C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Computing between different types.          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message hongseok.yoon@gmail.com Guest Posted: Wed Sep 14, 2005 7:22 am    Post subject: Computing between different types. Please let mem know the reason why bellow problem's occured... float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. int i = 100; f = i * f; // At this point, f is 29 not 30! // If I cast above i to float type ( f = static_cast< float >( i ) * f ), // its result become 30 exactly as I expected. What is the problem? [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ulrich Eckhardt Guest Posted: Wed Sep 14, 2005 9:46 am    Post subject: Re: Computing between different types. [email]hongseok.yoon (AT) gmail (DOT) com[/email] wrote: Quote: float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. It might print 0.3, but it can't be 0.3 exactly. The reason is that floats are still binary numbers (unless you have one of those rare BCD encoded machines). So, you can try to assemble the decimal value 0.3 from the fractions of two, i.e. 1/2, 1/4, 1/8, 1/6 etc. However far you go, your result will either be slightly larger or smaller. Quote: int i = 100; f = i * f; // At this point, f is 29 not 30! I don't believe you. It might me slightly larger or smaller than 30, but not exactly 29. Quote: What is the problem? Floating point inaccuracy. Do not depend on absolute accuracy for floating point operations. Uli [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top kwikius Guest Posted: Wed Sep 14, 2005 3:48 pm    Post subject: Re: Computing between different types. [email]hongseok.yoon (AT) gmail (DOT) com[/email] wrote: Quote: Please let mem know the reason why bellow problem's occured... [cut] Quote: What is the problem? It would be helpful to know what compiler you are using, but (incredibly) there is no problem (unless the int is not being converted to a float before the calculation) . The C++ standard has no specification describing what the results of a float computation should be. The same is true for integers(except for the binary requirement) and strings and characters. results of operations on all these types come under the category 'implementation defined'. This is a result of the C heritage, where abstract concepts were subservient to hardware implementation requirements. The 'implementation defined' feature distinguishes C++ from later languages such as Java and C# where a primary goal is hardware independence and where the specifications on results of operations on inbuilt types are much more tightly controlled. Hopefully though C++ is making some progress with ints (by means of the int32 etc specs) and strings (e.g by design of a unicode library). Unfortunately I dont think there are any proposals yet regarding floating point specifications regards Andy Little [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Roberto Waltman Guest Posted: Thu Sep 15, 2005 1:09 am    Post subject: Re: Computing between different types. [email]hongseok.yoon (AT) gmail (DOT) com[/email] wrote: Quote: Please let mem know the reason why bellow problem's occured... ... // At this point, f is 29 not 30! ... // its result become 30 exactly as I expected. What is the problem? Your expectations. - Read Goldberg's paper "What Every Computer Scientist Should Know About Floating-Point Arithmetic" ( There is a copy here: http://galileo.phys.virginia.edu/classes/551.jvn.fall01/goldberg.pdf ) Roberto Waltman [ Please reply to the group, ] [ return address is invalid. ] [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Jim Langston Guest Posted: Thu Sep 15, 2005 1:10 am    Post subject: Re: Computing between different types. <hongseok.yoon (AT) gmail (DOT) com> wrote Quote: Please let mem know the reason why bellow problem's occured... float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. int i = 100; f = i * f; // At this point, f is 29 not 30! // If I cast above i to float type ( f = static_cast< float >( i ) * f ), // its result become 30 exactly as I expected. What is the problem? [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] I think the answer can best be summarized by this program: #include <iostream> int main() { float MyFloat = 1.3; int MyInt = MyFloat * 1000000; std::cout << MyInt << std::endl; char x; std::cin >> x; return 0; } Output in Microsoft Visual C++ .net 2003 is: 1299999 It seems that 1.3 can't be accurately described in binary. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Carlos Moreno Guest Posted: Thu Sep 15, 2005 1:12 am    Post subject: Re: Computing between different types. [email]hongseok.yoon (AT) gmail (DOT) com[/email] wrote: Quote: float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. int i = 100; f = i * f; // At this point, f is 29 not 30! If f is 29, then you are not showing us the program that you are running. With the code that you posted, f *can not* be 29 at this point. If you do i = i * f, then it is perfectly possible that you get a value of 29 in i -- the reason is that the result of the operation is something *very close* to 30, but not exactly 30. When converting to int (which is necessary if you assign the result *to i*), the rule says that you *truncate* (as opposed to round to the nearest integer). If the result of the operation happens to be 29.9999999999 (which is possible), then when assigned to an int it would be converted to 29. HTH, Carlos -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Martin Bonner Guest Posted: Thu Sep 15, 2005 9:15 am    Post subject: Re: Computing between different types. Carlos Moreno wrote: Quote: hongseok.yoon (AT) gmail (DOT) com wrote: float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. int i = 100; f = i * f; // At this point, f is 29 not 30! If f is 29, then you are not showing us the program that you are running. With the code that you posted, f *can not* be 29 at this point. Nit: Actually, f COULD be 29.0 at this point - but that would imply an incredibly awful quality-of-implementation for floating point. (I agree that the OP almost certainly meant that i was 29 after i = i * f) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top kanze Guest Posted: Thu Sep 15, 2005 10:13 am    Post subject: Re: Computing between different types. Ulrich Eckhardt wrote: Quote: hongseok.yoon (AT) gmail (DOT) com wrote: float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. It might print 0.3, but it can't be 0.3 exactly. The reason is that floats are still binary numbers (unless you have one of those rare BCD encoded machines). So, you can try to assemble the decimal value 0.3 from the fractions of two, i.e. 1/2, 1/4, 1/8, 1/6 etc. However far you go, your result will either be slightly larger or smaller. int i = 100; f = i * f; // At this point, f is 29 not 30! I don't believe you. It might me slightly larger or smaller than 30, but not exactly 29. What is the problem? Floating point inaccuracy. Do not depend on absolute accuracy for floating point operations. I suspect that Carlos Moreno hit the right answer, and he is in fact assigning it to an int. If the results are slightly smaller than 30, assigning them to an int will give 29. A more interesting aspect is that he reported that casting i to a float gave different results. According to the standard, that's exactly what the compiler should do without the cast, so there should be no difference. Of course, if he assigns the results to a float, and then outputs them with the default precision of 6, the results will probably be 30. Which is different to the results of assigning the value to an int. But there is also §5/10: "The values of the floating operands and the results of floating expressions may be represented in greater precision and range than that required by the type; the types are not changed thereby." It's also possible that on his machine, the compiler actually uses double (or even long double) for the calculations, and then rounds them to float. This could also explain slightly different results. -- James Kanze GABI Software Conseils en informatique orientée objet/ Beratung in objektorientierter Datenverarbeitung 9 place Sémard, 78210 St.-Cyr-l'École, France, +33 (0)1 30 23 00 34 [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top hongseok.yoon@gmail.com Guest Posted: Thu Sep 15, 2005 10:17 am    Post subject: Re: Computing between different types. I'm so sorry, code what I showed is not exactly same with I've tried. You are right. I've casted ( i * f ) to int in order to get positive number as a result. I really appreciate your kind explanation include all of other people. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Carlos Moreno Guest Posted: Thu Sep 15, 2005 1:56 pm    Post subject: Re: Computing between different types. Martin Bonner wrote: Quote: Carlos Moreno wrote: [email]hongseok.yoon (AT) gmail (DOT) com[/email] wrote: float f = 1.3; f -= 1.0; // At this point, f is 0.3 if I print using cout. int i = 100; f = i * f; // At this point, f is 29 not 30! If f is 29, then you are not showing us the program that you are running. With the code that you posted, f *can not* be 29 at this point. Nit: Actually, f COULD be 29.0 at this point - but that would imply an incredibly awful quality-of-implementation for floating point. Incidentally, I had initially written a tiny paragraph saying that if it really was the case, then a severely-broken compiler etc etc; but then, when re-reading the message before hitting SEND, I thought it was simply too severely broken to be likely, so I deleted it :-) Carlos -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Carlos Moreno Guest Posted: Thu Sep 15, 2005 1:57 pm    Post subject: Re: Computing between different types. kanze wrote: Quote: [...] I suspect that Carlos Moreno hit the right answer Guys... guys... Why don't you call me just Carlos? If it was someone that doesn't know me, ok -- but coming from someone I'm sure knows me (from my posts, that is), it sounds so formal, so strange that you use my firstname lastname... Not that it is offensive or anything remotely similar, it just sounds strange :-) It could be necessary, to disambiguate if there were two different Carlos's that wrote in this thread, but that's not often the case (it's not in this thread, at least) Cheers, Carlos Mor... errhhh, I mean, Carlos -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         C++Talk.NET Forum Index -> C++ Language (Moderated) All times are GMT Page 1 of 1   Jump to: Select a forum C/C++----------------C LanguageC Language (Moderated)C++ Language (Moderated)C++ language (comp.lang.c++)C++ language, library and standards Non-English C/C++ Newsgroups----------------C++ (German)C++ (French)  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.