My knowledge of VEX is fairly newborn. So I'm having a hard time figuring out how I can achieve my goal here.
in my vex script, I have a series of float values. These values change depending on the geometry. for the sake of simplicity, let's say this is what I have:


What could I write here to look through those float values and automatically pick whichever one is closest to the target?
Thoughts and ideas are very much appreciated. Thanks!

Hi Poliigon,

you could iterate over an array and store the index and update the minimal absolute difference to the target value.


Alternatively you just set the float values as positions of a point cloud and use the nearpoint()-function.

Konstantin Magnus
Hi Poliigon,

you could iterate over an array and store the index and update the minimal absolute difference to the target value.

float vals[] = array(12.3, 34.1, 54.7, 67.9);
float target = 40.0;

float d_min = 10000.0;
int index = -1;

foreach(int i; float val; vals){
    float d = abs(val - target);
    if(d < d_min){
        d_min = d;
        index = i;
    }
}

f@closest_value = vals[index];

Alternatively you just set the float values as positions of a point cloud and use the nearpoint()-function.

Thanks a lot!

I'm very curious about the point cloud idea, as I had thought of that as well. Though I am unsure how to set the position of points based on those float values. If you have a minute, would you mind breaking that down?

You would overwrite the position:


and search for the closest match:

Konstantin Magnus
You would overwrite the position:

v@P = set(f@val, 0.0, 0.0);

and search for the closest match:

float target = chf('target');

vector pos = set(target, 0, 0);
int pt_near = nearpoint(0, pos);

i@group_target = pt_near == i@ptnum;

You're a legend. Thanks mate!

If you have so many values in a sorted state i.e. in the millions or billions, you can also use binary search to speed up the search exponentially:
https://en.wikipedia.org/wiki/Binary_search_algorithm [en.wikipedia.org]

However Konstantin's methods are already very established for practically any case.