Absolute relative approximate error

[chronicals]

I'm a bit confused, i have a question, it asks me to find ''the absolute relative approximate error'' at the end of each iteration. What's the formula of ''the absolute relative approximate error''?

 

[mrbohn1]

absolute means the size of the error (ie. the absolute value of the error), and relative means relative to the exact value. So the formula would be (absolute value of the error)/(exact value).

 

[D H]

The exact value is presumably unknown. Why go to the effort of finding an approximate solution if you already know the exact value? The exact value cannot be used to determine the relative error. An approximate value is available, so that is the appropriate thing to use in computing the scaled (or relative) error. Hence the term "absolute relative approximate error".

 

[chronicals]

I find lecture notes:

''the absolute relative approximate error'' = (( x_present - x_previous ) / x_present )*100