This study discusses how to find the best linear approximation y=mx+b to a fundamental function y=sqrt(x) on the interval [0,b], especially using the minimax error in Numerical Analysis. For this aim we employ two mathematical techniques: a) using the MATLAB code, positioning m and n values of the smallest maximum error on a broad range of m, and n value matrix in a rough scale and then repeatedly refining the regions in the smaller scales and b) Finding three-point fitting line to a set of non-colinear three points. We see that both results are successfully obtained and identical each other neglecting the error tolerance.


This research project is proceeded with a sponsor of "NSF-TIP grant (171944- PI: Atean Agegnehu): Targeted Infusion Project: Developing a Minor in Applied Mathematics at Savannah State University"