What exactly are the dp_fct mathematics and common names

[tmalone]

The DP_FCT config contains some common calculations, but some that I just can't find the common names. I am guessing that maybe the German docs are better than the English. I apologize as it was a few decades ago since I took higher level math and don't know the names. Can someone enlighten me on the following:

Integral 0 vs. Integral 1

Mean vs. Mean 0 vs. Mean 1

yes, I see the descriptions in the help, not much help as I have google-ed that and found no references that are exactly that description. Please don't cut-paste the help for me. I also found some rough math equations in the help for DYN array functions which seemed to give the formula for Mean 0 and Mean 1, but I don't recognize that either.

For example, is one of these Standard Deviation? Is that calculation possible with a combination from this list of dp_fct?

Is one of these "running average"? Exponentially weighted moving average? OMG!

Thanks,
Todd Malone
HMI CoC

[Gertjan van Schijndel]

Mean -> Average
Mean 0 -> Weighted average (for when there are many samples lost due to old/new comparision/smoothing)
Mean 1 -> Weighted average with interpolation between the samples

I do not see how you could calculate the standard deviation.

By setting the 'floating calculation' you could get a running/moving average.

[tmalone]

I found an obscure formula about the mean0 and mean1. If you look at the help for dynAvgWt(), you will find some interesting formulas for mean calculations. I am guessing that under the covers the dp_fct actually use these same calculations. (sorry but impossible to put this formula in this message, and it is not a good pic in the help!)

The description does not make a lot of sense in English ("penultimate", OMG) and looking at the actual equation, I really doubt that if the values are decreasing that the calculation fails? What I get from the formula is this:

Because OA logs only on change, a weighted average 0 is one in which the time a value stays the same has more weight if the time is longer than the others with shorter duration at that their value.

The difference between 0 and 1 seems to be that in the 1st order weighting, that it takes into account the change between the values in addition to the time, thus giving more weight to a larger jump or fall in value in addition to just the time spent at a particular value.

This really is very clever (must have been a CERN thing). These different ways to calculate mean give the user a lot of tools. The docs need an update in this area.

Here are my definitions for these terms
mean = average
mean0 = time weighted average
mean1 = time weighted average with interpolation
integral0 = area under stepped curve
integral1 = area under interpolated curve

Todd Malone
HMI CoC USA