CODEWITHSUNDEEP
cdecimalfloating-accuracy
Marievi
Marievi

Reputation: 5011

Compare small values with zero with specific accuracy

I have a program which prints out a matrix with very small values. An example of my matrix is

0.00000000000000004  0.12300000000000000

0.00000000011111114  0.00000000000038544

What I would like to do is compare each value with zero and accept it to be zero with specific accuracy, which is 9 decimal places. In other words, if a number has 9 zeros as its first decimal values, I want to consider it as a zero, otherwise not.

I have searched a lot but really found nothing about it. Any ideas?

Upvotes: 0

Views: 91

Answers (2)

nalzok
nalzok

Reputation: 16107

My idea is similar to what @MicroVirus has mentioned in his comment. You can simply compare it to a specific number:

if(num < 1E-9 && num > -1E-9)
    num = 0;

or

if(num * 1E9 < 1 && num * 1E9 > -1)
    num = 0;

Upvotes: 2

MicroVirus
MicroVirus

Reputation: 5477

As I noted in my comment, you can simply compare the float f via -1e-9 < f < 1e-9.

The you need both the positive and negative boundary to ensure it works properly for positive and negative numbers. You use 1e-9 and not 1e-10 because if a number is smaller than 1e-9, a number with 8 decimals 0, then it has 9 decimals or more zero.

Do note that due to float rounding of 1e-9 when going from decimal to binary, you might see some rounding error.

Upvotes: 5

Related Questions