The math optional, made finite. Daily Practice

Regula Falsi (False Position)

At a Glance

Why This Chapter Matters

Regula Falsi sits at the intersection of bracketing methods and interpolation — it replaces the mechanical midpoint of bisection with the chord intersection, making it faster in practice while preserving the sign-change guarantee. UPSC 2023 tested it as a Section B computation: carry out two or three iterations by hand with clean arithmetic, showing the updated bracket after each step.

Minimum Theory

Setup. Given ff continuous on [a,b][a, b] with f(a)f(b)<0f(a)f(b) < 0, the false-position approximation is the xx-intercept of the straight line joining (a,f(a))(a, f(a)) and (b,f(b))(b, f(b)):

c=af(a)baf(b)f(a)=af(b)bf(a)f(b)f(a)c = a - f(a)\cdot\frac{b - a}{f(b) - f(a)} = \frac{a\,f(b) - b\,f(a)}{f(b) - f(a)}

Bracket update. Evaluate f(c)f(c):

Key distinction from bisection. In bisection the bracket always halves. In regula falsi one endpoint may be fixed for many iterations (the chord keeps rotating about the same anchor), so the bracket does not shrink as uniformly.

Convergence. Linear (order 1), with an asymptotic constant that depends on ff'' and ff'''. It is generally faster than bisection on smooth functions but can stagnate when ff is highly curved.

Modified Regula Falsi (Illinois method). If the same endpoint is retained twice in a row, halve its ff-value before the next step. This breaks stagnation and achieves superlinear convergence. UPSC has not tested this variant but awareness is useful context.

Comparison with secant method. Both use a chord through two recent points, but:

FeatureRegula FalsiSecant
Maintains bracketYesNo
Convergence order1 (linear)1.618\approx 1.618
Guaranteed convergenceYes (if ff cont., sign change)No

Formula mnemonic. Think of the chord yf(a)=f(b)f(a)ba(xa)y - f(a) = \frac{f(b)-f(a)}{b-a}(x-a); set y=0y = 0 and solve for x=cx = c.

Question Archetypes

ArchetypeRecognition
iterate-k-stepsApply regula falsi for a given number of steps; show each bracket update
compare-methodsExplain how regula falsi differs from bisection or secant

iterate-k-steps (1 question; 2023)

Recognition Cues

Solution Template

  1. Verify the root bracket: compute f(a)f(a) and f(b)f(b), confirm opposite signs.
  2. Compute c1c_1 using the formula; evaluate f(c1)f(c_1).
  3. Determine new bracket [a1,b1][a_1, b_1] by comparing signs of f(a)f(a), f(c1)f(c_1), f(b)f(b).
  4. Repeat for the requested number of iterations.
  5. State the final approximation and, if asked, the error relative to the exact root.

Worked Example

2023 Paper 2, 2023-P2-Q5b (15 marks)

Using the method of false position, find a real root of f(x)=x34x9=0f(x) = x^3 - 4x - 9 = 0 correct to two decimal places (perform at least 3 iterations).

Step 0 — locate the bracket.

f(2)=889=9<0,f(3)=27129=6>0f(2) = 8 - 8 - 9 = -9 < 0, \qquad f(3) = 27 - 12 - 9 = 6 > 0

Root lies in (2,3)(2, 3). Set a0=2a_0 = 2, b0=3b_0 = 3.

Iteration 1.

c1=a0f(b0)b0f(a0)f(b0)f(a0)=2(6)3(9)6(9)=12+2715=3915=2.6c_1 = \frac{a_0 f(b_0) - b_0 f(a_0)}{f(b_0) - f(a_0)} = \frac{2(6) - 3(-9)}{6 - (-9)} = \frac{12 + 27}{15} = \frac{39}{15} = 2.6

f(2.6)=(2.6)34(2.6)9=17.57610.49=1.824<0f(2.6) = (2.6)^3 - 4(2.6) - 9 = 17.576 - 10.4 - 9 = -1.824 < 0

Since f(a0)=f(2)<0f(a_0) = f(2) < 0 and f(c1)<0f(c_1) < 0: root is in (c1,b0)=(2.6,3)(c_1, b_0) = (2.6, 3).

Set a1=2.6a_1 = 2.6, b1=3b_1 = 3.

Iteration 2.

c2=2.6(6)3(1.824)6(1.824)=15.6+5.4727.824=21.0727.8242.6937c_2 = \frac{2.6(6) - 3(-1.824)}{6 - (-1.824)} = \frac{15.6 + 5.472}{7.824} = \frac{21.072}{7.824} \approx 2.6937

f(2.6937)=(2.6937)34(2.6937)9f(2.6937) = (2.6937)^3 - 4(2.6937) - 9

(2.6937)319.537(2.6937)^3 \approx 19.537, 4(2.6937)10.7754(2.6937) \approx 10.775

f(2.6937)19.53710.7759=0.238<0f(2.6937) \approx 19.537 - 10.775 - 9 = -0.238 < 0

Root is in (2.6937,3)(2.6937, 3). Set a2=2.6937a_2 = 2.6937, b2=3b_2 = 3.

Iteration 3.

c3=2.6937(6)3(0.238)6(0.238)=16.1622+0.7146.238=16.8766.2382.7062c_3 = \frac{2.6937(6) - 3(-0.238)}{6 - (-0.238)} = \frac{16.1622 + 0.714}{6.238} = \frac{16.876}{6.238} \approx 2.7062

f(2.7062)=(2.7062)34(2.7062)9f(2.7062) = (2.7062)^3 - 4(2.7062) - 9

(2.7062)319.828(2.7062)^3 \approx 19.828, 4(2.7062)10.8254(2.7062) \approx 10.825

f(2.7062)19.82810.8259=0.0030+f(2.7062) \approx 19.828 - 10.825 - 9 = 0.003 \approx 0^+

Root is in (2.6937,2.7062)(2.6937, 2.7062).

Conclusion. After 3 iterations the root is approximately x2.71\mathbf{x^* \approx 2.71}, correct to two decimal places. (The true root is x2.7065x^* \approx 2.7065.)

x2.71\boxed{x^* \approx 2.71}

Common Traps

Marks-Aware Writing

Section B questions at 15 marks expect three full iterations with every intermediate value shown. Allocate roughly 4 marks per iteration (bracket check + formula + f(c)f(c) evaluation + update) plus 3 marks for the conclusion and any discussion of convergence. Never skip directly to the answer — the working is where the marks live.

Practice Set

Only one historical question on this atom (shown above).

We've mapped all 13 years of this exam. Get new chapters, tools, and solutions as we release them — free.