solving dense singular linear system

  • Permalink
  • submit to reddit
  • Email
  • Follow


Hello,
I am trying to solve a square system of linear equation Ax=b, with A
singular. Any suggestions about possible MATLAB built-in routines or
user-contributed routines are welcome.

Is it possible to use any of lsqr, cgs, gmres methods? do these give
accurate solution, is any method better than the others.

As I was reading, I came across the topic of matrix deflation and
inflation, anybody know about MATLAB resources related to this topic?

There was also a suggestion to reduce the number of equation by 1 and
then making the matrix consistent. Any insight on this is very
appreciated.

Thanks,
0
Reply fdawoud (1) 5/29/2007 2:49:47 AM

See related articles to this posting

"Fady Dawoud" <fdawoud@dal.ca> wrote in message 
news:ef58558.-1@webcrossing.raydaftYaTP...
> Hello,
> I am trying to solve a square system of linear equation Ax=b, with A
> singular. Any suggestions about possible MATLAB built-in routines or
> user-contributed routines are welcome.
>
> Is it possible to use any of lsqr, cgs, gmres methods? do these give
> accurate solution, is any method better than the others.
>
> As I was reading, I came across the topic of matrix deflation and
> inflation, anybody know about MATLAB resources related to this topic?
>
> There was also a suggestion to reduce the number of equation by 1 and
> then making the matrix consistent. Any insight on this is very
> appreciated.
>
> Thanks,

HELP MLDIVIDE

-- 
Steve Lord
slord@mathworks.com 


0
Reply slord (13686) 5/29/2007 3:44:07 AM

Fady Dawoud wrote:
>
>
> Hello,
> I am trying to solve a square system of linear equation Ax=b, with
> A
> singular. Any suggestions about possible MATLAB built-in routines
> or
> user-contributed routines are welcome.
>
> Is it possible to use any of lsqr, cgs, gmres methods? do these
> give
> accurate solution, is any method better than the others.
>
> As I was reading, I came across the topic of matrix deflation and
> inflation, anybody know about MATLAB resources related to this
> topic?
>
> There was also a suggestion to reduce the number of equation by 1
> and
> then making the matrix consistent. Any insight on this is very
> appreciated.
>
> Thanks,

x = pinv(A)*b

HTH,
John
0
Reply woodchips (7941) 5/29/2007 10:41:05 AM
comp.soft-sys.matlab 201246 articles. 513 followers. Post

2 Replies
38 Views

Similar Articles

[PageSpeed] 36

  • Permalink
  • submit to reddit
  • Email
  • Follow


Reply:

Similar Artilces:

Solve singular linear system
Hi, I'm trying to solve a linear system: Ax = B, A is a given n by n matrix It is known that A is singular, and if I give x1=a, x2=b then the solution is unique. How do I solve Ax = B provided x1=a and x2=b in matlab? Thank you. "Haoran " <chr@cms.zju.edu.cn> wrote in message <k43000$rkg$1@newscl01ah.mathworks.com>... > Hi, I'm trying to solve a linear system: > > Ax = B, A is a given n by n matrix > > It is known that A is singular, and if I give x1=a, x2=b then the solution is unique. How do I solve Ax = B provided x1=a and x2=b in ...

solving sparse singular (linear) system
I seem to remember that in previous version of matlab the backslash operator would try to give solutions to singular systems...I know I could just knock out the zero rows and solve that non-singular system but I kind of (lazily) expect there to be a one-line way to do this in matlab. Also, I am actually applying it to large sparse singular matrices so pinv (pseudoinverse) is no good. thanks ... >> A = [ 0 0; 1 0] A = 0 0 1 0 >> b = [ 0 1]' b = 0 1 >> pinv(A)*b ans = 1 0 >> A\b Warning: Matrix is singular to working precision...

Which method is used in MATLAB to solve square singular linear systems ?
Hello All. My question is in the subject. I wonder which numerical technique is used in MATLAB to solve square systems with singular coefficient matrix. Take e.g. A=[ 10 10 1 -18 10 20 18 -34 1 18 29 -29 -18 -34 -29 58]; b=[50 ; 100 ; 90 ; -170]; This 4x4 system has rank 2 and MATLAB returns the solution >> A\b Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 9.148276e-019. ans = 16.8889 -2.6667 9.8467 5.6705 This is not the least-squares solu...

Solving a recursive system of 3x3 linear systems...
Hi everybody, I'm looking at a recursive system where at each l-th step the three variables a[i,l] (1 <= i <= 3) have to be solved from a system of three linear equations involving the previously solved a[i,l+1], ...., a[i,k] (1 <= i <= 3) (so the iterator l runs back, starting from some value k). My goal is to find a closed form "direct formula" for the a[i,l], that is a formula that expresses each a[i,l] in terms of the starting values a[i,k]'s and the the other known constants involved. More precisely, the code for finding the first few s...

Solve Linear system with non-linear constrain
Hello! I would like to solve the following linear system: Ax = b, (where A is symmetric) but I would like to force the absolute value of x to be 1. That is <x,x>=1 How can I do that? Thanks a lot, and sorry: I am a newbie in optimization. "Max Brega" <matped@tiscali.it> wrote in message <hb4r52$5uj$1@fred.mathworks.com>... > Hello! > I would like to solve the following linear system: > Ax = b, (where A is symmetric) > > but I would like to force the absolute value of x to be 1. That is <x,x>=1 > How can I do that? > Thanks a lot, an...

Solving a recursive system of 3x3 linear systems... #2
Hello, apparently my previous email with the same subject wasn't readable since it contained a bitmap. Here's the text again with a link to the bitmap file to be viewed in a browser, hopefully this is better: Hi everybody, I'm looking at a recursive system where at each l-th step the three variables a[i,l] (1 <= i <= 3) have to be solved from a system of three linear equations involving the previously solved a[i,l+1], ...., a[i,k] (1 <= i <= 3) (so the iterator l runs back, starting from some value k). My goal is to find a closed form "direct ...

solve linear system
Dear all, I'm new to Matlab and so probably my question will be 'stupid' for some of you. I have an ODE system and I want to find the steady state (i.e. dx/dt=0). I think the way to achieve this task is using the symbolic toolbox and solve, but even with a reduced system, I can't get the solution for one of the variable, why? >> syms xa xb; eq1=-ReversibleMichaelis_Menten(xa, xb, p(2), p(3), p(4), p(5))*p(1)+ConstantFlux_irreversible_(p(7))*p(1) eq2=ReversibleMichaelis_Menten(xa, xb, p(2), p(3), p(4), p(5))*p(1)-ConstantFlux_irreversible_(p(6))*p(1) [xa xb]=solve(eq1,eq2...

Linear System solving
I have the system of equations Ax=b.How can I solve it for A if I know the x and b matrices?? And can I use the lscov() function too?? Hi Betty, How many rows and columns in the x and b matrices? Thanks. Rick "Betty Vag" <betty00arg@hotmail.com> wrote in message news:ievopg$1l9$1@fred.mathworks.com... > I have the system of equations Ax=b.How can I solve it for A if I know the > x and b matrices?? > And can I use the lscov() function too?? > On Dec 23, 10:10=A0am, "Betty Vag" <betty00...@hotmail.com> wrote: > I hav...

solving a linear system
Hi everyone, I have a linear equation of the form Ax = b which I solve for every pixel in an input image by the brilliant matlab \ operator. Right now, I am looping and solving for every pixel. The matrix A does not change from pixel to pixel. Is there a faster, better way to solve this rather than a loop through each pixel? For example, I looked at the LAPACK documentation for functions like sgels, dgesls etc. and it says the following: "Several right hand side vectors b and solution vectors x can be handled in a single call; they are stored as the columns of the M-by-NRHS right hand...