O(1) convert fraction part of double into int (no precision penalty)?

Question

1.00/5 (1 vote)

See more:

I am completly struck with this.

I was trying to search the internet, but gets nothing.
Is it even possible?
Looks like i need some real programming guru here...

Example:
i have 123456.65432 and i want to have a int value 65432 from it
using standart modf i will get 123456 and 0.65432

how to convert double 0.65432 into 65432 int in constant time?

There should be some CPU intrinstic or bits or math magic to do that.

P. s. please don't tell me how to do that in linear or log. time...

What I have tried:

I can use modf to get the fractional part without precision penalty, but what to do then?

The half-solution is to get a fractional part using modf and then multiple it on the value, based on the fractional part's size (100 for 0.32, 1000 for 0.456 etc), BUT still this involves precision losses:
for example, on my machine 999999999.99999905 * 10000000000000000000 woule lead to the last 5 changed to 4 and so on

Posted 23-Feb-16 12:25pm

Shvetsov Evgeniy

Updated 23-Feb-16 14:30pm

v3

Add a Solution

Comments

Patrice T 23-Feb-16 18:42pm

Can you rephrase your question and give example of what you want.

Shvetsov Evgeniy 23-Feb-16 19:02pm

I added an example, but the question itself is pretty straight and clear already..

Patrice T 23-Feb-16 19:13pm

Way better.
Can you tell what you want to do with the value ?
Do you always want 5 figures ?

Shvetsov Evgeniy 23-Feb-16 19:33pm

Nope, the digits count is undetermined

nv3 23-Feb-16 19:14pm

Your question is not clear. What do you mean by: convert fraction part to int? Obviously you want to interpret the fractional part in decimal - as your example shows. As the fractional part is internally represented in binary format, the conversion to a decimal representation will deliver an infinitely long sequence of fractional digits (in many cases). Functions like printf generate a rounded representation with a certain number of digits. So, do you always want the int to be equal to the first N digits of the fractional parts?

The other question is what you mean by linear or log time? With regard to what? The number of digits? You only have one value to operate on.

Shvetsov Evgeniy 23-Feb-16 19:39pm

It was 'convert fractional part of a double to int' until edited. 123.321 had 123 as an integral part and 0.321 as a fractional part. I need to get int = 321, as example shows. The precision of any type is finit, so i need to convert it to an unsigned long long int without lossing precision and regardless of the digit's count (constant complexity).

2 solutions

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

User 59241 · Answer 1 · 2016-02-23T13:27:00

This subject comes up frequently. There is no silver bullet. nv3 has pointed out the gotchas in the comments. You need to appreciate how doubles are stored. The binary representation of the real number value is usually just a close approximation of the number. When the number stored as a double is displayed (or the fractional part "converted" to an integer) it may not be quite what you expected. What has time got to do with it?

How to get the integer and fractional part of a decimal number as two integers in Arduino or C? - Stack Overflow[^]

c - Getting the fractional part of a double value in integer without losing precision - Stack Overflow[^]

floating point - Extract fractional part of double *efficiently* in C - Stack Overflow[^]

Patrice T · Answer 2 · 2016-02-23T14:12:00

The obvious answer is multiply by 100000 and you get an integer.
Since this is not rocket science, I guess you want something else but didn't explain what.

Since extracting the fractional part of a value has usually no meaning without the integer part, I guess it is part of something bigger.
There is heaps of methods to transform a fractional part to an integer, but if we don't know why, we can't tell what method to use.
Remember that we are not in your head and untold obvious things are obvious only to you, not us.

[edit]
there is no CPU instruction to do what you want because it simply have no meaning.
You can't have an exact conversion because you think with a fractional value in base 10 and the CPU know the value in base 2. This mean rounding in any case.

Quote:
for example, on my machine 999999999.99999905 * 10000000000000000000 would lead to the last 5 changed to 4 and so on

because of what I said above.
There is no solution to this exazct problem, the way you ask it.

O(1) convert fraction part of double into int (no precision penalty)?

2 solutions

Solution 1

Solution 2

Add your solution here

Preview 0