griddata 'v4' interpolation

[systems_eng]

I'm using an equation node so that I can use the griddata('v4') method for interpolating an (x,y,z) Data set.  For the most part, it seems like the interpolation is generating the right points.  But in some cases I'm running into some unexplained results.

I'll plot a series of points (x,y,z), with the max z value equal to -60.  Generally, when I plot (x,y) points with a z value of -60 (depending on the values of the neighboring points) I would expect to get a value of a little above -60 or a little below.  Most of the time these results are produced.  Other times, in the same data set, I have points that have z values of -110, -115, and -90, that end up creating a new, interpolated z value of -70.  This jump in the z value makes me question whether or not I'm using the proper interpolation technique.

The data that I'm graphing should look like a hill, with a single peak, as the altitude decreases with distance.  Smaller hills may appear as the distance from the center peak grows, but there should never be a peak near the value of the maximum peak.

I've attached a vi that shows 3 views of the data:
1. Plot - location of points with z values attached
2. 2d Mesh - 2d view of data after mesh/interpolation
3. 3d Graph - 3d view of data after mesh/interpolation

Some of the reasons for this sort of error may be due to the clustered nature of some of the lower z values.  My only reasoning for these erroneous peaks is that these lower values are some how stacking together, which end up creating poor results.  Does anyone have any ideas about how to get around this?

[DuffmanSays]

From your description, I assume that you have a large set of points (x y z coordinates) that you then want to form a 3D surface. The problem is that the points that are interpolated between the points in your dataset are not what you expect? Am I correct that this is the goal of your interpolation? Unfortunately I am not quite sure of the interpolation algorithm used by the griddata ('v4') method. What I would suggest is looking closer at the details of this algorithm to determine whether or not these values generated actually are incorrect. Please post back if we can offer additional advice. Thanks.

Mike D.

[playmaker]

Hello how are you?

I am really in need of help at this time. I appreciate it if someone could help me as I am new to Labview and have no idea what I am doing.

I am trying to start with an input string...12345678901, and convert into 36 bit binary. In this example, there will be 34 bits and two zeros will be placed in front of the binary converted string to fulfill the full 36 bits. Each time there is a new input, I would like for the original input string to increment by 1.  So 12345678902->converted to binary.

Any ideas

Thanks

dae_carrington@yahoo.com

[altenbach]

Please don't randomly crosspost to unrelated threads! This is the third identical offtopic post! If you look above the LabVIEW forum, there is a link called "New Message". If this is confusing you, please explain the problem with the forum so it can be improved.

Please keep the discussion all in one place:

http://forums.ni.com/ni/board/message?board.id=170&message.id=237246#M237246

Message Edited by altenbach on 03-22-2007 06:43 PM

[playmaker]

dude i said i was new to this whole forum / labview thing. The only reason I am trying it out is because tech support is closed and the engineers go home at 6pm. I would appreciate it if you could help me out.

Thanks

[systems_eng]

Yeah I've looked online at matlab interpolation function here:

http://www.mathworks.com/access/helpdesk/help/techdoc/index.html?/access/helpdesk/help/techdoc/ref/griddata.html&http://www.google.com/search?hl=en&q=matlab+griddata&btnG=Search

This looked okay to me.  I think I've determined that the number of samples and the spacing between the samples really does matter in the final output.  I am plotting then interpolating between nearly 200 points.  There are a few other posts online that I'm going to continue to look at to see if there are any differences between the interpolations styles.

I'm still having a hard time understanding the two resultant peaks that are generated after I interpolate over the entire data set (both of which are nearly equal in value).  To try and isolate the problem, I interpolated over part of the data set, the portion which I believe to generate the 2nd erroneous peak.  But when I did this, I saw a peak intensity which is 20 values lower than what was generated with all the data (this is what I expect).  So, I really don't see anything which might indicate there is anything wrong with the interpolation.  I'll post again, if I figure out what's going on.  Thanks for the response.

[craigc]

Hi,

I have also been using griddata for a while now, when i first started using it i and some others saw some of the Anomolies that you mentioned.  There was a patch detailed here.

http://digital.ni.com/public.nsf/allkb/D97AF9ECD0236D6F862571D9005EA79F

Although when i read it back it says it is just to do with the help files but why are there vi's in the zip file?  Anyway i installed it and things started to look better.  Now i am stuck on the problem of running out of memory when using griddata on large data sets eeek!

Hope that helps

Craig