logarithmic decimator?

[labuda]

Hi,

I am displaying an array of 1 million points (linearly spaced) on a graph with a logarithmic axis. That's a lot of points, and my program slows down consequently. Also, the points bunch up on the right side of the scale, so i can't see multiple graphs properly.

Solution: I would need a logarithmic decimator, to take a subset array from the original, with logarithmically spaced data points. That way, my data will look evenly spaced on a logarithmic graph.

Anyone have a subvi which will do this? Any suggestions for an algorithm that will do this?

Thanks

aleks.

[falkpl]

do a for loop log10(array size) get 10^i item and index it out.  This should index out 0, 10, 100,1000,10000, 100K and 1M element and run very fast.  I think this should work.

Paul

[labuda]

I've effectively implemented the same thing in a subvi just a moment ago.

It's  a while loop which runs, and creates an array of indices to remove from the original array. The algorithm is

INDEX_TO_REMOVE = LAST_INDEX_REMOVED - round(1.01^i)

LAST_INDEX_REMOVED starts at ARRAY_SIZE

1.01 is a fraction to be fine tuned

"i" is the while loop index.

The part which takes a long time is remove the entries one by one... i still need help with that.

I guess what you suggested simply keeps entries, whereas i've been removing them. I'll try to reverse my method, to reduce the time it takes.

Thanks   

[falkpl]

Here is what i was thinking, will give you the the solution in O(Log(n)) time very fast. 

[falkpl]

Oops forgon thew attachment