \sqrt{a^2 - 4b}}{2}, and the corresponding x-roots are x = \pm \sqrt{z} for each valid z. For example, the equation x^4 - 5x^2 + 4 = 0 yields z^2 - 5z + 4 = 0, with roots z = 4 and z = 1, so x = \pm 2, \pm 1. Reducible quartics factor into products of quadratics, such as (a x^2 + b x + c)(d x^2 + e x + f) = 0, allowing the roots to be found by solving each quadratic factor separately via the . To determine the factorization over , one computes the resolvent cubic z^3 + 2 c z^2 + (c^2 - 4 e) z - d^2 = 0 (for the depressed form x^4 + c x^2 + d x + e = 0) and checks for rational square roots among its roots, which provide the linear coefficients for the factors. Quasi-palindromic or quartics, of the form x^4 + p x^3 + q x^2 + p x + 1 = 0, are addressed by dividing by x^2 (assuming x \neq 0) to obtain x^2 + p x + q + p x^{-1} + x^{-2} = 0, or equivalently (x + x^{-1})^2 + p (x + x^{-1}) + (q - 2) = 0. The z = x + x^{-1} reduces this to the z^2 + p z + (q - 2) = 0. Solving for z and then solving x^2 - z x + 1 = 0 for each z yields , which come in reciprocal pairs. Trinomial quartics of the form x^4 + a x + b = 0 (a depressed case with no x^3 or x^2 terms) can be resolved into quadratics by assuming a (x^2 + r x + s)(x^2 - r x + t) = 0, which leads to the system s + t - r^2 = 0, r (s - t) = a, and s t = b. This requires solving the auxiliary u^3 - 4 b u - a^2 = 0 for u = r^2, after which the quadratics are solved.

Numerical and Computational Methods

Numerical methods provide efficient approximations for the roots of general quartic polynomials, particularly when symbolic solutions are cumbersome or when high precision is required for ill-conditioned cases. These techniques are essential in computational software and scientific applications where exact radical expressions may lead to numerical instability due to limitations. The Newton-Raphson iteration is a widely used root-finding method for quartic equations, applicable to the univariate case f(x) = a_4 x^4 + a_3 x^3 + a_2 x^2 + a_1 x + a_0 = 0. The update rule is given by x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}, where f'(x) = 4a_4 x^3 + 3a_3 x^2 + 2a_2 x + a_1. Starting from a suitable initial guess, such as one obtained from graphing or bounds like Cauchy's rule, the method converges quadratically to simple roots, making it effective for isolating individual real or complex roots of quartics. For quartics with multiple roots, convergence may slow, requiring deflation after finding one root to reduce the degree iteratively. For simultaneous approximation of all four roots, the Durand-Kerner method extends Newton's approach to multivariate . It initializes four guesses z_1^{(0)}, z_2^{(0)}, z_3^{(0)}, z_4^{(0)} (often on a circle enclosing the ) and updates them via z_k^{(n+1)} = z_k^{(n)} - \frac{f(z_k^{(n)})}{\prod_{j \neq k} (z_k^{(n)} - z_j^{(n)})}, \quad k = 1,2,3,4. This simultaneous refinement converges cubically for simple under suitable starting points, avoiding the need for sequential and handling naturally. The method is particularly robust for quartics with clustered , though it may require safeguards against stagnation in symmetric cases. Eigenvalue-based methods transform the root-finding problem into a matrix eigenvalue , leveraging stable linear algebra routines. For a monic quartic x^4 + a x^3 + b x^2 + c x + d = 0, the roots are the eigenvalues of the 4×4 C = \begin{pmatrix} 0 & 0 & 0 & -d \\ 1 & 0 & 0 & -c \\ 0 & 1 & 0 & -b \\ 0 & 0 & 1 & -a \end{pmatrix}. The , an iterative orthogonal , computes these eigenvalues efficiently with O(n^3) for n=4, ensuring backward stability where small perturbations correspond to small changes in the input . This approach excels for quartics, as the small allows rapid convergence, often in fewer than 20 iterations for double precision. Modern software implements these techniques seamlessly. In Mathematica, the NSolve function applies hybrid numerical methods, including eigenvalue solvers, to quartic polynomials, supporting arbitrary precision to mitigate rounding errors. Similarly, NumPy's numpy.roots function in computes roots via the and , but exhibits precision loss for ill-conditioned coefficients spanning multiple orders of magnitude, as seen in Wilkinson's examples where tiny perturbations shift roots dramatically. To illustrate sensitivity, consider the quartic x^4 + 0.0001 x^3 - x + 1 = 0, where the small cubic introduces near-degeneracy. Numerical evaluation using eigenvalue methods yields approximate around $0.25 \pm 0.93i and -0.75 \pm 0.25i, but perturbing the coefficient by even $10^{-6} can alter real parts by up to 0.1, highlighting multiple root vulnerability in floating-point computations. Such cases underscore the need for assessment via the before applying iterative solvers. Compared to symbolic methods like Ferrari's solution, numerical approaches offer advantages in speed and robustness for near-degenerate quartics, avoiding in radical expressions and enabling scalable on modern . They handle arbitrary coefficients without risks inherent in exact formulas, making them preferable for simulations and large-scale .

References

  1. [1]
    Cubics and quartics - Student Academic Success - Monash University
    Cubic and quartic function. s are the natural extension of polynomial. s after quadratics, representing higher-order.
  2. [2]
    Quartic Equation -- from Wolfram MathWorld
    A quartic equation is a fourth-order polynomial equation of the form z^4+a_3z^3+a_2z^2+a_1z+a_0=0.
  3. [3]
    Quartic Equation - (Intermediate Algebra) - Fiveable
    In engineering, quartic equations can describe the dynamics of mechanical systems or the characteristics of electrical circuits. In economics, quartic equations ...
  4. [4]
    Bob Gardner's "The Bicentennial of Evariste Galois" Brief History of ...
    Oct 19, 2011 · A Brief History of Equations: Quartics. In 1540, Lodovico Ferrari solved the quartic equation. Five years later, in 1545 Gerolamo Cardano ...
  5. [5]
    The Solution of the Quartic Equation in One Variable
    The purpose of this thesis has been to give a brief account of the history of the quartic equation, to develop and illustrate the particular solutions of the ...
  6. [6]
    BioMath: Polynomial Functions
    Degree 3, 4, and 5 polynomials also have special names: cubic, quartic, and quintic functions. Polynomials with degree n > 5 are just called nth degree ...
  7. [7]
    [PDF] Quartic.pdf
    Jun 17, 2020 · To demonstrate the method, Cardano gave rules for solving the depressed quartic equation and then worked out sample problems. Similar to ...
  8. [8]
    3.4: Graphs of Polynomial Functions
    ### Summary of Quartic Polynomials (Degree 4) from 3.4: Graphs of Polynomial Functions
  9. [9]
    [PDF] August 29 MATH 1113 sec. 51 Fall 2018 - Faculty Web Pages
    Even and Odd Symmetry. ▷ Polynomials with only even powers (including 0) are even functions. Polynomials with only odd powers are odd functions. ▷ Even ...
  10. [10]
    Graphing Using First and Second Derivatives - UC Davis Math
    Jun 3, 1998 · THE SECOND DERIVATIVE TEST FOR EXTREMA (This can be used in place of statements 5. and 6.) : Assume that y=f(x) is a twice-differentiable ...
  11. [11]
    Inflection Points of Fourth Degree Polynomials
    By removing the line through the inflection points of a fourth degree polynomial, the polynomial acquires a vertical axis of symmetry. The golden ratio pops ...
  12. [12]
    [PDF] The Golden Ratio and Quartic Polynomials 9-16-06
    Quartics without points of inflection. If a b. = then the quartic will not have any point of inflection. The result still holds with L. R. x a b x. = = = . Page ...
  13. [13]
    None
    ### Summary of Key Points on Quartics, Inflection Points, and Golden Ratio Connections
  14. [14]
    Polynomial Discriminant -- from Wolfram MathWorld
    A polynomial discriminant is the product of the squares of the differences of the polynomial roots r_i.<|control11|><|separator|>
  15. [15]
    Discriminant - AoPS Wiki
    ### Summary of Discriminant of a Quartic Polynomial
  16. [16]
    Galois group of a quartic polynomial - PlanetMath
    Mar 22, 2013 · In this case, the discriminant of the quadratic is not a rational square, but it is a rational square times D .Missing: formula | Show results with:formula
  17. [17]
    Quadratic, cubic and quartic equations - MacTutor
    ... the Babylonians (about 1800 BC) were the first to solve quadratic equations. This is an over simplification, for the Babylonians had no notion of 'equation'.
  18. [18]
    The Advanced Mathematics of the Babylonians - JSTOR Daily
    Mar 25, 2016 · In algebra, Babylonians apparently had the means to solve quadratic equations (remember those?) and perhaps even higher-order cubic equations.Missing: degree quartic
  19. [19]
    Archimedes - Biography - MacTutor - University of St Andrews
    Archimedes was the greatest mathematician of his age. His contributions in geometry revolutionised the subject and his methods anticipated the integral calculus ...
  20. [20]
    Archimedes Num An - MacTutor History of Mathematics
    His invention of the water-screw, still in use in Egypt, for irrigation, draining marshy land and pumping out water from the bilges of ships.Missing: quartic | Show results with:quartic
  21. [21]
  22. [22]
    (PDF) Omar Khayyam: Geometric Algebra and Cubic Equations
    Sep 3, 2020 · Khayyam presented methods to nd solutions of cubic equations using geometric constructions by identifying the intersection of hyperbolas, parabolas, circles, ...
  23. [23]
    Bhaskara II (1114 - Biography - MacTutor History of Mathematics
    The topics are: positive and negative numbers; zero; the unknown; surds; the kuttaka; indeterminate quadratic equations; simple equations; quadratic equations; ...Missing: biquadratic conics
  24. [24]
    Bhaskara II – Timeline of Mathematics - Mathigon
    Bhaskara also established that division by zero yields infinity, and solved various quadratic, cubic, quartic and Diophantine equations. Khayyam.Missing: biquadratic | Show results with:biquadratic
  25. [25]
    Algebra - Cardano, Cubic, Quartic - Britannica
    Oct 24, 2025 · Cardano and the solving of cubic ... Renaissance era's most systematic and comprehensive account of solving cubic and quartic equations.
  26. [26]
    Lodovico Ferrari (1522 - 1565) - Biography - MacTutor
    Ferrari discovered the solution of the quartic equation in 1540 with a quite beautiful argument but it relied on the solution of cubic equations so could ...
  27. [27]
    Lodovico Ferrari | Algebraic equations, Cubic equations, Quartic ...
    Oct 1, 2025 · In 1540 he succeeded Cardano as public mathematics lecturer in Milan, at which time he found the solution of the quartic equation, later ...
  28. [28]
    Descartes' Mathematics - Stanford Encyclopedia of Philosophy
    Nov 28, 2011 · In La Géométrie, Descartes details a groundbreaking program for geometrical problem-solving—what he refers to as a “geometrical calculus” ( ...Descartes' Early Mathematical... · La Géométrie (1637) · Book One: Descartes...
  29. [29]
    Mathematics - Analytic Geometry, Coordinates, Equations | Britannica
    Oct 1, 2025 · Descartes's La Géométrie appeared in 1637 as an appendix ... Descartes's goal in La Géométrie was to achieve the construction of solutions ...<|separator|>
  30. [30]
    Girard Desargues – Timeline of Mathematics - Mathigon
    Girard Desargues (1591 – 1661) was a French mathematician, engineer, and architect. He designed numerous buildings in Paris and Lyon, helped construct a dam.
  31. [31]
    [PDF] The quartic equation, invariants and Euler's solution revealed
    𝑟. (17). Thus the quartic discriminant 4𝐼3 − 𝐽2 is simply a constant multiple of 𝑦2. 𝑁𝑟. − ℎ2. 𝑟, the sign of which reflects whether the 𝑥-axis lies between (𝑦2.
  32. [32]
    [PDF] The Elementary Mathematical Works of Leonhard Euler (1707 – 1783)
    Solution of quartic equations. From Elements of Algebra, Sect. IV, Chapter XV ... Euler taught the solution of fourth degree equations by “the rule of ...
  33. [33]
    The quartic equation: Invariants and Euler's solution revealed
    Aug 5, 2025 · The central role of the resolvent cubic in the solution of the quartic was first appreciated by Leonard Euler (1707-1783). Euler's quartic ...
  34. [34]
    [PDF] Galois groups of cubics and quartics (all characteristics)
    We want to describe Galois groups of separable irreducible cubic and quartic polynomials in characteristic 2. For instance, if F is a field of characteristic 2 ...
  35. [35]
    [PDF] Polynomial Transformations of Tschirnhaus, Bring and Jerrard
    Tschirnhaus's transformation is a polynomial substitution y = Tk(x), in which the degree of the transformation k<n can be selected. Tschirnhaus demonstrated the ...
  36. [36]
    The fast quartic solver - ScienceDirect.com
    A fast and highly accurate algorithm for solving quartic equations is introduced. This new algorithm is more than six times as fast and several times more ...Missing: pqr | Show results with:pqr<|control11|><|separator|>
  37. [37]
    (PDF) A Classic New Method to Solve Quartic Equations
    Aug 8, 2025 · ≠ 0. Lodovico Ferrari is attributed with the discovery of the. solution to the quartic in 1540, but since this solution, like all. algebraic ...
  38. [38]
    From quartic anharmonic oscillator to double well potential - arXiv
    Oct 30, 2021 · It is already known that the quantum quartic single-well anharmonic oscillator V_{ao}(x)=x^2+g^2 x^4 and double-well anharmonic oscillator V_{ ...
  39. [39]
    [PDF] Solution of Euler-Bernoulli Beam Equation by Integral Transform ...
    The Euler-Bernoulli beam theory is a simple model for small lateral deflections in beams. This work derives the solution to the Euler-Bernoulli equation ...
  40. [40]
    Anharmonic vibrational computations with a quartic force field for ...
    Aug 8, 2019 · In the VPT2 theory, the anharmonic cubic and quartic terms are treated as a perturbation to the harmonic vibrational Hamiltonian, and the ...Missing: mechanics | Show results with:mechanics<|separator|>
  41. [41]
    Dynamics of a Duffing nanomechanical resonator coupled to a ...
    Nov 11, 2008 · Characteristic of the Duffing potential is a term that is quartic in the oscillator displacement. 35. Depending on this term being added to ...
  42. [42]
    [PDF] Geometry of Algebraic Curves - UChicago Math
    Joe Harris taught a course (Math 287y) on the geometry of algebraic curves at Harvard in Fall 2011. These are my “live-TEXed” notes from the course.
  43. [43]
    [PDF] FOUNDATIONS OF ALGEBRAIC GEOMETRY CLASS 46
    Apr 17, 2008 · all nonsingular complete intersections of quadrics and cubics are genus 4 non- hyperelliptic curves, canonically embedded. By Riemann-Roch ...
  44. [44]
    [PDF] CAYLEY-BACHARACH THEOREMS WITH EXCESS VANISHING ...
    A classical theorem of Cayley and Bacharach asserts that if D1,D2 ⊆ P2 are curves of degrees d1 and d2 meeting transversely, then any curve of degree d1+d2−3 ...
  45. [45]
    A Primer on Bézier Curves - Pomax
    Bézier curves are simply segments of some polynomial curve, so if we pick a wider interval we simply get to see more of the curve. So what do they look like?
  46. [46]
    Optimality conditions and optimization methods for quartic ...
    Apr 1, 2014 · In this paper multivariate quartic polynomial optimization program (QPOP) is considered. Quartic optimization problems arise in various practical applications ...
  47. [47]
    Higher Order Cumulants - an overview | ScienceDirect Topics
    Higher Order Cumulants (HOC) refer to a robust tool used in modelling for various applications involving linear and non-linear systems, which are not ...
  48. [48]
    Cassini Ovals -- from Wolfram MathWorld
    Cassini ovals are quartic curves where the product of distances from a point to two fixed points is constant. The shape depends on b/a.
  49. [49]
    Tschirnhausen Transformation -- from Wolfram MathWorld
    A transformation of a polynomial equation f(x)=0 which is of the form y=g(x)/h(x) where g and h are polynomials and h(x) does not vanish at a root
  50. [50]
  51. [51]
    [PDF] Factoring Quartic Polynomials: A Lost Art - Cal State LA
    If p/q is a rational number in lowest terms such that f(p/q) = 0, then p divides a0 and q divides an. These theorems suffice to factor any quadratic or cubic ...
  52. [52]
    Biquadratic Equation -- from Wolfram MathWorld
    The term "biquadratic equation" is sometimes used to as a synonym for quartic equation (Beyer 1987b, p. 34), but perhaps more commonly (e.g., Hazewinkel 1988; ...
  53. [53]
    Solving The General Quartic Equation
    ... quartic (i.e., fourth degree) equation. x 4 + a x 3 + b x 2 + c ... meaning we desire to see p 2 4 as the third term. Adding to the left side ...Missing: function | Show results with:function
  54. [54]
    Descartes' Method for Constructing Roots of Polynomials with ...
    So Descartes described a process for constructing the roots of any depressed quartic by intersecting the parabola y=x2 with a certain circle whose center and ...
  55. [55]
    [PDF] Mathematics and Its History, Third Edition - Index of /
    ... quartic equation is equivalent to one of the form x4 + px2 + qx + r = 0. 7.5 ... tan 4θ in terms of tan 2θ, and hence in terms of tanθ. 14.5.4 Letting ...
  56. [56]
    [PDF] 1. Lagrange's Solution to the Quartic
    In this lecture, we'll see how Lagrange made use of the theorem to give a solution to the general quartic equation. Lagrange's solution begins with the ...
  57. [57]
    [PDF] Computing the intersection of two quadrics through projection ... - arXiv
    Jun 25, 2019 · Abstract. This paper is devoted to presenting a new approach to determine the intersection of two quadrics based on the detailed analysis of ...
  58. [58]
    [PDF] Palindromic Polynomials - Dan Kalman Homepage
    In 1742, Euler factored the polynomial, showing Bernoulli to be wrong. substitution| thus obtaining a palindromic quartic.
  59. [59]
    symmetric quartic equation - PlanetMath
    Mar 22, 2013 · a0z4+a1z3+a2z2+a3z+a4=0(a0≠0), a 0 z 4 + a 1 z 3 + a 2 z 2 + a 3 z + a 4 = 0 ( a 0 ≠ 0 ) ,, (1). which can be reduced to quadratic equations ...
  60. [60]
    Solving the Quartic Equations by Newton's Method - ResearchGate
    In this note, I will describe an efficient way to solve a quartic equation a 4 λ 4 + a 3 λ 3 + a 2 λ 2 + a 1 λ + a 0 = 0 (1) using Newton's method.
  61. [61]
    Inverse power and Durand-Kerner iterations for univariate ...
    The Weierstrass (Durand-Kerner) approach and its variations as well as matrix methods based on the QR algorithm are among the most popular practical choices ...<|separator|>
  62. [62]
    [PDF] Polynomial Roots from Companion Matrix Eigenvalues 1 Introduction
    Jan 1, 1994 · Abstract. In classical linear algebra, the eigenvalues of a matrix are sometimes de ned as the roots of the char- acteristic polynomial.
  63. [63]
    Fast and Backward Stable Computation of Roots of Polynomials
    The roots are found by computing the eigenvalues of the associated companion matrix by Francis's implicitly shifted QR algorithm. A companion matrix is an upper ...