In false positions, instead of always using the midpoint of [a,b] as the iteration point, you use the x-intercept of the line between (a,f(a)) and (b,f(b)). That is, given (a,f(a)) and (b,f(b)), compute the equation of the line between these points; . Set y=0 and solve for x to find the x-intercept; call this intercept c. So . Now evaluate f(c) to determine whether it is positive or negative, and keep either a or b depending on the sign of f(c), just like in bisection.
To change your bisection code into a false positions code, copy bisect.f into another file, say falsa.f. Change the program name and comment line to reflect the change to false positions. The only line you need to change in the body of the code is the definition of y. Instead of y=(a+b)/2, you need to calculate the x-intercept, called c in the previous paragraph.
Run both bisection and false positions to find . False position starts out slower than bisection - after the first two steps, the interval in which the root lies is smaller under bisection. But once false positions gets close to the root, it zips in quickly. So the method of false position shares with bisection the advantage of trapping a root and, therefore, always converging. But false position has a decided advantage in the speed with which it converges.
We next turn to another algorithm, Newton's method for finding roots. Newton's method does not always converge. But, when it does converge, it is very fast (at least once it is ``close enough'' to the root). Newton's method also has the advantage of not requiring a bracketing interval with positive and negative values (so Newton's method will allow us to solve , whereas bisection and false positions will not).