How to find number of decimal places in Double

[sameer_thaj]

Hi,

Please help me how to find number of decimal places in a double value.

Example:
I have two double variables

double1 = 0.000000123450345454
double2 = 0.0000001234

i have to compare the above two doubles types and say that they are equal.

I have to omit the digits after 0.0000001234 in double1 variable.

Please help...


Thanks in advance

[jhen095]

multiply the number by 10,000,000,000 then parse it to an int,, it will cut off the decimal places,, then u can compare or turn to double and compare.

is the sneaky way,, but will work.

cheers

[sameer_thaj]

Thaks for your reply,

Using this solution i can compare only the above two variables.

I may get any number of decimals as i showed above. I need a generic solution which can satisfy any number of decimals...


Thanks

[dglienna]

Use the string length

[sameer_thaj]

When I convert the double to string, the contents is displayed in the Exponenet form. Using which i cannot find the length.

[Mutant_Fruit]

// Are the two doubles very nearly the same?
return Math.Abs(double1 - double2) < 0.0001;


Use that, or something similar.

[sameer_thaj]

Thanks for your support...


I got the alternate


I am using

str = double1.ToString("F 30");
str = str.TrimEnd('0');
int Length = str.Length;

[Mutant_Fruit]

Thats an extremely inefficient way of testing whether two doubles are equal, and also probably horribly inaccurate.

Console.WriteLine("Are these equal?: ", (0.2 + 0.2 + 0.2 + 0.2 + 0.2) != 1);

Funnily enough, this prints false. 0.2 can't be represented in binary exactly, therefore there's roundoff error. Your code would call these numbers not equal, because 0.9999999999 != 1.000000000. My code would call them equal.

Comparing floating point numbers for equality is the worst thing you can do, it's impossible to do. The best option is making sure they're about the same.

[laitinen]

As Mutant_Fruit said, comparing floating point numbers is difficult (at best).

Read this if you want to understand this very wide topic: What Every Computer Scientist Should Know About Floating-Point Arithmetic

Laitinen

[riscutiavlad]

.NET even provides an epsilon value to compare against: Double.Epsilon

[Mutant_Fruit]

Double.Epsilon is just the smallest value which can be represented in double precision. Adding 0.2 five times could easily result in a roundoff error greater than this. I wouldn't use it for comparisons.

[dglienna]

Thought he'd realize that I meant to use the string length to figure out how much to adjust the amount by.

[TheCPUWizard]

Double.Epsilon is just the smallest value which can be represented in double precision.

NO, Double.Epsilon is the smallest value which can be represented between 0.0 and 1.0. (ie it is 0.0 + 1 LSB of Mantissa)

This is an EXTREMELY important difference. As numbers get larger than 1.0 the increment between representable values becomes larger, and the potential for (absolute) error increases accordingly.

[Mutant_Fruit]

NO, Double.Epsilon is the smallest value which can be represented between 0.0 and 1.0. (ie it is 0.0 + 1 LSB of Mantissa)

This is an EXTREMELY important difference. As numbers get larger than 1.0 the increment between representable values becomes larger, and the potential for (absolute) error increases accordingly.

MSDN says:

Represents the smallest positive Double value greater than zero

I say:

Double.Epsilon is just the smallest value which can be represented in double precision

I don't see the difference here. I never said the Double.Epsilon is the increment between every floating point number.

[jhen095]

return Math.Abs(double1 - double2) < 0.0001;

I agree with this method also, its probably the easiest and is accurate to whatever value you put after the <

otherwise the method i suggested before with parsing it to an int combined with dglienna's string length idea (to get the shortest length of double, so you can trim the longer double to that length) would also work just fine but requires more code.

I would go with the first method!