# Multi-variable Continuity Applet Worksheet

To access the multi-variable continuity applet (based on work of Tom Banchoff) follow the link:
http://www.slu.edu/classes/maymk/banchoff/Continuity.html

(Before doing this worksheet, it would be wise to do the worksheet on continuity in 1 variable.  That worksheet is at:
http://www.slu.edu/classes/maymk/banchoff/Continuity1VarWS.html)

Spread the three graphs out where you can see them all, as well as the control panel.

The applet opens with the graph of the function f(x,y) = .5*(x^2-y^2)/(x^2+y^2), a rational function which is obviously continuous except at the origin. (This function is easier to understand in polar coordinates where it is cos(2θ)/2.) We will start at an easy point and me to harder points. The applet has an f(x,y) graph window, a close up graph window, and a domain window.  Visually the graph of the function in an δ-neighborhood of (x0, y0, f(x0, y0)) is a green patch. Two pink lines are drawn ε above and below the point. The definition of continuity says that no matter how small a value of ε we are given, δ can be chosen so that the green patch is trapped by the pink disks.

1) The example starts with f(x,y) = .5*(x^2-y^2)/(x^2+y^2), x0 = y0 = 0.4, and ε =  δ = 0.1. Find a value of δ that works with this ε at this point.  (Make δ smaller until the green patch is between between the two disks.)
Now reduce ε to 0.05 and find a δ that works for that ε and that point.
Repeat the process with ε = 0.01.
Notice that we have zoomed in far enough for the green patch to look like a piece of the plane.  Conjecture a rule that will produce a good δ for any ε < 0.01.  (Find a ratio of   δ/ ε that always produces a good δ if ε is small enough for the graph to look linear.

2) Repeat the process above with a new point.  Set x0=0.2, y0=0.1.  Find values of δ that work at this point when ε = 0.1, 0.05, and 0.01.  Conjecture a rule that will produce a good δ for any ε < 0.01.

3) The applet won't let us examine the function at the origin directly since the function is not defined there.  Instead we will look at a sequnce of settings to conclude the divisior needed to turn an epsilon into a good delta grows without bound as you approach the origin.
On the Close-up window, change the view to "Up Z Axis".  If delta is small enough we will see only pink.  If delta is too big we will be able to see green.  We will look at the points (x0, y0) = (10^-n, 10^-n), where n = 2, 3, 4, ...
Set ε =  δ = 0.1.  For x0 and y0 enter 1E-2.  You should see a lot of green.  Change  δ to 1E-2 and note that that is not small enough.  Change  δ to 1E-3 to see that that is small enough.  Change  δ to 2E-3, which is not small enough to be a good delta, but is close enough to see that a good delta exists.
Repeat the process with x0 = y0 = 1E-3 and   δ varying between 1E-3 and 1E-4.
Repeat the process with x0 = y0 = 1E-5 and   δ varying between 1E-5 and 1E-6.