[sage-support] Re: Seg fault with determinant calculation
On Sep 22, 4:23 pm, phil [EMAIL PROTECTED] wrote: On Sep 15, 10:26 am, Martin Albrecht [EMAIL PROTECTED] wrote: The original machine I was using was needed for other things. So, I ran it on sage 3.1.2rc4 on sage.math.washington.edu and it completed successfully after 169446 seconds. So, the problem was specific to the setup I was using or it was fixed in 3.1.2rc4. The scaling of the problem seems worse than it should be though. The 9x9 problem takes 40 seconds while the 10x0 problem takes 4236 times longer. That's worse than O(n!) let along O(n^3). If your curious, the test problem is athttp://sage.math.washington.edu/home/fongpwf/sage_work/determinant_10... One more thing I've noticed is that loading the saved result either doesn't work is is extremely inefficient. On Monday, I started up 3.1.2 final and ran load detMp.sobj. It's been rough 3 days and it's still going. This means loading takes longer than the original computation. Phil --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Sep 25, 12:34 pm, phil [EMAIL PROTECTED] wrote: Hi, I did valgrind a couple of the example sessions you gave for various n (n=4,5 and 8 IIRC) and the good news is that there are no leaks going on and Singular does not seem to do anything naughty. One more thing I've noticed is that loading the saved result either doesn't work is is extremely inefficient. On Monday, I started up 3.1.2 final and ran load detMp.sobj. It's been rough 3 days and it's still going. This means loading takes longer than the original computation. Yeah, pickling for those matrices over the mv polynomial ring probably uses generic code and hence sucks speed wise. If malb or anyone else familiar with that are of the code agrees please open a ticket so we remember to fix this. Phil Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Sep 15, 10:26 am, Martin Albrecht [EMAIL PROTECTED] wrote: On Monday 15 September 2008, phil wrote: I've been pushing the limits of determinant calculation over multivariate polynomial rings. I can calculate determinants of matrices up to 9x9 of the form [[x_0_0, x_0_1],[x_1_0, x_1_1]] (each element is a single unique variable). When I get to 10x10 is runs for a while the crashes with: Unhandled SIGSEGV: A segmentation fault occured in SAGE. This probably occured because a *compiled* component of SAGE has a bug in it (typically accessing invalid memory) or is not properly wrapped with _sig_on, _sig_off. You might want to run SAGE under gdb with 'sage -gdb' to debug this. SAGE will now terminate (sorry). I'll try to reproduce the crash and see what I can do about it. You could help by running sage -gdb (if you have gdb installed) and send me the backtrace off list. Thanks. The original machine I was using was needed for other things. So, I ran it on sage 3.1.2rc4 on sage.math.washington.edu and it completed successfully after 169446 seconds. So, the problem was specific to the setup I was using or it was fixed in 3.1.2rc4. The scaling of the problem seems worse than it should be though. The 9x9 problem takes 40 seconds while the 10x0 problem takes 4236 times longer. That's worse than O(n!) let along O(n^3). If your curious, the test problem is at http://sage.math.washington.edu/home/fongpwf/sage_work/determinant_10_poly.sage Phil --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Sep 22, 4:23 pm, phil [EMAIL PROTECTED] wrote: On Sep 15, 10:26 am, Martin Albrecht [EMAIL PROTECTED] wrote: SNIP I'll try to reproduce the crash and see what I can do about it. You could help by running sage -gdb (if you have gdb installed) and send me the backtrace off list. Thanks. The original machine I was using was needed for other things. So, I ran it on sage 3.1.2rc4 on sage.math.washington.edu and it completed successfully after 169446 seconds. So, the problem was specific to the setup I was using or it was fixed in 3.1.2rc4. The scaling of the problem seems worse than it should be though. The 9x9 problem takes 40 seconds while the 10x0 problem takes 4236 times longer. That's worse than O(n!) let along O(n^3). If your curious, the test problem is athttp://sage.math.washington.edu/home/fongpwf/sage_work/determinant_10... It would be worth it to check foe either memory leaks or something else fishy going on here via valgrind. Let me poke around to see if I can find anything either in Sage or in libSingular. Phil Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Mon, Sep 15, 2008 at 10:05 AM, phil [EMAIL PROTECTED] wrote: I've been pushing the limits of determinant calculation over multivariate polynomial rings. I can calculate determinants of matrices up to 9x9 of the form [[x_0_0, x_0_1],[x_1_0, x_1_1]] (each element is a single unique variable). When I get to 10x10 is runs for a while the crashes with: Unhandled SIGSEGV: A segmentation fault occured in SAGE. This probably occured because a *compiled* component of SAGE has a bug in it (typically accessing invalid memory) or is not properly wrapped with _sig_on, _sig_off. You might want to run SAGE under gdb with 'sage -gdb' to debug this. SAGE will now terminate (sorry). How much RAM do you have? Write to me off list if you want access to a machine with more :-) 9x9 matrices only take about 40 seconds. The 10x10 calculation runs for a long time (1 hr) before crashing. This is on 64 bit Ubuntu with the patch in Trac #4068 (http://trac.sagemath.org/sage_trac/ ticket/4068) applied. BTW, what is the underlying algorithm used for the determinants? As I understand it, the naive way is O(N!) while the recursive way is O(N^3) for a NxN matrix. Also, occured is misspelled in the error message. It should be occurred. -- William Stein Associate Professor of Mathematics University of Washington http://wstein.org --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Monday 15 September 2008, phil wrote: I've been pushing the limits of determinant calculation over multivariate polynomial rings. I can calculate determinants of matrices up to 9x9 of the form [[x_0_0, x_0_1],[x_1_0, x_1_1]] (each element is a single unique variable). When I get to 10x10 is runs for a while the crashes with: Unhandled SIGSEGV: A segmentation fault occured in SAGE. This probably occured because a *compiled* component of SAGE has a bug in it (typically accessing invalid memory) or is not properly wrapped with _sig_on, _sig_off. You might want to run SAGE under gdb with 'sage -gdb' to debug this. SAGE will now terminate (sorry). I'll try to reproduce the crash and see what I can do about it. You could help by running sage -gdb (if you have gdb installed) and send me the backtrace off list. Thanks. 9x9 matrices only take about 40 seconds. The 10x10 calculation runs for a long time (1 hr) before crashing. This is on 64 bit Ubuntu with the patch in Trac #4068 (http://trac.sagemath.org/sage_trac/ ticket/4068) applied. BTW, what is the underlying algorithm used for the determinants? As I understand it, the naive way is O(N!) while the recursive way is O(N^3) for a NxN matrix. See: http://groups.google.com/group/sage-devel/browse_thread/thread/7aa1bd1e945ff372/ Cheers, Martin -- name: Martin Albrecht _pgp: http://pgp.mit.edu:11371/pks/lookup?op=getsearch=0x8EF0DC99 _www: http://www.informatik.uni-bremen.de/~malb _jab: [EMAIL PROTECTED] --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
[sage-support] Re: Seg fault with determinant calculation
On Sep 15, 10:08 am, William Stein [EMAIL PROTECTED] wrote: How much RAM do you have? Write to me off list if you want access to a machine with more :-) Ok. I'll send you an off list message. I'm running Sage on 64 bit Ubuntu installed in a VMWare Infrastructure virtual machine with 6 GB of memory allocated. The underlying physical machine has 8 GB. However, free and top only show about 5 GB of total memory. Some how the kernel does not use all 6 GB that is allocated. --~--~-~--~~~---~--~~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~--~~~~--~~--~--~---
