Michael Chen wrote:
> --this is my first version for the purpose of storing sparse numerical
> matrix in sql
> --please let me know how to fix the bug at the end of the file, and how to
> tune the performance
> --or any better reference, thanks!
>
> .explain--
yes, that's what I am thinking of too. This big table is in charge of store
all matrix, keep track of all index changes, and rollback when needed. I
will only extract a tiny part from this big table in format like a sparse
matrix, and put it in C array, then the available numerical routines, such
On Fri, Oct 16, 2009 at 1:41 PM, Michael Chen
wrote:
> rdbms is indeed not a place for store a single sparse matrix like this.
> However I have hundreds of them, and I need to break them and recombine them
> frequently; furthermore, I need to drop a few rows or
rdbms is indeed not a place for store a single sparse matrix like this.
However I have hundreds of them, and I need to break them and recombine them
frequently; furthermore, I need to drop a few rows or columns successively,
and need to be able to trace back what's a row's original index. I think
thanks Pavel !
On Fri, Oct 16, 2009 at 1:24 PM, Pavel Ivanov wrote:
> > -- IA = [ 1 3 5 7 ] // IA(i) = Index of the first nonzero element
> of
> > row i in A
>
> Why 4th element if A has only 3 rows?
>
> > create temp view rowwiseC as
> > select a1.rowid, sum(a2.ct) +1
Hi All,
Simon answered:
Here's a less gruesome version - no cases. I've given no thought to
performance comparisons.
Thanks for the two great solutions you posted.
Upon further investigation, those solutions assume that we want all
like occurrences together, effectively sorting records
At 1:54 PM +1000 8/23/07, T wrote:
Hi Darren,
It seems to me that you have a flawed design.
Displaying sparse like that should be a function of your
application display code, not the database
I had to chuckle that when I asked "How do I use this to do that",
your solution was "you shouldn't
Hi Darren,
It seems to me that you have a flawed design.
Displaying sparse like that should be a function of your
application display code, not the database
I had to chuckle that when I asked "How do I use this to do that",
your solution was "you shouldn't have that and you should do it
Hi Simon,
Here's a less gruesome version - no cases. I've given no thought to
performance comparisons.
Thanks for the two great solutions you posted. They certainly achieve
the desired result with the simplified sample I gave. However, in the
broader reality, it doesn't quite satisfy my
Hi Tom,
Here's a less gruesome version - no cases. I've given no thought to
performance comparisons.
C:\Joinerysoft\JMS\dev\trunk> sqlite3 tst.db
SQLite version 3.4.0
Enter ".help" for instructions
sqlite> insert into List values( 'a' );
sqlite> insert into List values( 'a' );
sqlite> insert
Hi Tom,
Its a pretty gruesome bit of sql...
C:\Joinerysoft\JMS\dev\trunk> sqlite3 tst.db
SQLite version 3.4.0
Enter ".help" for instructions
sqlite> create table List( Code text );
sqlite> insert into List values( 'a' );
sqlite> insert into List values( 'a' );
sqlite> insert into List values(
It seems to me that you have a flawed design.
You should just have a 2 column database to begin with, with a table like this:
Code Count
a 4
b 2
c 3
Rather than plain inserting or deleting rows, just sometimes insert
or delete, you should instead increment or decrement
Hi all,
OK, this one has me stumped. I've been staring at it too long.
I have a table with a column like this:
Code
a
a
a
a
b
b
c
c
c
I want to output two columns, one showing the Code, another showing
the count of each code, but with a null wherever it was a repeat,
like this:
Code
13 matches
Mail list logo