#I forget how we found this problem
1324 messages · Page 2 of 2 (latest)
I didnt do it
But im rage quiting
I tried taking the list of tables and just like remove duplicates but its now even bigger number
If im coding ever again it will be in a purely functional programming lang
noo
LOLLLLLLLL
My code found 563 unique combinations for f(5 , 5)
I dont know how
I removed all duplicates and i dont think my code can give wrong combintions
I would have to code now a program that takes a combination , validates it ,and ....
@tough kettle is there some part of your code that takes a combination and gives like True if it really is and false if it isnt
Or like do you have that
It can be slow it would be really helpfull to debug my program with that
it doesn't work
well it doesn't work in c
i think it works in python
you generate some bitmasks and do some bitwise ops
f.py is somewhere up there
Whats the input?
i'm working on my algorithm rn
do you have f.py?
ok the thing inside the if inside the nested loops validates stuff
No but im asking what kind of structure does it take
My table is basically a list of white cells
lets say g is 0b10010110100101110, m is 4, n is 4
so the grid here is 1001 \n 0110 \n 1001 \n 0110
this is how my program achieves its speed lol
the 1s are white cells, the 0s are black cells
Hmm i think for my case i will just like write a loop that loops through whole table
And just does a check if any whites can be placed
It should be all false
Ok i found the extra ones
They are just wrong
The new problem now is that without those extra ones my program undercounts
For every larger n im just fixing my code
Thats how many flaws it has
First it did f(3 , 3) wrong and i fixed it. Then it did f(4 , 4) and i fixed it now f(5 , 5) is wrong
Im kind of asking myself is my algorithm even consistent or will i have to fix it for every laeger n
Oh i see the problem
I thought that this cell was always safe in my algorithm
So i didnt bother to check if white cell can be placed there
I think thats the last flaw in my algorithm
Ok i should prob rewrite my whole algorithm because its getting ugly
This problem has given me nothing other than pain and suffering

If my algorithm isnt polynomial time then this wasnt worth it
I mean on paper it souns fast. As everything you do will be to get a combination so there is no extra checking stuff
so it's supposed to be O(however many grids there are)?
Well it would seem to me that it would be something like that but idk in practice how would you calculate big O of that algo
I mean there are more operations done
that doesn't seem very fast on paper to me
I mean it is how a human would manually calculate this
i wouldn't
How would you?
the same way my algorithm does (which i've just been mysteriously mentioning a lot because i'm not sure if people want it spoiled yet)
Ok well imma try and finish this
My program is undercounting....
Again
How do i see a difference between 2 lists?
difference in what way?
Like [1 , 2 ,3 , 4] and [1 , 2 ,3] it would be [4]
python?
Yes
you could make them sets and set difference
might lose some information like the order though
and like if you want [1 , 2 , 3 , 3, 4] and [1 , 2 ,3] to return [3, 4] then that wouldn't work ofc
you could use Counter if you wanted that ig
Will try that prob
Even better
My algorithm has so many edge cased
I did it
It works

Damn this solved 7x7 in seconds and im on my phone
I like it more for scripting
Like i code on a computer only when its multiple file project
Like svelte or something like that
Can anyone speed test
Pastebin.com is the number one paste tool since 2002. Pastebin is a website where you can store text online for a set period of time.
Idk how
That.... is slow right?
Is it also written in python?
Well my time wasnt that well spent then ig
Is your algorithm pure python or like uses alot of functions from C?
but yea at least one operation for each valid grid is not going to be pretty limiting
just python
just simple python objects, no numpy or anything
Thats cool
I mean how does your algorithm scale
What about f(9 , 9) or something like thaz
f(n+1, n+1) takes about 3x as long as f(n,n). but f(2n,n) only takes (about) twice as long as f(n,n)
so as long as min(m,n) isn't too big, one of the numbers can go really high
like f(5,10^6) for example i've computed
So like f(n , n) has big O of 3^n?
yea
But im not checking every combination
Im only checking every combination that makes sense
Aka every combination that will give me a solution
but the number of valid grids still grows fast
polynomial in what?
And due to that being exponential then it probably also is
f(n+1,n+1) is at least 2*f(n,n) so that's already at least exponential for your algorithm
but it's much worse than that message makes it sound because the multiplier is much bigger than 2 (and grows)
Or maybe more like f(n , n) *n^6
f(8,8) still hasn't finished and i need to leave my computer
Yeah expected
Whats max you computed?
with my algorithm?
i don't think i've bothered going higher than 15
i could tell you f(19,19) but it would have to be later today 
So you got to 15. In how much time?
like 4 minutes probably?
i will compute this btw just for fun
and because it's past the oeis page
Damn...
@cunning stump hows your algorithm going
oh wow
nice
now its time to turn it into c
How is that even possible? It would take a supercomputer 1 day to permorfm ammount the number of calculations that is the solutions.
Im guessing previous calculated f(n , n) would be given as an input?
i don’t go through them all
and no i don’t use f(n-1,n-1) to assist in computing f(n,n)
So basically magic
pretty much
Are you like calculating some other sequence and then getting this sequence in terms of that?
it shouldn’t be that unbelievable though. to count the subsets of a size 100 set that are size 50 you don’t need to go through all 2^100 sets and check their size
So is your algorithm like lego pieces and then you just connect them?
kinda
not really
this is like 20 questions 😂
I mean it looks like you arent actually calculting every solution. But rather something else and then number of solutions is in terms of that thing.
there are sequences involved but that’s not really the main idea. they just happen to be made along the way
I mean by https://sequencedb.net/s/A288956
This sequence can be written in terms of other sequences
Which may be easier to compute
what exactly are you referring to?
well in any case i didn’t do anything related to anything on that page
but that is interesting 
Inverse of 1334th cyclotomic polynomial.
wtf
maam
u literally called yourself "no one"
in the sense that u said if someone looks you in the eyes they'll have "nothing" staring back at them
that implies self deprecation in the sense that you're nothing
Instead of f(n , n)
You compute $$F\left( \sum_{k = 1}^n (4k ) + 1\right)$$
Pluton
i didn’t do that but now i’m sad i won’t be at my computer for a while because i want to see how high i can go with that
What even is that seq
i love it when i see people that are smarter when it comes to maths then me
"Array read by antidiagonals: T(m,n) is the number of maximal independent sets in the m X n king graph."
so many weird english words to discover
it’s just this table https://oeis.org/A332347
but in a sequence
i may be able to abuse it
yea
then yea that’s useless lol
thats not self deprecating
elaborate
i know my worth .
your worth is nothing then
Take that back
im joking haha
not affirming anything
either way maybe u meant it in a more "I won't give you attention" way
like u wont look back at the person, i dunno
ok
It seems like the lcm one could be useful
maybe. i do fear now that they all do this though
just in convoluted ways
a(n)=lcm(A015343(n), A332347(A001844(n)))
Like if A001844(n) is less than sum of 4n + 1
Then it should be easier to compute maybe
And if A015343(n) is easier to compute
Yeah but like that one says f(n ,n ) is on f(n , m) this one says f(n , n) has some connection with f(n , m)
unrelated but i found this funni
And maybe possibly f(n , m) that is easier to compute than f(n , n)
my hopes are not very high
You have to compute A332347(2n(n + 1) + 1)
Now just because its a big number doesnt mean its hard to compute
what pairs (m,n) do those correspond to?
It could be f(2 , m) as far as we know
Wouldnt know
Wont be that hard to get i think
problem seems fun, i'll think about it
yea i’ve had a lot of fun with it
It coressponds to (n , n)
lol rip
so this was probably right
I guess its saying that the random thing divides that number
the fact that they all used the array read by anti-diagonals and that already has the sequence we want embedded inside it was sus
Yeah
easy to just grab the (n,n) parts out of it
lol. so taking lcm with 1 or -1 did nothing
But its 0 sometimes
not before 18 or whatever
And lcm( 0 , x) ....
Oh yeah
Does it mean a_(19) is undefined 
Makes no sense ik

A finite grid with non finite ammount of combinations
it means we can scrap that because it’s just coincidence
true
now you sleep
i think the closest i can get to what i mean is that im somewhere else
hmmmm
so like
head in the clouds?
like if someone stares at you you're not actually looking back cause you're distracted?
f(20,20) = 75332748233894465141809677114505805331037
What
just posting some numbers that go past the oeis page for f(n,n)
how long do these take to compute
owie
Goated
what does that mean
lol
run it using cpython not jupyter
actually run it with dandidapython
jupyter uses cpython
x
@tough kettle hi!!!!
is it ok if i participate here????
yeah
i call them solutions
oh ok
keepalive
so the thread doesn't close
same
i did like half my prob 5 at 3-4 am today so not it looks so jumpleb😭
lol ye
im doing prob 4 last omg its so hard to explain
no discus
ok
