Re: [Help-gsl] Re: gsl c++ wrapper ?
Hi! I'd like to use gslWrap - downloaded it from SourceForge - , but it only supplies an installation procedure for Linux / Unix. Since I'm using Windows, I'm a bit lost. For one thing, I can't find an installation exe-file in the zip downloaded at SourceForge. Secondly, (assuming that the files might just need to be copied), the install help tells to fill in info for the following lines in the GSLWrap Makefile, which I cannot find in any of the makefile files: GSLLIBDIR = GSLINCDIR = (yet, looking through the files, it looks like some installation procedure must be run through) Thanks a lot in advance, cheers Carola -- This message was sent on behalf of c.smuld...@gmail.com at openSubscriber.com http://www.opensubscriber.com/message/help-gsl@gnu.org/12617065.html ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
On Sat, Sep 12, 2009 at 3:19 AM, Leo Razoumov slonik...@gmail.com wrote: On 9/10/09, Rohit Garg rpg@gmail.com wrote: [..snip..] But that still begs the question. Why would you want to use GSL's linear algebra subset? Here are few: (1) Simplicity of GSL design (2) No expression templates !! I do not agree with you here, but if you like GSL's approach better here, then good for you. (3) support for BLAS with vectorized optimizations (SSE2 and later) I was under the impression that gsl's blas was fairly generic and you had to use ATLAS and the like to get good performance and vectorization. Or are you referring to gsl compiled with auto-vectorization? (4) total manual control of resources I certainly have less experience than you, and I _demand_ total manual control over my code when I am optimizing it. But since eigen is a purely template library, I think it offers you more choices. BTW, I mostly link with fedora's gsl packages, so our POV may be different. How do you use gsl ( in terms of compiling/compile options/linking etc. ) ? I do not want to derogate your project. It could well be very good, indeed. But my 15+ years of large scale numerical simulations taught me few lessons. I would pick a simple, clean library with manual resource management numeric and flat debugging anytime over a smart library like Boost or eigen for that matter. Agreed, overall choice depends on many factors. smartness is one of them. more crucial. And when you optimize algorithms you want to be very close to your hardware and have full manual control. In numerics I see people going from C to asm more often than to C++. How about this for a manual control over optimization http://listengine.tuxfamily.org/lists.tuxfamily.org/eigen/2009/01/msg00234.html The product would have been auto vectorized if the size was a multiple of 2. (3 in this case). Just my 1 cent. :) PS: GSL's non linear algebra subset is great, and I am really thankful to the devs for this. -- Rohit Garg http://rpg-314.blogspot.com/ Senior Undergraduate Department of Physics Indian Institute of Technology Bombay ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
Frankly, you have slowly drifted off the subject since your previous post. We are talking about how to wrap GSL and make GSL work better with C++, rather than do advertisement for yet another C++ vector/matrix library. For the original question, my personal opinion on this subject is that just build the program in C++ with some favorite style and narrow the interface between the C++ part and GSL facilities. By contrast, wrapping GSL is not an easy choice because GSL utilizes so many classic C style coding tricks, e.g., void* pointer conversion etc. Specifically, for vector/matrix support library, a third part one out of GSL can be used, because most parts of GSL facilities can work well with built-in array (which are widely supported in other vector/matrix libraries) and do not rely on specific GSL vector/matrix interfaces. On Thu, Sep 10, 2009 at 1:25 PM, Rohit Garg rpg@gmail.com wrote: Well, you can of course use ublas and the like, but after looking at these benchmarks, http://eigen.tuxfamily.org/index.php?title=Benchmark-August2008 I wouldn't touch them with a bargepole if I can help it. As far as C++ey API goes, may be this http://eigen.tuxfamily.org/index.php?title=API_Showcase will convince to think twice, if not more. -- HZ ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
On Thu, Sep 10, 2009 at 1:17 PM, Hongzheng Wang wan...@gmail.com wrote: Frankly, you have slowly drifted off the subject since your previous post. With the benefit of hindsight, perhaps yes. My intention is not to troll at all (even for things that may or may not be better), so please bear with me. For the original question, my personal opinion on this subject is that just build the program in C++ with some favorite style and narrow the interface between the C++ part and GSL facilities. By contrast, wrapping GSL is not an easy choice because GSL utilizes so many classic C style coding tricks, e.g., void* pointer conversion etc. Specifically, for vector/matrix support library, a third part one out of GSL can be used, because most parts of GSL facilities can work well with built-in array (which are widely supported in other vector/matrix libraries) and do not rely on specific GSL vector/matrix interfaces. But that still begs the question. Why would you want to use GSL's linear algebra subset? GSL's comprehensive coverage makes it a valuable tool for many tasks. I for one, mix it's histogram routines with some operations from eigen trivially. Considering that there is some community interest in writing C++ish wrappers to GSL, I want to know what motivates such need? I mean why invest effort here, when it is possible to take something that, atleast appears to be, is better? Is there a particular reason/benefit to investing effort in writing GSL's C++ bindings over simply using eigen? I might be missing something here, but is the interest in writing c++ wrappers is for just the linear algebra subset or for the whole GSL? Full GSL c++ wrappers would be a good idea, I agree. I am not so enthused about wrappers for just the linear algebra subset. -- Rohit Garg http://rpg-314.blogspot.com/ Senior Undergraduate Department of Physics Indian Institute of Technology Bombay ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
I think using eigen over gsl (for linear algebra) is a no brainer. a) gsl is dog slow compared to eigen (no vectorization whatsoever). b) gsl has no float option c) eigen offers both column major and row major storage d) eigen *does* expression templates e) eigen will happily map your data and perform operations on it. f) eigen has way cooler (aka C++-ey) api over gsl/lapack g) eigen is adding support for sparse matrices at a rapid pace. For (nearly full, I think) lapack to C++ bindings, FLENS is a great idea too. But I have never used flens. http://eigen.tuxfamily.org/index.php?title=Main_Page I would recommend using eigen/FLENS over gsl anyday for linear algebra purposes. BTW, it also does small vectors/matrices. :) DISCLAIMER: I have contributed and continue to contribute to eigen. On Wed, Sep 2, 2009 at 1:38 AM, Rodney Sparapanirspar...@mcw.edu wrote: John D Lamb wrote: I’ve written some wrappers for GSL matrices and vectors and might can help if someone has a suitable project. While Rodney is right that there are some issues with creating a practical OO interface, C++ is not the same as OO and OO is not the same as operator overloading (assuming OO means object orientation). I was talking about both OO and operator overloading. Another approach to wrapping GSL functions for C++ is to use shared pointers, namespaces and exceptions. Typically I might create a class gsl::vector that is an extended shared pointer and whose member functions look much like the original functions. The main benefit is not having to deallocate memory explicitly and only once. Well, it seems like there is plenty of interest. We should all work together and create an add-on. We can call it GSL++. I'm in the process of working on my PhD which will mean that I'll have an ample opportunity to use it. However, I strongly recommend that the test-bed be integral rather than an after-thought. For each feature, there should be a corresponding check. I don't really understand dejaGNU/etc. so I can't contribute much to this aspect. Rodney ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl -- Rohit Garg http://rpg-314.blogspot.com/ Senior Undergraduate Department of Physics Indian Institute of Technology Bombay ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
If I really need a C++ style vector/matrix library, I would refer to boost uBlas + bindings. But for daily normal numerical work I encountered, gsl is easily used and better because vector/matrix support is not the whole world; we usually need more numerical computation facilities beyond basic vector/matrix support and gsl is such a quite complete solution. On Sat, Sep 5, 2009 at 10:26 PM, Rohit Gargrpg@gmail.com wrote: I think using eigen over gsl (for linear algebra) is a no brainer. a) gsl is dog slow compared to eigen (no vectorization whatsoever). b) gsl has no float option c) eigen offers both column major and row major storage d) eigen *does* expression templates e) eigen will happily map your data and perform operations on it. f) eigen has way cooler (aka C++-ey) api over gsl/lapack g) eigen is adding support for sparse matrices at a rapid pace. For (nearly full, I think) lapack to C++ bindings, FLENS is a great idea too. But I have never used flens. http://eigen.tuxfamily.org/index.php?title=Main_Page I would recommend using eigen/FLENS over gsl anyday for linear algebra purposes. BTW, it also does small vectors/matrices. :) DISCLAIMER: I have contributed and continue to contribute to eigen. On Wed, Sep 2, 2009 at 1:38 AM, Rodney Sparapanirspar...@mcw.edu wrote: John D Lamb wrote: I’ve written some wrappers for GSL matrices and vectors and might can help if someone has a suitable project. While Rodney is right that there are some issues with creating a practical OO interface, C++ is not the same as OO and OO is not the same as operator overloading (assuming OO means object orientation). I was talking about both OO and operator overloading. Another approach to wrapping GSL functions for C++ is to use shared pointers, namespaces and exceptions. Typically I might create a class gsl::vector that is an extended shared pointer and whose member functions look much like the original functions. The main benefit is not having to deallocate memory explicitly and only once. Well, it seems like there is plenty of interest. We should all work together and create an add-on. We can call it GSL++. I'm in the process of working on my PhD which will mean that I'll have an ample opportunity to use it. However, I strongly recommend that the test-bed be integral rather than an after-thought. For each feature, there should be a corresponding check. I don't really understand dejaGNU/etc. so I can't contribute much to this aspect. Rodney ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl -- Rohit Garg http://rpg-314.blogspot.com/ Senior Undergraduate Department of Physics Indian Institute of Technology Bombay ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl -- HZ ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
On Sun, 2009-09-06 at 23:29 +0800, Hongzheng Wang wrote: If I really need a C++ style vector/matrix library, I would refer to boost uBlas + bindings. But for daily normal numerical work I encountered, gsl is easily used and better because vector/matrix support is not the whole world; we usually need more numerical computation facilities beyond basic vector/matrix support and gsl is such a quite complete solution. This is more or less my view. I like GSL for what it can do. Sometimes I use other libraries or write my own versions of GSL functions when they are not enough. What I’d like is a simple C++ wrapper for GSL that provides C++ features not readily available in C: (1) shared pointer or similar so that you don’t have to worry about freeing pointers (2) Exception handling instead of int error values from functions. (3) Better use of const (4) Templates where appropriate, e.g. gsl::vectordouble, gsl::vectorfloat and the like, even if the only implementations are only specialisations. (5) Use of namespaces to simplify #includes. Occasionally other C++ features might be useful: Perhaps iterators for gsl::vector so that it’s easier to use the STL. Perhaps operator overloading for GSL::complex. Perhaps function objects where this is more natural than using a function. However, the main advantage for me is that more errors are caught at compile-time in C++. I’m only interested in object orientation, operator overloading and the like if there’s some obvious advantage to using them. If anyone wants to work on wrappers like this, I’m happy to help. -- JDL ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
Designing a good C++ wrapper to any matrix library is difficult provided one tries to keep C++ semantics. It may be easier, if you have a favorite C++ matrix/vector library, to design a GSL-wrapper for the matrix/vector library itself. For example, write wrappers that accept ublas [1] or eigen2 [2] matrices and vectors, wrap the same storage with gsl_matrix and gsl_vector views, and then call the GSL methods directly. Such GSL-ublas or GSL-eigen2 bindings would be independent of both projects, mainly boilerplate, and easy to test for correctness. - Rhys [1] http://www.boost.org/doc/libs/1_40_0/libs/numeric/ublas/doc/index.htm [2] http://eigen.tuxfamily.org/index.php ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
Re: [Help-gsl] Re: gsl c++ wrapper ?
On 8/31/09, Rhys Ulerich rhys.uler...@gmail.com wrote: Designing a good C++ wrapper to any matrix library is difficult provided one tries to keep C++ semantics. It may be easier, if you have a favorite C++ matrix/vector library, to design a GSL-wrapper for the matrix/vector library itself. For example, write wrappers that accept ublas [1] or eigen2 [2] matrices and vectors, wrap the same storage with gsl_matrix and gsl_vector views, and then call the GSL methods directly. Such GSL-ublas or GSL-eigen2 bindings would be independent of both projects, mainly boilerplate, and easy to test for correctness. - Rhys Many good and tested libraries like LAPACK have come from Fortran and use column-major matrix storage incompatible with GSL. One cannot simply wrap them up in some sort of GSL objects that can be passed to GSL functions. --Leo-- ___ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl
