Manipulating f(x) may result in discontinuities

pixelzoom commented 1 year ago

In https://github.com/phetsims/qa/issues/898#issuecomment-1437631257, @liammulh reported the issue below. I'd be surprised if it's specific to iPad. I have not been able to reproduce on any platform, and don't have any hypothesis about the cause. This is more in @veillette wheelhouse, so assigning to him to investigate.

These notes apply to:

Tester = LM
Device =
    (Seemingly has no name. Also has no cart sticker. Black with black case. Lion logo on the case.)
    iPad Model MR7F2LL/A
    iPadOS 16.3
    Safari 16.3 Mobile
Time = 1.5 hour

...

Integral Screen

With the area under curve checkbox checked, I've seen some weirdness. If you click buttons A and B, make a tall "high frequency" curve, then press button C, make the curve as vertical as possible, then click button D to move the curve to the top of the screen (or the bottom), then press button C again to mess around with the curve, you start to see some interesting dots. These dots move around as you move the curve, and they're not completely solid; you can see the background color coming through with some of them.

kathy-phet commented 1 year ago

These are probably just the dots that appear when the curve is discontinuous. I believe it is behaving as expected (with the current approach).

pixelzoom commented 1 year ago

f(x) is not supposed to be discontinuous.

veillette commented 1 year ago

Although @liammulh created a curve that has clearly no pedagogical purposes, I see that he is trying to create an odd curve to test the limits of this simulation.

For the first two screens, one should not be able to create a discontinuous plot. However, the algorithm to detect the discontinuities is heuristic and based on the jump between two adjacent points. So it is possible in principle to create a curve with discontinuities on the first two screens.

@liammulh : I'd be curious to know the process of creating a "tall high frequency curve" as I could not reproduce it.

liammulh commented 1 year ago

This bug is very hard to reproduce.

I am swamped with translation utility work, QA work, and miscellaneous sprint work, so I likely won't be able to comment further today.

veillette commented 1 year ago

In the end, I was able to reproduce discontinuities on demand following these steps.

Select shift mode and shift curve to the lowest possible value
Select tilt mode and tilt curve to maximum positive value
Set the slider to the narrowest width.
Select "hill" mode and create some hills.

veillette commented 1 year ago

One pleasant side effect of the fix proposed in #245, means that it is one cannot readily create a large tilt. The large tilt of the slope (seen near x=0 in the screenshot above) was an essential component to create high frequency. Repeating the steps in https://github.com/phetsims/calculus-grapher/issues/241#issuecomment-1452678153, yields a rather well behave function for f(x)

pixelzoom commented 1 year ago

@veillette and I discussed in 3/7/2023 pairing. @veillette has made this problem impossible to create, so we're going to close this issue.

phetsims / calculus-grapher

Manipulating f(x) may result in discontinuities #241

Integral Screen