Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.c++ > #121142

Re: Stange bucket-count for unordered_map

Show all headers | View raw

On 12/3/2024 1:30 PM, Ross Finlayson wrote:
> On 12/03/2024 12:25 PM, Chris M. Thomasson wrote:
>> On 12/2/2024 10:23 PM, Bonita Montero wrote:
>>> Am 02.12.2024 um 20:49 schrieb Ross Finlayson:
>>>
>>>> The prime rings make a pseudo-random or ring buffer?
>>>
>>> A prime bucket count could make a better equal distribution
>>> for poor hash-functions.
>>>
>>>
>>
>> For some reason I remember way back wrt using Cantor Pairing for a hash
>> function in a fun test. ;^)
> 
> Wow, next you'll invent arithmetic coding.

lol! :^) I was not trying to invent anything wrt using Cantor Pairing 
for a hash. It was just fun to me. I also know a lot about arithmetic 
coding, been a while, but think I can still do it from scratch. Iirc, 
was trying to relate it to some of my algorithms that store real data in 
the roots of complex numbers under reverse iteration of a Julia set. 
Fwiw, here is some more fun I had with Cantor Pairing:

https://youtu.be/XkwgJt5bxKI

Using it for music? ;^)


> 
> Of course everybody knows Huffman coding,
> because everybody learned the principles
> of Morse code, and the idea that the
> most occurring terms get the shortest
> codes in what's called entropy coding,
> is so obvious, that just because Huffman
> proved it was optimal, has it was already
> a given long, long, ago.
> 
> Once you know arithmetic coding then
> "a large vector-model language-model"
> is very much simplified. It's just a
> geometrization as after arithmetic coding,
> into the corpus.
> 
> Prime numbers with extended-precision make
> a very simple model of a multi-set, giving
> the terms the n'th prime by their occurrence
> rank, with 1 being an empty set and just
> multiplying to add and after trial division,
> or rather, a divisibility test, establishing
> membership and count.
> 
> There are lots of divisibility tests after
> approaches like "digit summation congruence"
> which is well-known in decimal as casting out 9's,
> besides that things like find-first-set count
> multiples of two, and so on.
> 
> Here it's just called "1-to-1 functions".
> 
> 1-to-1 and onto, usually.
> 
> Counting isn't memoryless.
> 
> 

Back to comp.lang.c++ | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Stange bucket-count for unordered_map Bonita Montero <Bonita.Montero@gmail.com> - 2024-12-02 17:30 +0100
  Re: Stange bucket-count for unordered_map Ross Finlayson <ross.a.finlayson@gmail.com> - 2024-12-02 11:49 -0800
    Re: Stange bucket-count for unordered_map Bonita Montero <Bonita.Montero@gmail.com> - 2024-12-03 07:23 +0100
      Re: Stange bucket-count for unordered_map "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com> - 2024-12-03 12:25 -0800
        Re: Stange bucket-count for unordered_map Ross Finlayson <ross.a.finlayson@gmail.com> - 2024-12-03 13:30 -0800
          Re: Stange bucket-count for unordered_map "Chris M. Thomasson" <chris.m.thomasson.1@gmail.com> - 2024-12-03 14:08 -0800
        Re: Stange bucket-count for unordered_map Bonita Montero <Bonita.Montero@gmail.com> - 2024-12-09 13:05 +0100

csiph-web