Suggested change to frequency resolution

[mgold]

I'm referring to this page, figure 3, which seems to be working as intended but I found to be misleading.

At first I thought the samples better approximated an underlying analog curve. But it soon dawned on my that the horizontal time scale is not constant, and the curve compresses in from the right. This is because the sampling frequency is fixed and the time domain increases.

I think a more honest visualization is to show the underlying curve with sample points placed only on the left side. More points are added left to right as the slider is moved. Conceptually, equal horizontal spacing corresponds to the constant sampling rate.

This is unsolicited advice that you are free to take or leave as you like. I can also try implementing it myself if you assure me you'll merge it.

[mgold]

Although, doing so would break the charts on the next page, so I guess it's a wash. Maybe try to introduce it in the text a little more explicitly, which is certainly an easier solution than writing code.

[jackschaedler]

Hi Max,

This is a good point, and I initially had the same concern when writing these two sections. I'll see what I can do! I think a little visualization showing how much of the input signal is 'consumed' might help.

[mgold]

Thanks Jack. It's a really neat project, and enjoyed reading it.

[jackschaedler]

Thanks Max,

I appreciate it! Your visualization of Melkman's algorithm is really nicely done. And it's nice to see your work on the explorable explanations wiki.