Question

What causes ‘round off’ error in numerical solution?

(a) Finite grid points

(b) Rounding number to a significant figure

(c) Solving nonlinear equation

(d) Incoherent boundary conditions

I had been asked this question during an internship interview.

I'm obligated to ask this question of Philosophy of Method of Characteristics topic in chapter Numerical Techniques for Steady Supersonic Flow of Aerodynamics

Answer 1

The correct answer is (b) Rounding number to a significant figure

To explain I would say: Numerical solutions are obtained using the computers which often rounds off every number to a certain significant digit. Each of these rounding off results in small errors. If we decrease the truncation errors, the round – off error increases and these errors is a function of step size ∆x.