Have i found a new way to generate a square wave?

Have i found a new way to generate a square wave?

desmos.com/calculator/qodirgafgw

Other urls found in this thread:

desmos.com/calculator/kdak7fogpj
twitter.com/NSFWRedditGif

Maybe, congratulations. It's useless, but I bet it was all worth it.

Not a new formula by a long shot, but it's always good to discover something yourself

is a retard

no u

Great. Now how do you vary the duty cycle?

more fiber

>the period is not 1

Dropped.

not op but trying to find out of this is possible with this setup. probably not how its setup but ill try

Wat?

This is just a logistic function, but composed with a periodic function. It is only as useful as a logistic function.

Wow faggot I'm really impressed, the general expression seems to be a lot easier and simpler than an usual Fourier series. It seems that it converges really fast, congratulations. This is a legit publication, go ahead.

This is why Veeky Forums hates your engineering filth.

Try it now

desmos.com/calculator/kdak7fogpj

Pretty neat

Or you can just use a piecewise function or modular arithmetic, you fucking nerd.

Very nice, but the amplitude is half of what it should be

Engineer mathlet here, getting into fourier series. Why can a "square wave" be not perfectly square?
Even if it's infinity close to being square it still isn't "square".

Because straight upward slopes (with derivaate = infinity) aren't possible in anything that is an actual function, as you'd have multiple results for the same value.

you mean gibbs phenomenon? fourier series aren't equal in the point-wise sense, they are equal "almost everywhere" i.e. on all sets of non-zero measure.

That isn't true, the cube root function has an infinite slope at x=0, despite it being defined everywhere. The reason a "square wave" can't be perfectly square is the result of definitions. Typically, people want a square wave to be a continuous line, where the bottom cusps are connected by an unbroken line, to the respective top cusps. However, well-defined functions that vary with respect to x cannot do that. Because people want an actual continuous function to describe their system, they pick faux square-wave functions to do the job, when they could easially design a discontinuous function to do the same thing.

>That isn't true, the cube root function has an infinite slope at x=0
The length of said slope is also 0, so it's not the same case at all.

>implying real life square waves with sufficient rise times show up continuous without interpolation between samples
>implying x^3 is not differentiable on its entire domain

Re read that post
>Because straight upward slopes (with derivative = infinity) aren't possible in anything that is an actual function
the cube root of x is an actual function that has a "straight upward slope (with derivative = infinity)" at x = 0.

>cube root function

I don't think so DESU