The maths ot Max Msp

[maxforlive]

Hi,

I think I realize that if I should quit fucking around with sound programming, and really learn this stuff, I need to get my math skills up to a certain level.
What do I need to know if most topics of Msp except fft should make sense to me? Do I need to be expert in general algebra and geometry.

[emmanuel_2]

Hi,

I think I realize that if I should quit fucking around with sound programming, and really learn this stuff, I need to get my math skills up to a certain level.
What do I need to know if most topics of Msp except fft should make sense to me? Do I need to be expert in general algebra and geometry.

In my experience, you need to master addition and multiplication... I'm not being sarcastic, with that I can do 99% of my patching.

[josquin2000]

Perhaps what is lacking is not arithmetical knowledge, but logic and set theory
that are the basis of all programming:
Basic Symbolic Logic and simple Set Theory
(the basis of what my generation of US students were told is the "new math")
are the skills most used in all programming.

Set theory:
Can you describe what is in done in
a) enumerating a set?
b) applying a filter to a set of data?
c) mapping a function over a set of data? (!)
d) making decisions based on content of data?
e) union / intersection of data sets?

Logic:
Can you:
a) Combine logical values into logical 'product'?
b) reduce real world problems to their base data and logical connections,
and reduce these to their possible truth values ("boolean values"?).
c) use these truth values in controlling process decisions?
d) perform mental logical conversions such as "(not (A and B)) is logically the same as ((not A) or (not B))"?

if so, you have almost all the advanced knowledge you'll need for MaxMSp..now there are advantages in knowing and
applying trig. and calculus in using tools based in these knowledge sets (Fourier Transforms, digital filter theory, etc.). but honestly, a simple intuitive understanding is often all that is needed of these matters, as long as the basic manipulation skills of arithmetic and logic are available to the user...EG: hey the fft results are an array of floating point value pairs that hold the "complex" amp values for each 'band' of analyzed frequency and an iterative "unrolling"
of the relationship between the real and imaginary values contains the full phase information for that freq band...BUT
It's just not needed to play with the values and get cool results! Zack&Cort's cool 'cross-synth' just blindly substitutes one of the pairs with the same values from a different analysis, and VOILA! an EXTREMELY cool "spectral cross synthesis". no fancy knowledge, just hook it up, whoooa, KEWL! ( and it definately not standard "signal convolution")...

just go for it, and if you have not had one, take a course in programming or logic or set theory or database (databases = set theory writ large in the real world.)

Calculus / Trig., well there're useful, but not as important as simple logic.

YMMV,
L&K,
J2K

[pukunui]

I'm with EJ. If you can figure out how to go back and forth between milliseconds and bpm you'll be fine most of the time.

I guess I feel that math is like music in this respect. You'll end up using what you know, or finding your own way round things you don't. It's all yours in the end. If you learn how to transcribe Lute tablature, some of that will end up in your music. If you want to poke around in number theory, that will probably end up in your patch.

Apart from that, forums are like large, semi intelligent symbolic calculators.

-A

[S4racen]

I don't know a thing about math, but logic i can do, combine this with the graphical interface of max and i'm in geekery heaven!

Cheers
D

[henke]

I would go one step further: MAX forces you to learn math, step by step. But for a start you don't need to understand much of it. You make a patch, and suddenly there is a problem. A great example: if you want a crossfade between two signals, you need to raise the volume of one signal while lowering the other one. The most straight forward solution is:

Output A = Input A * X
Output B = Input B * (1-x)

and x is the crossfader from 0 to 1

This works perfectly well. However, you'll notice a lack of energy with most signals, if you set X to 0.5. And to overcome this, you'll need a bit more math. And you don't even have to invent it. Simply Google for "panning laws" and you'll find some usefull formulas....

Just try them out. If it works it works

Robert

[LOFA]

I have collaborated on a projects were a handful of arithmetic ended up being an asymptotic filter. I studied calc but I still don't quite follow how it ended up in that territory.

I have to say that math seems so much easier now that I have worked with max. Nothing intimidates. You start to realize that it is all just a means to an end. Sometimes knowing too much just gets you hung-up in details anyway. There are objects for that.

[maxforlive]

math seems so much easier now that I have worked with max.

Would you also perform better on an IQ test, you think?

[LOFA]

math seems so much easier now that I have worked with max.

Would you also perform better on an IQ test, you think?

Seems like a loaded question or at least impossible to answer in a non-flighty manner. Having said that, I believe anyone who had the interest in something like max or the potential to develop in parallel with their time investment with such a thing would be a good candidate for demonstrating arguments of conditioned development over natural merits.

Give a kid legos first, but with something like max lying around... ... .... ....

[sacredgeometry]

I don't know a thing about math, but logic i can do, combine this with the graphical interface of max and i'm in geekery heaven!

Cheers
D

im the same, logic i can do i leave the maths for the computer to worry about

[maxforlive]

What to do to understand the Slide help file :

slide filters an input value logarithmically between changes. The formula is y (n) = y (n-1) + ((x (n) - y (n-1))/slide). It's particularly useful for envelope following and lowpass filtering.

[Filch]

math seems so much easier now that I have worked with max.

Would you also perform better on an IQ test, you think?

IQ tests are specifically built to preclude training, education, upbringing, etc. They assess intelligence. Math can be learned and through use and practice, one can improve their ability to use it more effectively and efficiently.

The answer to your question, in short, is "no".

[emmanuel_2]

What to do to understand the Slide help file :

slide filters an input value logarithmically between changes. The formula is y (n) = y (n-1) + ((x (n) - y (n-1))/slide). It's particularly useful for envelope following and lowpass filtering.

you need to understand what y(n), and x(n) means and that's it ;-) x(n) is your input value, y(n) is the result, y(n-1) is the previous result. If you have slide == 1, you have y(n) = x(n). x(n) - y(n-1) is the delta between the input and the last output value. You basically apply a coefficient on the delta, the higher the coefficient is the lower will be the result (the lower you add to the previous value) so it goes slower to the target (input) value.