Using 1 dimension of a multidimensional wave as a 1D wave

Posts: 11
Joined: 2017-06-12
Location: United States

I have several multidimensional waves that I sometimes need to analyze by dimension.

For example, I would like to do

 LOESS/CONF={.8, WAVENAME[%FIRSTROWLABEL][%FIRSTCOLUMNLABEL][%SECONDLAYERLABEL][]}/R=1/TIME=45/V=6/Z=1 WAVENAME[%FIRSTROWLABEL][%FIRSTCOLUMNLABEL][%FIRSTLAYERLABEL][] </IGOR> smooth the first layer using the second layer as the confidence interval. Since I've held the row, column, and layer constant in each case, what's left is a single dimension 'slice' of the wave. However, instead of completing the line, I get an error message that highlights the first <igor> % 
and reads "Expected a name."

The manual seems to say contradictory things about how to make a multidimensional wave behave like a 1D wave (using p and q are required; using p and q are not allowed) and either way from the example in the manual ( DisplayHelpTopic "Treating Multidimensional Waves as 1D" ) start/end is not supported in functions like Loess.

Interestingly, though DISPLAY/M/W=(1,13,52,29) WAVENAME[%FIRSTROWLABEL][%FIRSTCOLUMNLABEL][%FIRSTLAYERLABEL][] works just fine. Since the concepts are similar, it seems like the Loess function should work with this nomenclature, yet it doesn't. Why?

[ last edited September 10, 2017 - 18:01 ]
JimProuty's picture
Posts: 619
Joined: 2007-07-19
Location: United States

Only a few of the commands have this ability to treat a row or column of a multi-dimensional wave as a 1D wave.

The commands that were coded before Igor had multi-dimensional waves tend to not support this feature.

It was important for Display to handle the idea, other commands haven't been retrofitted.

The Duplicate command is your friend.

--Jim Prouty
Software Engineer, WaveMetrics, Inc.

Back to top