Bisection Root Bracketing
Bisection Root Bracketing
University numerical methods. Bisection Root Bracketing. Use a continuous real function on [a,b], with a<b and opposite endpoint signs, or an endpoint root. Enter a,b,f(a),f(b),N; N is an integer from 1 to 50, not a tolerance. Each round displays X=(a+b)/2; press EXE, then enter the current f(X) at Y?. The function is not stored automatically. All three exercises use Rad, essential for cos(x)−x.
- 235 bytes
- COMP mode
Worked examples
Example 1
Original exercise. University numerical methods. Data: f(x)=x²−2; [a,b]=[1,2]; f∈C[a,b]; N=4. Find: [a₄,b₄], m=(a₄+b₄)/2, h=(b₄−a₄)/2, |m−α|≤h.
Answer
Standard solution (without a calculator)
- Build the mathematical model and substitute the data.
- f(x)=x²−2; f∈C[1,2]
- f(1)=-1; f(2)=2; f(1)f(2)<0
- 1: c=1.5; f(c)=0.25; [a,b]=[1,1.5]
- 2: c=1.25; f(c)=-0.4375; [a,b]=[1.25,1.5]
- 3: c=1.375; f(c)=-0.109375; [a,b]=[1.375,1.5]
- 4: c=1.4375; f(c)=0.06640625; [a,b]=[1.375,1.4375]
- m=(1.375+1.4375)/2=1.40625
- h=(1.4375−1.375)/2=0.03125; |m−α|≤0.03125
- Combine the preceding results to obtain the answer.
Example 2
Original exercise. University numerical methods. Data: f(x)=x³−x−2; [a,b]=[1,2]; f∈C[a,b]; N=4. Find: [a₄,b₄], m=(a₄+b₄)/2, h=(b₄−a₄)/2, |m−α|≤h.
Answer
Standard solution (without a calculator)
- Build the mathematical model and substitute the data.
- f(x)=x³−x−2; f∈C[1,2]
- f(1)=-2; f(2)=4; f(1)f(2)<0
- 1: c=1.5; f(c)=-0.125; [a,b]=[1.5,2]
- 2: c=1.75; f(c)=1.609375; [a,b]=[1.5,1.75]
- 3: c=1.625; f(c)=0.666015625; [a,b]=[1.5,1.625]
- 4: c=1.5625; f(c)=0.2521972656; [a,b]=[1.5,1.5625]
- m=(1.5+1.5625)/2=1.53125
- h=(1.5625−1.5)/2=0.03125; |m−α|≤0.03125
- Combine the preceding results to obtain the answer.
Example 3
Original exercise. University numerical methods. Data: f(x)=cos(x)−x; [a,b]=[0,1]; Rad; f∈C[a,b]; N=4. Find: [a₄,b₄], m=(a₄+b₄)/2, h=(b₄−a₄)/2, |m−α|≤h.
Answer
Standard solution (without a calculator)
- Build the mathematical model and substitute the data.
- f(x)=cos(x)−x; f∈C[0,1]
- f(0)=1; f(1)=-0.4596976941; f(0)f(1)<0
- 1: c=0.5; f(c)=0.3775825619; [a,b]=[0.5,1]
- 2: c=0.75; f(c)=-0.01831113113; [a,b]=[0.5,0.75]
- 3: c=0.625; f(c)=0.1859631195; [a,b]=[0.625,0.75]
- 4: c=0.6875; f(c)=0.08533494615; [a,b]=[0.6875,0.75]
- m=(0.6875+0.75)/2=0.71875
- h=(0.75−0.6875)/2=0.03125; |m−α|≤0.03125
- Combine the preceding results to obtain the answer.
When to use it
University numerical methods. Bisection Root Bracketing. Use a continuous real function on [a,b], with a<b and opposite endpoint signs, or an endpoint root. Enter a,b,f(a),f(b),N; N is an integer from 1 to 50, not a tolerance. Each round displays X=(a+b)/2; press EXE, then enter the current f(X) at Y?. The function is not stored automatically. All three exercises use Rad, essential for cos(x)−x.
Input order
- A=a
- B=b
- C=f(a)
- D=f(b)
- M=N
- Y=f(X)
- Enter every required value again, including 0. Blank EXE retains memory; it is not zero. Used registers and Ans are overwritten.
Output order
- c₁,…,cₙ
- 0,aₙ,bₙ,m,h
- Record the value and press EXE.
- The program has ended. Press AC to leave; EXE runs it again.
Conditions and limitations
- a<b; f∈C[a,b]; f(a)f(b)<0 ∨ f(a)f(b)=0; N∈ℤ,1≤N≤50
- Y≠9E99
- Use a continuous real function on [a,b], with a<b and opposite endpoint signs, or an endpoint root. Enter a,b,f(a),f(b),N; N is an integer from 1 to 50, not a tolerance. Each round displays X=(a+b)/2; press EXE, then enter the current f(X) at Y?. The function is not stored automatically. All three exercises use Rad, essential for cos(x)−x.
- After the iterations read 0,a,b,midpoint,half-width. A zero function value gives a zero-width bracket. A single 1 means invalid bracket, signs or N; 2 means precision stagnation; 3 means blank Y or reserved 9E99. The half-width bounds root error only when continuity and correct function values hold. It does not prove uniqueness. Discontinuities and numerical cancellation can invalidate the sign test.
- Enter every required value again, including 0. Blank EXE retains memory; it is not zero. Used registers and Ans are overwritten.
- Finite precision can lose small differences between large numbers or probabilities extremely near 0 or 1. An error marker is never a valid answer.
Checking access…