256 dcdft fwhm as a measure of period error

[dbenn]

Okay, I've implemented FWHM @mpyat2 if you would not mind reviewing it.

Uncertainty results appear in the models dialog along with std errors now.

I have not yet added a button to show FWHM independent of model creation even though it is not tied to that, because it seems most useful in combination with the standard error of the frequency which is what the original context was.

[dbenn]

@mpyat2, here is one way to address your suggestions:

For frequency 0.060918525, period 16.415367836, power 262.29192674, semi-amplitude 0.397430612: 
  FWHM lower bound: 0.060650162
  FWHM upper bound: 0.061455252
  Standard Error of the Frequency: 0.000016567
  Standard Error of the Semi-Amplitude: 0.011405734

The FWHM and standard error metrics apply to a frequency, period, amplitude triple, or put another way: a row in the top-hits table. Do you agree?

[mpyat2]

Ok, probably the line 'For frequency 0.060918525, period 16.415367836, power 262.29192674, semi-amplitude 0.397430612: ' is sufficient, I would add 'for frequency' near FWHM to avoid possible ambiguity.

[dbenn]

Something like this, also spreading things out a bit?

For frequency 0.060918525, period 16.415367836, power 262.29192674, semi-amplitude 0.397430612:

  FWHM for frequency:
    Lower bound: 0.060650162
    Upper bound: 0.061455252

  Standard Error of the Frequency: 0.000016567
  Standard Error of the Semi-Amplitude: 0.011405734

[mpyat2]

Very good!

[dbenn]

Feel free to review the changes to src/org/aavso/tools/vstar/util/model/PeriodAnalysisDerivedMultiPeriodicModel.java @mpyat2.

[dbenn]

@mpyat2, I could also add the FWHM +/- delta, making it easier to compare with standard error of the frequency. There's a couple of ways this could be done:

• Half of the FWHM span: (upper-lower)/2 = (0.061455252-0.060650162)/2 = 0.0004025450 so:

  • +/- 0.0004025450 or simply 0.0004025450

• Or, next to lower and upper bound, frequency-lower = 0.060918525-0.060650162 and upper-frequency = 0.061455252-0.060918525:

  • Lower bound: 0.060650162 (0.000268363)
  • Upper bound: 0.061455252 (0.000536727)

As a sanity check, note that (0.000536727+0.000268363)/2 = 0.0004025450 which is the same as we get from (upper-lower)/2.

To reduce complexity, I tend to think simply 0.0004025450, e.g.

FWHM for frequency:
    Lower bound: 0.060650162
    Upper bound: 0.061455252
    Resulting error: 0.0004025450

Compare this against:

Standard Error of the Frequency: 0.000016567

What do you think?

[mpyat2]

Looks good, @dbenn

[dbenn]

@mpyat2 if you can review the latest commits, I think we're basically done with this issue

[mpyat2]

Hi @dbenn , I suspect you forgot abs(Resulting error)

[dbenn]

Try the latest commit @mpyat2

[dbenn]

Thanks @mpyat2!