brunatius

Is there a data structure that is matrix like, Similar to what we have in jit.gen?

What are the options if I want sample rate control over a 1000x1000 matrix in a codebox?

Is there a data structure that is matrix like, Similar to what we have in jit.gen? 

What are the options if I want sample rate control over a 1000x1000 matrix in a codebox?


matrixarray-in-gen

the 2nd and 3rd arguments are like the dimensions for columns and rows(2nd argument is length of each channel in samples, 3rd argument is number of channels).

i just tried and i can make a data that is 1000x1000 by creating, for example, "data wut 1000 1000", and then to read, you can use many objects like wave, sample, etc. but you can either choose one which specifies number of channels to output(i just tried with sample and it's amazing, it actually created 1000 outlets, but then crashed shortly after, haha :D ...but i was just testing with the visual op objects, probably behaves easier in codebox ;), or you can choose an op like wave which can allow you to access a channel at a time by specifying channel offsets... see reference and docs for 'data', 'sample', 'wave', etc. (use them in codebox of course :)

hehe, yes, it's called 'data' :D
the 2nd and 3rd arguments are like the dimensions for columns and rows(2nd argument is length of each channel in samples, 3rd argument is number of channels).
i just tried and i can make a data that is 1000x1000 by creating, for example, "data wut 1000 1000", and then to read, you can use many objects like wave, sample, etc. but you can either choose one which specifies number of channels to output(i just tried with sample and it's amazing, it actually created 1000 outlets, but then crashed shortly after, haha :D ...but i was just testing with the visual op objects, probably behaves easier in codebox ;), or you can choose an op like wave which can allow you to access a channel at a time by specifying channel offsets... see reference and docs for 'data', 'sample', 'wave', etc. (use them in codebox of course :)

Well shit. This is pretty huge, and I find myself craving that sort of data access... but how do I begin to dive into this? Raja, any ideas on how to get me past the "hacking" stage of gen?

we do have this in vanilla msp as well, it works a little differently to setup, but can be accessed sample-accurately: 2d.wave~ (and i've not even worked at all with polybuffer~ yet :p, nor have i gone beyond 4-channels of buffer~ access... are there more? probably... so much damn stuff, whatever shall i do with these 1st-world probs ;D)

Raja, any ideas on how to get me past the "hacking" stage of gen?

I think it's just a matter of spending the time on sample-accurate operations in msp(so basically, i think you're getting there just fine... going from my experience, every single person i've seen here these days is learning faster than i ever did, no exceptions to anyone despite what anyone might think of the ogre-character i attempt to play here often :p(the docs have improved greatly since max2 :D))

I've been extremely haphazard at learning, my DSP understanding started with grooveduck, then went to sampling(granular synthesis), then expanded from there to simple delays, filters, eventually got fed up and decided to read through books by Charles Dodge, Curtis Roads, Gareth Loy, Eric Lyon, etc. (even then, those books for awhile helped me backtrack to understand better what i had done before... but also beyond that, helped me understand things like anti-aliasing, interpolation, FFT, filter-coefficients, etc. much much bettererer)

If you're looking for DSP stuff, it's more in the books(or maybe even stanford-style courses), than online(you will have gaps in your knowledge if you only go by examples from others, no one will annotate everything perfectly for everyone else's language and learning style).

If you're looking for how to code, that just takes time: just like learning a regular speaking language of any kind.

(if you have more specifics of something in particular you're looking to do, i might be able to give more specific links or somethin...)

But basically(and truthfully) i dunno... only idea i have is to keep going at it(i believe you, in particular, to be much smarter than me anyways: you have a good sense of social-etiquette(not just here in text form, but i've also seen you in those ableton and lemur vids from awhile back, too... very personable)... which i do not have the maturity yet to grasp ;D there's a bit of a prob with american-style communication which has infected me: blunt communication is honest and direct communication without a good sense of timing(usually caused by the impetuousness of pride), fakeness can be well-timed communication but without the honesty or directness... good communication involves both honesty and the humility to wait for the right timing. we americans have a problem with cultivating the thoughtful use of both.)

So have faith in yourself! You're already ahead of the game in many ways you probably don't realize yet(or probably do) :)

I think it's just a matter of spending the time on sample-accurate operations in msp(so basically, i think you're getting there just fine... going from my experience, every single person i've seen here these days is learning faster than i ever did, no exceptions to anyone despite what anyone might think of the ogre-character i attempt to play here often :p(the docs have improved greatly since max2 :D))
I've been extremely haphazard at learning, my DSP understanding started with grooveduck, then went to sampling(granular synthesis), then expanded from there to simple delays, filters, eventually got fed up and decided to read through books by Charles Dodge, Curtis Roads, Gareth Loy, Eric Lyon, etc. (even then, those books for awhile helped me backtrack to understand better what i had done before... but also beyond that, helped me understand things like anti-aliasing, interpolation, FFT, filter-coefficients, etc. much much bettererer)
If you're looking for DSP stuff, it's more in the books(or maybe even stanford-style courses), than online(you will have gaps in your knowledge if you only go by examples from others, no one will annotate everything perfectly for everyone else's language and learning style).
If you're looking for how to code, that just takes time: just like learning a regular speaking language of any kind.
(if you have more specifics of something in particular you're looking to do, i might be able to give more specific links or somethin...)

But basically(and truthfully) i dunno... only idea i have is to keep going at it(i believe you, in particular, to be much smarter than me anyways: you have a good sense of social-etiquette(not just here in text form, but i've also seen you in those ableton and lemur vids from awhile back, too... very personable)... which i do not have the maturity yet to grasp ;D there's a bit of a prob with american-style communication which has infected me: blunt communication is honest and direct communication without a good sense of timing(usually caused by the impetuousness of pride), fakeness can be well-timed communication but without the honesty or directness... good communication involves both honesty and the humility to wait for the right timing. we americans have a problem with cultivating the thoughtful use of both.)
So have faith in yourself! You're already ahead of the game in many ways you probably don't realize yet(or probably do) :)
Cheers!


Hi Raja, thanks for the reply. My question was ill posed, let me be a bit more concrete and clear. I am basically trying to test run different physical modeling models, ie, a plate in 2D. So, this means that not only do I need to have a way to apply a function to a 2d array (trivial with while/for loops, just not optimal) but I also need to iterate (at least) at each sample and keep the previous value. This works for History but I'm not sure if there's a way to do it combining history and data?

I read about data but when I saw "channels", had alarm bells in my mind saying this is likely not optimized for acting as a 2d matrix.

I'm wondering if the right answer is an external.. :-/

Btw, I think it would help a lot if cycling included in the help files how to properly implement these things in codebox, a simple function "header" (.h style) would do as opposed to us trying to figure out the proper syntax...

Hi Raja, thanks for the reply. My question was ill posed, let me be a bit more concrete and clear. I am basically trying to test run different physical modeling models, ie, a plate in 2D. So, this means that not only do I need to have a way to apply a function to a 2d array (trivial with while/for loops, just not optimal) but I also need to iterate (at least) at each sample and keep the previous value. This works for History but I'm not sure if there's a way to do it combining history and data? 

Btw, I think it would help a lot if cycling included in the help files how to properly implement these things in codebox, a simple function "header" (.h style) would do as opposed to us trying to figure out the proper syntax...


data is basically just an array. channels can be thought of as the y-dimension.

how you read from data is where it all matters. 'sample' and 'lookup' can allow you to @index samples (with @interp none if you also need; see reference doc for these ops' attributes and messages)... the 'channel offset' input to either one can be changed at sample-rate allowing you to read columns just like rows(this is optimal for audio).

for your history, you can either have 2 read-ops that access 2 consecutive points within data('consecutiveness' determined by how you control the index(x) and channel(y) inputs of course... edit: or maybe it's index(y) and channel(x), yess, actually... that's probably more like it :p you can test and figure it out), or add the history op to the output of 1 read-op.

I’m wondering if the right answer is an external.. :-/

Btw, I think it would help a lot if cycling included in the help files how to properly implement these things in codebox, a simple function "header" (.h style) would do as opposed to us trying to figure out the proper syntax…

Must admit, sometimes writing an external seems quick and easy enough.

Agreed the gen docs could be improved, but there's much there already(i'm looking in Max7)... you can search 'GenExpr' in the help docs which is where you'll find basics for codebox. It's very helpful as is, even some Jitter things for codebox there.

You can also create a patch with visual op objects, real simple bullshit that uses the ops you want to learn more about, then from outside the gen~ patcher, send the message 'exportcode' to it(specify the folder for the code-export first in the inspector for the gen~ patcher), and although they are cpp.h, they will tell you quite a bit(whether you decide to try to continue to learn codebox or write your own external ;).

data is basically just an array. channels can be thought of as the y-dimension.
how you read from data is where it all matters. 'sample' and 'lookup' can allow you to @index samples (with @interp none if you also need; see reference doc for these ops' attributes and messages)... the 'channel offset' input to either one can be changed at sample-rate allowing you to read columns just like rows(this is optimal for audio).
for your history, you can either have 2 read-ops that access 2 consecutive points within data('consecutiveness' determined by how you control the index(x) and channel(y) inputs of course... edit: or maybe it's index(y) and channel(x), yess, actually... that's probably more like it :p you can test and figure it out), or add the history op to the output of 1 read-op.

I’m wondering if the right answer is an external.. :-/
Btw, I think it would help a lot if cycling included in the help files how to properly implement these things in codebox, a simple function "header" (.h style) would do as opposed to us trying to figure out the proper syntax…

Must admit, sometimes writing an external seems quick and easy enough.
Agreed the gen docs could be improved, but there's much there already(i'm looking in Max7)... you can search 'GenExpr' in the help docs which is where you'll find basics for codebox. It's very helpful as is, even some Jitter things for codebox there.

You can also create a patch with visual op objects, real simple bullshit that uses the ops you want to learn more about, then from outside the gen~ patcher, send the message 'exportcode' to it(specify the folder for the code-export first in the inspector for the gen~ patcher), and although they are cpp.h, they will tell you quite a bit(whether you decide to try to continue to learn codebox or write your own external ;).


Hello, perhaps any news in this regard? I just approached gen/gen~ and while I was happy to see for(), while() loops and conditional statements, I'm pretty disappointed there's no way to use arrays or matrices...

data and buffer are arrays. if you want to treat them as 2d or Nd that is pretty straightforward. as far as jitter matrix - one could use jit.buffer~ to convert a matrix to a buffer and then read/write to it in gen~ and then convert back to a matrix.

Matrix/array in gen~