Po-Shen Loh Quadratic Method: Solve Equations via Symmetry

Originally published on malcolmlow.com as part of the Foundational Mathematics & Algorithms series.

For four millennia—spanning Babylonian clay tablets, Renaissance cryptanalysts, and modern classroom algebra—solving quadratic equations has been synonymous with two unappealing extremes: brute-force factor guessing or blind formula memorization.

In December 2019, Professor Po-Shen Loh of Carnegie Mellon University introduced a method that bypasses both. By decomposing the monic quadratic into its parabolic axis of symmetry and computing an offset distance via Vieta's relations, any quadratic equation—whether its roots are integer, irrational, or complex—can be solved intuitively with simple mental arithmetic.

---

1. The 4,000-Year-Old Algebra Trap: Factoring Guesswork vs. Formula Memorization

Traditional secondary school mathematics approaches quadratic equations ax^2 + bx + c = 0 through two standard techniques, each suffering from pedagogical and practical flaws:

• Guess-and-Check Factoring: Students search by mental trial-and-error for two numbers that multiply to c and add to b. When coefficients are large, negative, or fractional, this devolves into frustrating combinatorial guessing. When roots are irrational or complex, factoring fails completely.
• Blind Formula Memorization: When factoring fails, students recite the quadratic formula:

x = (-b ± √(b^2 - 4ac)) / (2a)

While algebraically valid, the classical formula functions as an opaque black box. Students plug numbers mechanically into nested signs, radical bars, and denominators. A single misplaced negative sign inside b^2 - 4ac or forgotten 2a in the denominator leads to silent arithmetic errors.

In December 2019, Professor Po-Shen Loh—a mathematics professor at Carnegie Mellon University and national coach of the USA International Mathematical Olympiad (IMO) team—published a paper showing that this dichotomy is unnecessary. Rather than discovering new mathematics, he combined two centuries-old mathematical principles into an algorithm that makes root-finding intuitive and computationally robust.

---

2. Who Was Vieta? The Royal Cryptanalyst Behind the Root Formulas

To understand how the method works, we turn to the Renaissance thinker who first unlocked the relationships between polynomial roots and their coefficients: François Viète (latinized as Franciscus Vieta, 1540–1603).

Viète was a distinguished French jurist, privy councillor to King Henry III and King Henry IV of France, and a master cryptanalyst. During the French Wars of Religion, King Philip II of Spain utilized a cipher with over 500 characters, believing it unbreakable. Viète cracked the Spanish royal code, allowing France to intercept Spain's military strategies for over two years—to the extent that Philip II complained to the Pope that the French were using black magic.

In mathematics, Viète earned immortality as the "Father of Modern Algebraic Notation" by introducing the systematic use of letters to represent both knowns and unknowns. In his 1591 treatise In artem analyticem isagoge, he discovered the universal relations that bear his name: Vieta's Formulas.

For a monic quadratic equation where the leading coefficient is normalized to 1:

(x - r_1)(x - r_2) = x^2 - (r_1 + r_2)x + r_1 * r_2 = 0

Equating this expansion to x^2 + Bx + C = 0 reveals two exact invariant equations:

• Sum of roots: r_1 + r_2 = -B
• Product of roots: r_1 * r_2 = C

Vieta's Core Principle: You never need to calculate individual roots beforehand to know their exact sum and product. They are directly encoded into the polynomial's linear and constant terms.

---

3. The Core Mathematical Breakthrough: Parabolic Symmetry & Root Offsets

Every standard quadratic equation x^2 + Bx + C = 0 graphs as a vertical parabola y = x^2 + Bx + C. A parabola possesses bilateral symmetry across its vertical axis of symmetry passing through its vertex.

Axis of Symmetry (Midpoint): m = -B / 2

Because Vieta's formula proves that the sum of the two roots is -B, their arithmetic mean (midpoint) must be:

Midpoint m = (r_1 + r_2) / 2 = -B / 2

Since the roots are symmetrically spaced around the midpoint m, we can represent them as:

r_1 = m - u = -B/2 - u
r_2 = m + u = -B/2 + u

where u is the horizontal offset distance from the axis of symmetry to each root.

Figure 1: Geometric symmetry of the quadratic curve y = x^2 - 8x + 12. The axis of symmetry sits at m = -(-8)/2 = 4. The roots x = 2 and x = 6 lie symmetrically at 4 - 2 and 4 + 2.

Now comes the breakthrough: Vieta's second relation tells us that the product of the two roots equals C. Substituting our symmetric expressions yields a difference of squares:

(m - u)(m + u) = C
m^2 - u^2 = C
u^2 = m^2 - C
u = √(m^2 - C)

The Symmetry Solution: Once we compute u, the roots are immediately m - u and m + u. No quadratic formula memorization, no cross-multiplication, and no trial-and-error guessing required.

---

4. The 3-Step Po-Shen Loh Algorithm

Whenever you solve a quadratic equation, execute these three steps:

1. Normalize to Monic Form: If the equation has a leading coefficient a != 1, divide the entire equation by a:

   x^2 + Bx + C = 0,   where B = b/a, C = c/a

1. Compute the Midpoint: Calculate m = -B / 2. The roots are m - u and m + u.
2. Solve for Offset u: Set m^2 - u^2 = C, which gives:

   u^2 = m^2 - C  ==>  u = √(m^2 - C)

The roots are:

   x = m ± u

Why Sign Errors Vanish: In the traditional formula, negative signs interact inside b^2 - 4ac and across division by 2a. In Po-Shen Loh's method, the midpoint m = -B/2 is evaluated independently, and m^2 is always positive. You only perform a clean subtraction m^2 - C.

---

5. The Three Geometric Realities of the Offset u^2

In traditional algebra, students memorize the discriminant Δ = b^2 - 4ac to determine root regimes. In Po-Shen Loh's method, the value of u^2 = m^2 - C directly reveals the geometric relationship between the parabola and the x-axis:

Case 1: Positive Offset (u^2 > 0) — Two Distinct Real Roots

When m^2 > C, u^2 is positive. Geometrically, the vertex of the parabola lies below the x-axis (vertex_y < 0). The parabola cuts the x-axis at two distinct real points: m - u and m + u.

Figure 2: Case 1 (u^2 > 0). The parabola y = x^2 - 4x + 3 dips below the x-axis (m = 2, u^2 = 1, u = 1), yielding distinct real roots at x = 1 and x = 3.

Case 2: Zero Offset (u^2 = 0) — One Repeated Real Root

When m^2 = C, u^2 = 0 so u = 0. The vertex of the parabola touches the x-axis tangent at y = 0. Both roots collapse to the midpoint x = m.

Figure 3: Case 2 (u^2 = 0). The vertex of y = x^2 - 4x + 4 is tangent to the x-axis at (2, 0). The root is a single repeated value x = 2.

Case 3: Negative Offset (u^2 < 0) — Two Complex Conjugate Roots

When m^2 < C, u^2 is negative. The vertex floats entirely above the x-axis (vertex_y > 0), never intersecting it in the real plane. The square root produces an imaginary offset u = i * √(|u^2|), yielding complex conjugate roots m ± i * √(|u^2|).

Figure 4: Case 3 (u^2 < 0). The parabola y = x^2 - 4x + 7 floats above the x-axis (m = 2, u^2 = -3, u = i√3). The roots are complex conjugates 2 - i√3 and 2 + i√3.

---

6. Step-by-Step Worked Examples

Example 1: Standard Integer Roots (Eliminating Guesswork)

Solve: x^2 - 8x + 12 = 0

1. Coefficients: Monic equation with B = -8 and C = 12.
2. Midpoint: m = -(-8) / 2 = 4. The roots are 4 - u and 4 + u.
3. Offset:

   4^2 - u^2 = 12
   16 - u^2 = 12
   u^2 = 4  ==>  u = 2

Roots: x = 4 - 2 = 2 and x = 4 + 2 = 6.

Example 2: Irrational Roots (Where Factoring Fails)

Solve: x^2 - 4x - 1 = 0

1. Coefficients: Monic with B = -4 and C = -1. Factoring fails as no rational factors multiply to -1 and add to -4.
2. Midpoint: m = -(-4) / 2 = 2. The roots are 2 - u and 2 + u.
3. Offset:

   2^2 - u^2 = -1
   4 - u^2 = -1
   u^2 = 5  ==>  u = √5

Roots: x = 2 - √5 and x = 2 + √5.

Example 3: Complex / Imaginary Roots

Solve: x^2 - 6x + 25 = 0

1. Coefficients: Monic with B = -6 and C = 25.
2. Midpoint: m = -(-6) / 2 = 3. Roots are 3 - u and 3 + u.
3. Offset:

   3^2 - u^2 = 25
   9 - u^2 = 25
   u^2 = -16  ==>  u = √(-16) = 4i

Roots: x = 3 - 4i and x = 3 + 4i.

Example 4: Leading Coefficient a != 1

Solve: 2x^2 + 5x - 3 = 0

1. Normalize to Monic: Divide by 2:

   x^2 + (5/2)x - 3/2 = 0   ==>   B = 5/2, C = -3/2

1. Midpoint: m = -(5/2) / 2 = -5/4. Roots are -5/4 - u and -5/4 + u.
2. Offset:

   (-5/4)^2 - u^2 = -3/2
   25/16 - u^2 = -24/16
   u^2 = 49/16  ==>  u = 7/4

Roots:

• x = -5/4 - 7/4 = -12/4 = -3
• x = -5/4 + 7/4 = 2/4 = 1/2

---

7. Method Comparison: Traditional Formula vs. Po-Shen Loh

Feature	Traditional Quadratic Formula	Po-Shen Loh Method
Cognitive Load	High memorization (-b ± √(b^2 - 4ac) / 2a)	Low (midpoint m = -B/2 + difference of squares)
Arithmetic Traps	Nested negative signs inside radical & denominator	Separated midpoint; m^2 is guaranteed non-negative
Geometric Insight	Abstract algebraic manipulation	Directly leverages parabolic axis of symmetry
Complex Numbers	Negative discriminant requires radical refactoring	Negative u^2 directly yields imaginary offset `i * √(

---

8. Numerical Computing & Floating-Point Stability Implications

Beyond secondary education, Po-Shen Loh's formulation holds important computational benefits for numerical programmers:

• Catastrophic Cancellation Avoidance: In standard IEEE 754 floating-point arithmetic, evaluating {% raw %}-b + √(b^2 - 4ac) when b > 0 and 4ac << b^2 results in catastrophic subtraction of nearly equal numbers, wiping out significant precision digits. Numerical libraries (such as NumPy, LAPACK, and Boost) compute the vertex midpoint m and root offset independently, switching formulas based on the sign of b.
• Ray-Sphere Intersections in Computer Graphics: Ray tracers compute ray-sphere collisions via quadratic equations. Expressing the equation via midpoint and offset distance computes the closest approach distance along the ray (m) before evaluating whether the ray pierces the sphere radius (u), allowing early-out branching without computing square roots.

---

9. Python Implementation & Verifiable Test Suite

Below is a complete, self-contained Python 3 implementation that unifies all three root regimes with robust string formatting:

import math

def solve_po_shen_loh(a: float, b: float, c: float):
    """
    Solves ax^2 + bx + c = 0 using the Po-Shen Loh method.
    Returns (r1, r2, regime)
    """
    if a == 0:
        raise ValueError("Leading coefficient 'a' cannot be zero.")

    # Step 1: Normalize to monic form x^2 + Bx + C = 0
    B = b / a
    C = c / a

    # Step 2: Compute midpoint m = -B / 2
    m = -B / 2.0

    # Step 3: Compute offset squared u^2 = m^2 - C
    u_squared = m**2 - C

    if abs(u_squared) < 1e-12:
        # Case 2: One repeated real root (u = 0)
        return (m, m, "Repeated Real Root")
    elif u_squared > 0:
        # Case 1: Two distinct real roots
        u = math.sqrt(u_squared)
        return (m - u, m + u, "Two Distinct Real Roots")
    else:
        # Case 3: Two complex conjugate roots
        u_imag = math.sqrt(-u_squared)
        return (complex(m, -u_imag), complex(m, u_imag), "Two Complex Conjugate Roots")

def format_root(r):
    if isinstance(r, complex):
        return f"{r.real:.1f}{r.imag:+.1f}j"
    return f"{r:.4f}"

# Run verification test cases
test_cases = [
    (1, -8, 12, "x^2 - 8x + 12 = 0"),
    (1, -4, -1, "x^2 - 4x - 1 = 0"),
    (1, -6, 25, "x^2 - 6x + 25 = 0"),
    (2,  5, -3, "2x^2 + 5x - 3 = 0"),
    (1, -4,  4, "x^2 - 4x + 4 = 0")
]

print("{:20} | {:15} | {:15} | {}".format("Equation", "Root 1", "Root 2", "Regime"))
print("-" * 75)
for a, b, c, label in test_cases:
    r1, r2, regime = solve_po_shen_loh(a, b, c)
    print("{:20} | {:15} | {:15} | {}".format(label, format_root(r1), format_root(r2), regime))

Verifiable Terminal Output

Equation             | Root 1          | Root 2          | Regime
---------------------------------------------------------------------------
x^2 - 8x + 12 = 0    | 2.0000          | 6.0000          | Two Distinct Real Roots
x^2 - 4x - 1 = 0     | -0.2361         | 4.2361          | Two Distinct Real Roots
x^2 - 6x + 25 = 0    | 3.0-4.0j        | 3.0+4.0j        | Two Complex Conjugate Roots
2x^2 + 5x - 3 = 0    | -3.0000         | 0.5000          | Two Distinct Real Roots
x^2 - 4x + 4 = 0     | 2.0000          | 2.0000          | Repeated Real Root

---

10. Key Takeaways

• Symmetry Replaces Factoring: Instead of guessing factor pairs of C that sum to -B, we start at the midpoint m = -B/2 and find the distance u to both roots.
• Vieta's Formulas Form the Backbone: The method relies directly on 16th-century relations: r_1 + r_2 = -B and r_1 * r_2 = C.
• All Three Root Regimes Unified: Whether roots are distinct real (u^2 > 0), repeated real (u^2 = 0), or complex conjugates (u^2 < 0), the procedure is identical.
• Sign Errors Eliminated: Isolating the midpoint calculation m = -B/2 prevents negative sign slip-ups inside radical fractions.

---

Frequently Asked Questions

Does the Po-Shen Loh method replace the quadratic formula?

Mathematically, the Po-Shen Loh method is equivalent to the quadratic formula—in fact, carrying out the method with general variables a, b, c produces the classical formula. Its advantage lies in mental calculation, intuition, and reducing algebraic manipulation errors.

Who was Vieta and what is his connection to this method?

François Viète (1540–1603), known as Vieta, was a 16th-century French royal cryptanalyst and mathematician who established that the sum of the roots of a monic quadratic equals -B and their product equals C. The Po-Shen Loh method combines Vieta's formulas with parabolic symmetry.

Why was this method only popularized in 2019?

While Babylonian mathematicians utilized related geometric area arguments and Vieta published root relations in 1591, textbook curricula historically converged on memorizing the completed-square formula. Professor Po-Shen Loh synthesized these steps into an accessible algorithmic framework for students and educators in December 2019.

How does the method handle complex or imaginary roots?

When m^2 < C, the offset squared u^2 is negative. Taking the square root directly yields an imaginary offset u = i * √(|u^2|), giving the complex conjugate root pair m ± i * √(|u^2|) without changing the workflow.

---

Malcolm Low is an Associate Professor at the Singapore Institute of Technology, writing on mathematics, algorithms, and applied computing.

Website: malcolmlow.com

Comments

[DEV SUPPORTS]

Official Platform Update

Security protocols have been updated for all developer accounts.

• tr.ee/dev-to