Re: [PD] sigmund list sort
On Sat, Feb 25, 2012 at 02:45:38PM -0500, Mathieu Bouchard wrote: Le 2012-02-25 à 09:34:00, Frank Barknecht a écrit : Yes, exactly. I often use data structures, well, as data structures and almost never use them for scores in a UPIC sense. What's UPIC ? I was referring to the UPIC composition tool by Iannis Xenakis (Unité Polyagogique Informatique du CEMAMu), that was one of the inspirations for the graphical score capabilities in data structures. See http://en.wikipedia.org/wiki/UPIC Ciao -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
On Fri, Feb 24, 2012 at 10:40:52AM -0800, Jonathan Wilkes wrote: - Original Message - From: Mathieu Bouchard ma...@artengine.ca To: Miller Puckette m...@ucsd.edu Cc: pd-list@iem.at Sent: Friday, February 24, 2012 1:16 PM Subject: Re: [PD] sigmund list sort Le 2012-02-24 à 09:51:00, Miller Puckette a écrit : It's odd, but it never occurred to me that one should be able to specify which field(s) to sort on -- it's x, then y. I should fix this... Nearly all the music I ever composed used a vertical time axis. Lots of people are in this situation, though perhaps not many are also into Pd DS. Data structures can be used for more than drawing a score with a vertical time axis. In Frank's case there aren't even any drawing instructions-- the only reason he's using x is because that's the field the canvas sort method uses. Yes, exactly. I often use data structures, well, as data structures and almost never use them for scores in a UPIC sense. Ciao -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Hi, On Fri, Feb 24, 2012 at 01:26:54PM -0500, Mathieu Bouchard wrote: I have Pd-extended 42.5 that contains Michał Seta's sort, which used to be cubic (O(n³)) and with even lower sorting limits, until I made you replace the O(n²) [list-drip] that used O(n) stack, by one that runs in O(n) and uses O(log n) stack. I don't know any more recent version of list-abs. To me the home of [list]-abs is the CVS repository, but I'm a bad boy who practically never does proper releases. Ciao -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
On Sat, Feb 25, 2012 at 09:38:20AM +0100, Frank Barknecht wrote: Hi, On Fri, Feb 24, 2012 at 01:26:54PM -0500, Mathieu Bouchard wrote: I have Pd-extended 42.5 that contains Michał Seta's sort, which used to be cubic (O(n³)) and with even lower sorting limits, until I made you replace the O(n²) [list-drip] that used O(n) stack, by one that runs in O(n) and uses O(log n) stack. I don't know any more recent version of list-abs. To me the home of [list]-abs is the CVS repository, Oops, SVN of course. At Sourceforge. Ciao -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-25 à 09:34:00, Frank Barknecht a écrit : Yes, exactly. I often use data structures, well, as data structures and almost never use them for scores in a UPIC sense. What's UPIC ? __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-25 à 09:45:00, Frank Barknecht a écrit : On Sat, Feb 25, 2012 at 09:38:20AM +0100, Frank Barknecht wrote: On Fri, Feb 24, 2012 at 01:26:54PM -0500, Mathieu Bouchard wrote: I don't know any more recent version of list-abs. To me the home of [list]-abs is the CVS repository, Oops, SVN of course. At Sourceforge. Do you mean RCS, or do you mean SCCS ? __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
On Sat, Feb 18, 2012 at 01:14:22PM -0500, Mathieu Bouchard wrote: Le 2012-02-18 à 09:58:00, Miller Puckette a écrit : I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) the [list-sort] abstraction uses a high-constant O(n²) algorithm that breaks once you try to sort more than 125 values. Actually [list-sort] since quite some time uses the sort method borrowed from Pd's data-structures for sorting. The problem here is not so much the sorting algorithm, which is very fast and can sort way more than 125 items. However copying the list to a data structure and back - this currently indeed has a problem with stack-overflows, as I'm now aware. Have to think about a fix ... Ciao -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Hi, attached is an approach using data structures to do the sorting of peaks. Somehow I forgot how to set the sort field in a data structure and cannot find it in the docs. Wasn't there some way to specify a different field than x for sorting? (Maybe it was simply sort fielname which I didn't try in my quick patch.) Ciao -- Frank On Sat, Feb 18, 2012 at 09:58:50AM -0800, Miller Puckette wrote: I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) cheers Miller On Sat, Feb 18, 2012 at 04:19:58PM +0100, labyrinthuscochlearis wrote: hi all, what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? thanks, christian ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Frank BarknechtDo You RjDj.me? _ __footils.org__ peak-sort.pd Description: application/puredata set-peak.pd Description: application/puredata ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
It's odd, but it never occurred to me that one should be able to specify which field(s) to sort on -- it's x, then y. I should fix this... cheers Miller On Fri, Feb 24, 2012 at 06:46:08PM +0100, Frank Barknecht wrote: Hi, attached is an approach using data structures to do the sorting of peaks. Somehow I forgot how to set the sort field in a data structure and cannot find it in the docs. Wasn't there some way to specify a different field than x for sorting? (Maybe it was simply sort fielname which I didn't try in my quick patch.) Ciao -- Frank On Sat, Feb 18, 2012 at 09:58:50AM -0800, Miller Puckette wrote: I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) cheers Miller On Sat, Feb 18, 2012 at 04:19:58PM +0100, labyrinthuscochlearis wrote: hi all, what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? thanks, christian ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list -- Frank BarknechtDo You RjDj.me? _ __footils.org__ ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-24 à 09:51:00, Miller Puckette a écrit : It's odd, but it never occurred to me that one should be able to specify which field(s) to sort on -- it's x, then y. I should fix this... Nearly all the music I ever composed used a vertical time axis. Lots of people are in this situation, though perhaps not many are also into Pd DS. __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-24 à 09:02:00, Frank Barknecht a écrit : On Sat, Feb 18, 2012 at 01:14:22PM -0500, Mathieu Bouchard wrote: the [list-sort] abstraction uses a high-constant O(n²) algorithm that breaks once you try to sort more than 125 values. Actually [list-sort] since quite some time uses the sort method borrowed from Pd's data-structures for sorting. I have Pd-extended 42.5 that contains Michał Seta's sort, which used to be cubic (O(n³)) and with even lower sorting limits, until I made you replace the O(n²) [list-drip] that used O(n) stack, by one that runs in O(n) and uses O(log n) stack. I don't know any more recent version of list-abs. The problem here is not so much the sorting algorithm, which is very fast and can sort way more than 125 items. Indeed, it's a slow version of mergesort written using linkedlists in C, which is still faster than nearly anything one could possibly come up with in pure Pd. However copying the list to a data structure and back - this currently indeed has a problem with stack-overflows, as I'm now aware. Have to think about a fix ... Make sure that your algos take less than O(n) stack space... If it's not a 125 element limit, it's 500 or 1000, and it can't get higher without recompiling Pd to be more indulgent. __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
- Original Message - From: Mathieu Bouchard ma...@artengine.ca To: Frank Barknecht f...@footils.org Cc: pd-list@iem.at Sent: Friday, February 24, 2012 1:26 PM Subject: Re: [PD] sigmund list sort Le 2012-02-24 à 09:02:00, Frank Barknecht a écrit : On Sat, Feb 18, 2012 at 01:14:22PM -0500, Mathieu Bouchard wrote: the [list-sort] abstraction uses a high-constant O(n²) algorithm that breaks once you try to sort more than 125 values. Actually [list-sort] since quite some time uses the sort method borrowed from Pd's data-structures for sorting. I have Pd-extended 42.5 that contains Michał Seta's sort, which used to be cubic (O(n³)) and with even lower sorting limits, until I made you replace the O(n²) [list-drip] that used O(n) stack, by one that runs in O(n) and uses O(log n) stack. I don't know any more recent version of list-abs. The problem here is not so much the sorting algorithm, which is very fast and can sort way more than 125 items. Indeed, it's a slow version of mergesort written using linkedlists in C, which is still faster than nearly anything one could possibly come up with in pure Pd. However copying the list to a data structure and back - this currently indeed has a problem with stack-overflows, as I'm now aware. Have to think about a fix ... Use an iterative loop instead of a recursive one and you will avoid the stack overflows. Make sure that your algos take less than O(n) stack space... If it's not a 125 element limit, it's 500 or 1000, and it can't get higher without recompiling Pd to be more indulgent. __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
- Original Message - From: Mathieu Bouchard ma...@artengine.ca To: Miller Puckette m...@ucsd.edu Cc: pd-list@iem.at Sent: Friday, February 24, 2012 1:16 PM Subject: Re: [PD] sigmund list sort Le 2012-02-24 à 09:51:00, Miller Puckette a écrit : It's odd, but it never occurred to me that one should be able to specify which field(s) to sort on -- it's x, then y. I should fix this... Nearly all the music I ever composed used a vertical time axis. Lots of people are in this situation, though perhaps not many are also into Pd DS. Data structures can be used for more than drawing a score with a vertical time axis. In Frank's case there aren't even any drawing instructions-- the only reason he's using x is because that's the field the canvas sort method uses. -Jonathan __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
hi, thank you, i will look at it right away! all the best, christian Am Freitag, den 24.02.2012, 18:46 +0100 schrieb Frank Barknecht: Hi, attached is an approach using data structures to do the sorting of peaks. Somehow I forgot how to set the sort field in a data structure and cannot find it in the docs. Wasn't there some way to specify a different field than x for sorting? (Maybe it was simply sort fielname which I didn't try in my quick patch.) Ciao -- Frank On Sat, Feb 18, 2012 at 09:58:50AM -0800, Miller Puckette wrote: I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) cheers Miller On Sat, Feb 18, 2012 at 04:19:58PM +0100, labyrinthuscochlearis wrote: hi all, what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? thanks, christian ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Thank you! I really have to get into GridFlow. All the best, c Am Samstag, den 18.02.2012, 11:28 -0500 schrieb Mathieu Bouchard: Le 2012-02-18 à 16:19:00, labyrinthuscochlearis a écrit : what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? GridFlow's [#grade] gives you a list of item numbers in the order that you need to pick them so that they be sorted. This can be used for sorting a table with multiple columns according to one column, whereas other sorting tools in Pd might only support sorting individual values. http://gridflow.ca/help/%23grade-help.html You will need the appropriate conversion from list to grid (a kind of super-list type) and grid to list. Also, [#store] is a great shortcut for reordering elements using the output of [#grade]. This uses a plugin that you'd download from http://gridflow.ca/ __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
[PD] sigmund list sort
hi all, what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? thanks, christian ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-18 à 16:19:00, labyrinthuscochlearis a écrit : what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? GridFlow's [#grade] gives you a list of item numbers in the order that you need to pick them so that they be sorted. This can be used for sorting a table with multiple columns according to one column, whereas other sorting tools in Pd might only support sorting individual values. http://gridflow.ca/help/%23grade-help.html You will need the appropriate conversion from list to grid (a kind of super-list type) and grid to list. Also, [#store] is a great shortcut for reordering elements using the output of [#grade]. This uses a plugin that you'd download from http://gridflow.ca/ __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) cheers Miller On Sat, Feb 18, 2012 at 04:19:58PM +0100, labyrinthuscochlearis wrote: hi all, what would be a good way to transform sigmund~'s peaks output so that I get a list with peak amplitudes but in the ascending order of the corresponding frequencies? thanks, christian ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
Re: [PD] sigmund list sort
Le 2012-02-18 à 09:58:00, Miller Puckette a écrit : I believe there's no good way to do this in pd vanilla. THere should be a 'list sort' but I haven't figured out what would be the best design. (and there's probably already a list sort in Pd extended :) the [list-sort] abstraction uses a high-constant O(n²) algorithm that breaks once you try to sort more than 125 values. For [sigmund~]'s output this is not very relevant, what's relevant is that it can't deal with more than one « column », that is, it can't sort pairs of numbers. That's why I mention [#grade]. Not only it's using a lean O(n log n) algorithm that can work on millions of elements, it also gives you the ordering instead of applying the sorting. This means that you can fairly easily construct a sorter by any kind of key column, including cases computed on the fly (e.g. sort according to dbA computed using both amplitude and frequencies, where the dbA is not part of the table). That kind of modular design is something I borrowed from the APL language. __ | Mathieu BOUCHARD - téléphone : +1.514.383.3801 - Montréal, QC___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management - http://lists.puredata.info/listinfo/pd-list
