From theory to practice - page 556

[Evgeniy Chumakov]

Novaja:
Well, yes, it's beautiful.))

And before that there was such a deal.

[Alexander_K]

Evgeniy Chumakov:

Eugene - don't stall, open the signal.

The eager ones are waiting.

[Alexander_K]

Novaja:

Where do these distributions come from?

[Yuriy Asaulenko]

Alexander_K:

О!

I'll probably wait for the Automat to write an article about ACF and then I'll rush off again in pursuit of the healing Grail.

Yes, I too am wondering what new things Automat can tell you about ACF that aren't in the textbooks.

So you don't have to wait. Why don't you just read something? And go for the Grail.)

[Alexander_K]

Yuriy Asaulenko:
Yes, I too am interested in what new things the Automaton can tell us about the ACF that are not in the textbooks.

So you don't have to wait. Why don't you just do some reading? And go for the grail.))

Autocorrelation could help sort out the following junk (pardon the below portent):

I.e. VisSim has actually 3 ways of calculating the discrete autocorrelation estimate.

Which one should be used so that:

1. there are no conditions for applying tau=const. I'm interested solely and only in dynamic exponential tau.

2. there is an explicit criterion that for a given sample N the ACF is exponentially decreasing.

???

[Yuriy Asaulenko]

Alexander_K:

Which one should be used so that:

1. there are no conditions for applying tau=const. I am interested solely and only in dynamic exponential tau.

2. there was an explicit criterion that for a given sample N the ACF is exponentially decreasing.

???

1. There is no limit here. A single ACF point is calculated.

You want the whole thing, count for each value of tau.

2.ACF does not give such a criterion, it only states a fact. Why stick to the expectation?))

[Alexander_K]

Yuriy Asaulenko:
1. There is no limitation here. A single ACF point is calculated.

If you want the whole thing, count for each tau value.

2.ACF does not give such criteria, it only states the fact. What are you pestering me about expectation?)

Don't ruin my last hope, Yuri...

I believe in ACF like a child, that it's exactly what I need - a "trend/flop" parameter like Hearst.

I just can't apply it properly.

[[Deleted]]

Blessed are the believers.

;)))

[Yuriy Asaulenko]

Alexander_K:

Don't ruin my last hope, Yuri...

I believe in ACF like a child, that it's exactly what I need - a Hearst-like trend/flat parameter.

Just can't apply it properly.

Doesn't solve the issue.

An ordinary wizard will do all this better, faster and easier. In the trend sense.

But once you have solved this issue, the next ones will arise.) But they will arise anyway.)

[secret]

Alexander_K:

exponentially decreasing ACF.

We have to transform the data so that this condition is always satisfied.

I laughed for a long time) great joke, keep on laughing)

OK, you're basically right of course. Peasant control question, for three points: is such a transformation possible for SB?