Multivariable calculus
Limit of sin(x^2+y^2)/(x^2+y^2) at the Origin
The limit of sin(x^2 + y^2) / (x^2 + y^2) as (x, y) approaches the origin is 1. The function depends on x and y only through u = x^2 + y^2, and u tends to 0 from above under every approach, so the two variable limit equals the one variable limit of sin(u)/u, which is 1.
The limit exists
| Path | Limit along it |
|---|---|
| along y = 0 | 1 |
| along y = x | 1 |
| along y = 3x | 1 |
Every path returns 1
Start with three lines of different slopes, to see whether the direction of approach changes anything.
Each of these is the same one variable fact in disguise, with a different constant multiplying . The constant cancels, which is a hint that the direction never entered the problem in the first place.
Substitute the single variable u = x^2 + y^2
The function depends on the point only through the combination . Write , so .
The substitution is legitimate for two reasons: is continuous, and exactly when , with at every point other than the origin itself. So the two variable limit is the one sided one variable limit.
Polar coordinates say the same thing even more plainly. Since , the function is with no anywhere in it. A function that does not mention the angle cannot depend on the direction of approach.
The mistake: reaching for L'Hopital in two variables
L'Hopital's rule is a theorem about functions of one variable. There is no two variable version, and when the input is a point there is nothing obvious to differentiate with respect to.
Students often differentiate top and bottom with respect to , getting , and land on the right answer by luck. The step is unjustified, and on a function that does not reduce to a single variable it will produce a confidently wrong number.
The honest route is the substitution above. It converts the problem into a one variable limit, and there L'Hopital, or the standard fact, applies for real.
Frequently asked questions
Is the substitution u = x^2 + y^2 always allowed?
It is allowed when the function genuinely depends on the point only through , and when precisely as . Both hold here. If a stray or survives the substitution, as in , the reduction is incomplete and you need a bound instead.
What is the limit of (1 - cos(x^2+y^2))/(x^2+y^2)?
It is . The same substitution gives , and since behaves like for small , the quotient behaves like and tends to .