the calculation I am using returns 0 0 so I didn't need it . Before I figured that out, I was using an identity matrix multiplied by the table and it worked out fine
@nathanrogers If you set ⎕DIV←1 then all divisions by 0 will give 0.
@nathanrogers If you want to error on divisions by 0, keep ⎕DIV←0 (the default) and just use ×∘÷ instead of ÷. For code golf, use can use ⌹ instead of ÷ for scalars, and ⌹¨ instead of ÷ on flat arrays.
@nathanrogers Any particular reason you don't backtick APL in your messages? You are of course free to do as you please, but I find it harder to read. Besides, the graduation userscript makes code have darker background, which makes it easier to spot…
for clarification I'm really looking for how to do the ∘.{⍺ attract ⍵} version of that. I can't seem to get it, because when I try I'm getting ? ? as the result in cases where the magnitude is 0. But in this version, where the magnitude is 0, there are no issues.
@Cowsquack a fancy (colors for points, bounds checking, different masses) version of this in pure processing is just above 30 lines, so removing the fanciness would prolly bring it to 30 lines of code, compare that with the apl equivalent
well I also stored the points as objects, that might contribute to some of the verbosity, but still there a lot
@dzaima oh I just realised I had a rogue ⎕div←1 in there forgot to remove it >_>
@nathanrogers if the two parts are different types of information, I would suggest not even joining them up ever (until needed, e.g. for the drawing functions)
just do l2-l1 & m2×m1, it's not that much work and is probably shorter than anything one could come up with with them joined together
I transform it into a column for simple operations accross the rest of the columns, the thing is I don't have a simple way of passing both location and mass to a function taht I need to call
no they're all single vectors
but I have to join location and mass because I need them both in a function for a single calculation
In linear algebra, a rotation matrix is a matrix that is used to perform a rotation in Euclidean space. For example, using the convention below, the matrix
R
=
[
cos
θ
−
sin
θ
sin
θ
...
that's a perfect example of when I might want a matrix of functions
@nathanrogers that takes only one arg, not multiple, each for each column.
multiple functions for one input makes sense. one function for multiple collumns makes sense. Different functions per different columns doesn't make sense, as there's no reason to join them up, if they each need distinct operations.
So, for that reason, I was asking if it was possible in APL. I think its silly that functions aren't actually first class, and that only operators are when the other 2 anscestors of the language saw fit to add them.
heh, this conversation made me think again about implementing first-class functions again and it made me realize that that's a bad decision
also from a memory perspective: imagine if you want to store objects, each with a boolean and a number property. Storing them in a matrix would force the booleans to all be stored as full numbers (persumably 64-bit ones), whereas a vector of the booleans & a vector of the numbers could easily store the booleans as bit booleans
also if another property you'd want would be a string (aka a char vector), the matrix of everything couldn't even be stored as a number array
but, at some point I need to perform calculations which pertain to more than 1 vector
for instance when I need to multiply the distance by something to do with the mass. I need locations and masses of each object, so... it would be COOL if I could find some kind of way to apply functions accross those vectors while I still had them joined together
@Adám I like the array of namespace solution. In addition you can put the vector in the namespace and associate other metadata with it. Effectively a class. Very useful.
as I said this morning, I'm open to any suggestions for how to improve this
I was playing around with the attract function, and thought I saw an opportunity for (-,×) or some other such thing, so I was asking questions
I was trying to make a mce or w/e @dzaima because that's what I was told I should do
and specifically, I need help with the bug having to do with a a⍴... when generating forces. becuase if there are locations < 6, (a< 3) then there is only 1 result but 2 pairs? or something... and the balls disappear off into infinity land
I'm all ears as to coding style and how to fix the thing so there's less of this "join" and "vecs" nonsense every which way. then there's the sun...
@dzaima why is it that (sun←origin,100 100) is not the same as (sun←↓2 2⍴origin, 100 100) then (,/sun) ? they both have length 4
So there's that question, but then generally, I'm interested in understanding better APL writing. In this example, I have lms∘attract¨join lms I was trying to do a ∘.{⍺ attract ⍵} ⍨join lms or something to that effect, but I don't know if my current implementation may be faster? since I'm applying ...¨join lms to lms all at the same time?
So there's the question of efficiency, but that a a⍴... bug where a must be > 3 or the balls disappear is gnawing at me
sorry @dzaima I'm really enjoying using your app thing, so I'm not complaining
except that whole no vector of functions thing... but elsewise, I'm just trying to get things working, I don't mind a few bugs, it's all water under a ducks back
