Limit and continuity of a multivariable function

The basic concept remains the same. However, with 2D things are more complicated than in 1D. In 1D you either walk forwards or backwards. In 2D we can circle around a point, meaning that, sometimes, the limit may not exist in one direction while it does exist in another. In 3D think about optical illusions. An object may appear continuous from one angle, yet it's discontinuous from another. One such example is the Penrose triangle.

For one variable we take one step to the right or one step to the left. For two variables we can take one step up or down, in addition to left and right. If we can take one small step in any direction on a plane what we are describing is a circle. Going to the 3D space and we have a sphere.

We have an equation of a circle in 2D and the equation of a sphere in 3D. The equation for the circle is: [math]\displaystyle{ (x - x_0)^2 + (y - y_0)^2 = \delta^2 \iff \delta = \sqrt{(x - x_0)^2 + (y - y_0)^2} }[/math] (we aren't interested in a negative radius, we can disregard the negative root). Suppose that [math]\displaystyle{ P = (x,y) }[/math] is located anywhere in that circle, including the circle's perimeter. If you know how to calculate the distance between two points from analytical geometry you are going to notice that we just wrote it. With [math]\displaystyle{ \delta }[/math] being the radius, [math]\displaystyle{ (x_0, y_0) }[/math] the circle's origin and [math]\displaystyle{ P = (x,y) }[/math] any point inside that circle.

Notice how the figure is also a graphical depiction of the property: [math]\displaystyle{ |a - b| = \sqrt{(a - b)^2} }[/math]. Distance cannot be negative. We can view the coordinates of the points as displacement vectors. Both points being displaced from the origin of the Cartesian plane to their positions shown in the graph. The radius of that circle can also be interpreted as [math]\displaystyle{ ||\overrightarrow{P} - \overrightarrow{C}|| = \delta }[/math], where [math]\displaystyle{ \overrightarrow{C} }[/math] is displacement vector for the circle's origin.

Notation is really the same idea from limits of single variable functions:

[math]\displaystyle{ \lim_{(x,y) \to (x_0,y_0)} f(x,y) = L }[/math] (same for any number of variables)

For each [math]\displaystyle{ \epsilon \gt 0 }[/math], there is a [math]\displaystyle{ \delta \gt 0 }[/math], such that every [math]\displaystyle{ (x,y) \in D_f }[/math], [math]\displaystyle{ 0 \lt \sqrt{(x - x_0)^2 + (y - y_0)^2} \lt \delta \implies |f(x,y) - L| \lt \epsilon }[/math]. (some textbooks replace the square root with a norm and difference between the coordinates, it's really the same thing)

The concept is virtually the same used for single variable functions. We are considering the smallest distance between two points in 2D that is as close as possible to zero. While the error, the distance between the image and the limit, is the lowest possible value. Note that the definition of a limit for many variables is not considering the path to the point. The concept of one sided limits for many variables is a bit more complicated because when we expand to 2D and 3D there is up and down, front and back, there are way more sides and directions to account for.

Terminology: when a limit does exist at a point [math]\displaystyle{ (x_0, y_0) }[/math], that point is called a limit point or accumulation point.