#15820: Implement sequences of bounded integers
--------------------------------------------+------------------------
Reporter: SimonKing | Owner:
Type: enhancement | Status: new
Priority: major | Milestone: sage-6.2
Component: algebra | Resolution:
Keywords: sequence bounded integer | Merged in:
Authors: | Reviewers:
Report Upstream: N/A | Work issues:
Branch: | Commit:
Dependencies: | Stopgaps:
--------------------------------------------+------------------------
Comment (by SimonKing):
Replying to [comment:25 ncohen]:
> Hmmmm... I see. Well, isn't your only parameter the number of bits you
need to store each entry ?
No. We have
{{{
#!c
cdef Integer NumberArrows
cdef Integer IntegerMask
cdef size_t ArrowBitSize, BigArrowBitSize
cdef block* bitmask = NULL
cdef block DataMask, BitMask
cdef size_t ArrowsPerBlock, BigArrowsPerBlock
cdef size_t ModBAPB, DivBAPB, TimesBABS, TimesBlockSize
}}}
For example, `ModBAPB` is used to replace `n%BigArrowsPerBlock` by a bit-
wise `and`, which is noticeably faster.
> If it is, did you decided whether you want to allow it to be arbitrary,
or if you want it to be a power of 2 ?
Depending on the implementation...
> If it is a power of 2, then perhaps templates are sufficient, i.e. "one
data structure per value of this parameter". There are not so many powers
of 2 between 1 and 64 that make sense to store something `:-)`
Isn't duplication of code supposed to smell?
> If your data structure is not an object, you could also add this width
parameter as an input of the functions you use ? I guess all Bounded
Sequence Integer that you use in the same function would have the same
bound.
Do I understand correctly: You suggest that we should ''not'' provide a
class `BoundedIntegerSequence` that can be used interactively in Sage, but
you suggest that we should just provide a couple of functions that operate
on C structures (say, `mpz_t` or `usigned long*`) and can only be used in
Cython code?
Then it would be better to revert this ticket to its original purpose:
There would be...
1. ... cdef functions operating on `mpz_t` resp. on `unsigned long*`,
2. ... a Cython class `Path` using `mpz_t` resp. `unsigned long*` as a
cdef attribute, using the afore mentioned functions to implement
concatenation and iteration,
3. ... a subsequent ticket, implementing an F5 style algorithm to compute
standard bases, operating not with `Path` but with `mpz_t*` resp `unsigned
long**` (which will be the case anyway).
> But admittedly I love to split hairs. Tell me if you think that this is
going too far `:-P`
No problem. I just want to know if people see the need to have a Cython
implementation of sequences of bounded integers, for use in Python.
> Ahahahah. That's up to you. I just hate Sage's parents/elements. I would
live happy with a C pointer I can control, but not witha framework that I
do not understand.
Making it something like a container rather than a fully fledged parent
would be an option, from my perspective.
--
Ticket URL: <http://trac.sagemath.org/ticket/15820#comment:26>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/groups/opt_out.
