Cholskey's Inversion Error

Discussion of MicroSurvey CAD related issues and questions.

Moderators: Brian Sloman, Jason Poitras, James Johnston

Post Reply
Posts: 1
Joined: Mon Apr 04, 2005 7:22 am

Cholskey's Inversion Error

Post by ejm » Mon Apr 04, 2005 7:33 am

I'm trying to use the LSA routine, when it starts to process an Error window comes up with "Cholesky's Inversion Error". I've had alook through the data and there's no obvious problem's. Any idea's with what's causing this?

Glen Cameron
Posts: 1395
Joined: Fri Nov 08, 2002 12:18 pm
Location: Corbeil, Ontario, Canada


Post by Glen Cameron » Mon Apr 04, 2005 7:51 am

When the data is being inverted in the matrix, it is running into a zero value on the bottom (which is invalid of course). This tells me that there may not be enough data in the job to have enough redundancy, or there may be some invalid data.
You may not be able to do a least squares adjustment on this data, as it is right now, in our program. You may need to use another adjustment option.

Glen W. Cameron, C.E.T.
City of North Bay, Ontario

Posts: 139
Joined: Sun Nov 24, 2002 6:45 am
Location: Temecula, California

Post by ianw2 » Mon Apr 11, 2005 6:29 am

Cholesky inversion is a technique of processing matrices in solving a least squares network. If the inversion of the matrix yields zero, it implies that there is no correlation between the data.

All that is a fancy way of saying that one or more of the points pretty well “blows chunks”. Typically, because of the massive level of garbage needed to cause an inversion error in the matrix, the problems will be with a major control point or two.

If you are using GPS, is there a massive reflective glass window or a billboard or a metal building or surface near one or more of the points? Mulipath error could cause the kind of error that would result in a matrix inversion error. A major part of the sky being cut of (without a lot of multipath) would be another source.

With a total station network, the only time I’ve ever seen this is when field techniques were so bad that the work was useless. For example, tribrachs not being leveled properly, guns not over points, rough sighting of the target without proper focus, etc. Of course, each of the offending points would have to have multiple set-ups on it to cause even this level of problem.

Unless there’s a glitch in the Least Squares Adjustment routine in MicroSurvey (and I haven’t seen one), the data almost has to be so bad to cause a matrix inversion error that nothing will resurrect it.

I’m thinking you’ve got someone else collecting this data and they weren’t briefed properly or knew how to set up the work. Perhaps a new instrument person or someone newly promoted?
Ian Wilson
Ian Wilson Land Surveying, Inc.
Temecula CA 92591
(951) 684-1044

Posts: 11
Joined: Thu Dec 25, 2008 1:37 pm

Least Squares Adjustment-Cholskey`s Inversion Error

Post by CARTOGRAF » Thu Mar 19, 2009 6:30 am

I`m having a problem with the LSA routine,tried to run my traverse and it comes up with the Cholskey`s Inversion Error.The interesting thing is,that i previously adjusted the same traverse with another program using the LSA method and the result was OK.
I used Sokkia total station.
I can think about one thing: most of the time a use orbitrary coordinates and on my first setup station i orientate to north - only an angle direction of zero,recorded as an azimuth.I can`t re-coordinate the traverse with this north azimuth,because it`s asking me to enter the coordinates of my first backsite.To keep it north i entered N:11000,E:1000,Elevation:0.After the re-coordination it looks fine,but failing to run the LSA.
Any ideas?

Post Reply