4200 by 3 matrix has "overflow error" on input

[gwhowell]

   _       _ _(_)_     |  A fresh approach to technical computing
  (_)     | (_) (_)    |  Documentation: http://docs.julialang.org
   _ _   _| |_  __ _   |  Type "help()" for help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 0.4.0-dev+5040 (2015-05-27 08:31 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit cfbd68c (0 days old master)
|__/                   |  x86_64-unknown-linux-gnu

julia> include("b1ss.m")
15x3 Array{Float64,2}:
 5.0  1.0  -0.0359994 
 6.0  1.0  -0.0176371 
 7.0  1.0  -0.00772178
 1.0  2.0   1.0       
 2.0  2.0  -1.0       
 1.0  3.0   1.0       
 3.0  3.0  -1.0       
 1.0  4.0   1.0       
 4.0  4.0  -1.0       
 2.0  5.0   0.45      
 5.0  5.0   1.0       
 3.0  6.0   0.1       
 6.0  6.0   1.0       
 4.0  7.0   0.45      
 7.0  7.0   1.0       

julia> include("b2ss.m")
ERROR: LoadError: OverflowError()
 in include at ./boot.jl:252
 in include_from_node1 at ./loading.jl:133
while loading /home/howell/howell/progs/matdirs/simpleLUjulia/b2ss.m, in expression starting on line 1

---

Here's the file b1ss.m

a = [ 5 1 -.03599942 ; 
6 1 -.0176371 ; 
7 1 -.007721779 ; 
1 2 1 ; 
2 2 -1 ; 
1 3 1 ; 
3 3 -1 ; 
1 4 1 ; 
4 4 -1 ; 
2 5 .45 ; 
5 5 1 ; 
3 6 .1 ; 
6 6 1 ; 
4 7 .45 ; 
7 7 
] ;

---

the file b2ss.m is similar (but has around 4228 rows in the matrix )

too big to drag and drop .. so posted it online

http://ncsu.edu/hpc/Courses/CSC302/b2ss.html

Note,
julia>include("b222.m")
is fine in 0.3.7

0.3.7 was from
yum install julia
in fedora 19,

The 0.4.0 is today's generic linux binary (fedora 20 )

Thanks,
Gary Howell

[gwhowell]

include("b222.m") was a misprint, should be include("b2ss.m")

[andreasnoack]

These is a typo in the link at http://ncsu.edu/hpc/Courses/CSC302/b2ss.html

[timholy]

@gwhowell, I added backticks for you with your post, it makes it much more readable.

In b1ss.m, is it really true that the final row contains only 2 numbers? How does that final 1 get in there?

[andreasnoack]

For matrices of that size, I think you should read the numbers from a file. Is it a sparse matrix in triplet format? If you add the dimensions and the number of lines in the top you can read it with

A = sparse(Base.SparseMatrix.CHOLMOD.Sparse(b2ss.m"))

[JeffBezanson]

Probably same issue as #11449

[gwhowell]

I fixed the link.
Yes, it is a triplet matrix for a sparse matrix (produced from Davis
collection by a perl
script).

On Wed, May 27, 2015 at 3:36 PM, Andreas Noack [email protected]
wrote:

These is a typo in the link at
http://ncsu.edu/hpc/Courses/CSC302/b2ss.html

—
Reply to this email directly or view it on GitHub
#11458 (comment).

[gwhowell]

Yes, there should be a 3rd entry in the final row (failed cut and paste).

On Wed, May 27, 2015 at 3:37 PM, Tim Holy [email protected] wrote:

@gwhowell https://github.com/gwhowell, I added backticks for you with
your post, it makes it much more readable.

In b1ss.m, is it really true that the final row contains only 2 numbers?
How does that final 1 get in there?

—
Reply to this email directly or view it on GitHub
#11458 (comment).

[gwhowell]

Hi Andreas,

converting b1ss.m to b1ss.jl

7 7 15
5 1 -.03599942 ;
6 1 -.0176371 ;
7 1 -.007721779 ;
1 2 1 ;
2 2 -1 ;
1 3 1 ;
3 3 -1 ;
1 4 1 ;
4 4 -1 ;
2 5 .45 ;
5 5 1 ;
3 6 .1 ;
6 6 1 ;
4 7 .45 ;
7 7 1 ;

and using
A = sparse(Base.SparseMatrix.CHOLMOD.Sparse("b1ss.jl"))

did work fine (and similarly for b2ss.m and b2ss.jl)
Thanks,
Gary

On Wed, May 27, 2015 at 3:44 PM, Andreas Noack [email protected]
wrote:

For matrices of that size, I think you should read the numbers from a
file. Is it a sparse matrix in triplet format? If you add the dimensions
and the number of lines in the top you can read it with

A = sparse(Base.SparseMatrix.CHOLMOD.Sparse(b2ss.m"))

—
Reply to this email directly or view it on GitHub
#11458 (comment).

[JeffBezanson]

Closing as dup of #11320.

[gwhowell]

HI Andreas,

I haven't been able to figure out how to reconstruct sparse A from lufact(A).

The formula give by help(lufact) does not appear to work

F[:L]F[:U] == F[:Rs] . A[F[:p], F[:q]]

but

julia> A = sparse(Base.SparseMatrix.CHOLMOD.Sparse("b1ss.jl"))
7x7 sparse matrix with 15 Float64 entries:
[5, 1] = -0.0359994
[6, 1] = -0.0176371
[7, 1] = -0.00772178
[1, 2] = 1.0
[2, 2] = -1.0
[1, 3] = 1.0
[3, 3] = -1.0
[1, 4] = 1.0
[4, 4] = -1.0
[2, 5] = 0.45
[5, 5] = 1.0
[3, 6] = 0.1
[6, 6] = 1.0
[4, 7] = 0.45
[7, 7] = 1.0

julia> F = lufact(A)
UMFPACK LU Factorization of a 7-by-7 sparse matrix
Ptr{Void} @0x0000000002c85540

julia> F[:L]F[:U] - F[:Rs] . A[F[:p], F[:q]]
7x7 sparse matrix with 13 Float64 entries:
[1, 1] = 0.631918
[6, 1] = -0.131856
[2, 2] = 0.293013
[5, 2] = -0.00561606
[3, 3] = 0.0832465
[7, 4] = -0.659004
[5, 5] = 0.0561606
[7, 5] = -0.659004
[6, 6] = 0.293013
[7, 6] = -0.659004
[1, 7] = -0.0227487
[2, 7] = -0.00516791
[3, 7] = -0.000642811

Also I'd like a way to get all 3 of the following

just row pivoting,

just row and column pivoting

-- row and column pivoting with row scaling ..

In matlab or octave, these options can be acomplished by a combination of
how many arguments specified on the left hand side, e.g. to get LU
factorization with
only row pivoting

[L,U,P] = lu(a,"vector")

(vector specifes that P is a permutation vector, else P would be the matrix
such that

L_U = P_A

This (vector P) happens to be the option I was attempting to use.

I'm willing to help on this, but am not very good at julia.

Regards,
Gary Howell

The b1ss.jl file is

7 7 15
5 1 -.03599942 ;
6 1 -.0176371 ;
7 1 -.007721779 ;
1 2 1 ;
2 2 -1 ;
1 3 1 ;
3 3 -1 ;
1 4 1 ;
4 4 -1 ;
2 5 .45 ;
5 5 1 ;
3 6 .1 ;
6 6 1 ;
4 7 .45 ;
7 7 1 ;

[howell@gary2 julia]$ julia --version
julia version 0.4.0-dev

I had also tried this with some versions of julia I downloaded a few months
ago.

On Wed, May 27, 2015 at 3:44 PM, Andreas Noack [email protected]
wrote:

For matrices of that size, I think you should read the numbers from a
file. Is it a sparse matrix in triplet format? If you add the dimensions
and the number of lines in the top you can read it with

A = sparse(Base.SparseMatrix.CHOLMOD.Sparse(b2ss.m"))

—
Reply to this email directly or view it on GitHub
#11458 (comment).

[andreasnoack]

There is a set of parentheses missing in the documentation. It should be F[:L]*F[:U] == (F[:Rs] .* A)[F[:p], F[:q]] such that the scaling is done before the permutations. I'll update the documentation.

The easiest way to get the extra pivoting options you are asking for is to try implement it. I can try to help with comments, but I will probably not do implement it. There are more important issues.

[gwhowell]

Thanks,
That worked ..

On Mon, Jul 13, 2015 at 12:33 PM, Andreas Noack [email protected]
wrote:

There is a set of parentheses missing in the documentation. It should be F[:L]F[:U]
== (F[:Rs] . A)[F[:p], F[:q]] such that the scaling is done before the
permutations. I'll update the documentation.

The easiest way to get the extra pivoting options you are asking for is to
try implement it. I can try to help with comments, but I will probably not
do implement it. There are more important issues.

—
Reply to this email directly or view it on GitHub
#11458 (comment).