Continuous Functions Module

[MarcelloPoletti]

Hi,
There are many cases where there is a need to represent continuous trends rather than discrete ones. Some examples include:

• Trend lines, which are not necessarily linear.
• Line annotations, which are not necessarily straight lines.

These types of series or annotations are essentially continuous functions rather than a set of points. The topic is evidently complex and broad. There are challenges in defining the domain and codomain of functions, and functions can be expressed in various ways - explicitly, parametrically, and so on.

However, by imposing certain restrictions (at least initially), we could create a powerful tool. For instance, we could consider only:

• Explicit functions, where y = f(x).
• A domain interval, possibly open (e.g., [-∞, +∞]).
• An optional density parameter that specifies the drawing precision (perhaps unnecessary?).

A series or annotation associated with a function could then be drawn by the engine. Naturally, this leads to some special behaviors:

• A function-based series would not have markers.
• A function-based series would not automatically define any axis visibility range.
• Crosshairs, tooltips, etc., would require careful consideration.

In summary, this is indeed a non-trivial task, but if someone willing takes it up, it would benefit the entire community! 😊

Marc.

[karolkolodziej]

Hi @MarcelloPoletti!

Thank you for sharing this idea!

Right now with a bit of code, you can create such a thing- see the y = f(x) demo.

But if I understand you correctly your idea is to have instead of the data in the series something like this:

  series: [{
    type: 'function',
    equation: 'x^2 + 5'
    range: [-10, 10]
  }]

If you have any other idea how it might look or work please feel free to share that!

[MarcelloPoletti]

Hi karolkolodziej,

Yes, a function type series it's probably the best way. And it can have some params, in example:

series: [{
    type: 'function',
    equation: function(x){ return x^2 + 5},
    range /*optional*/: [-10, 10], //or [-10, null] or [null, 10]...
    density /*optional*/: 10, //or nothing, draw a point every 10 pixel 
  }]

marc.

[karolkolodziej]

Thank you for clarifying! We are going to discuss it internally and we will keep you posted!

[MarcelloPoletti]

Another way to think the problem, perhaps more powerfull, is work on data instead of series.
Suppose "data" depends on xAxys range now visible for a generic series:

series: [{
    type: anytype!!
    data:{
	function: function(o){
		//o contains some information about newminx, newmax, oldminx, oldmax, olddata
		//Here we can produce new dataset in any possible way, using any possibile optimization strategy
	}
    }
}]

marc.