Easy delta/epsilon proof of a multivariable limit

Click For Summary

Discussion Overview

The discussion revolves around the epsilon/delta proof for a multivariable limit, specifically the limit of the function \(\frac{xy}{x+y}\) as \((x,y)\) approaches \((1,1)\). Participants explore the steps necessary to establish the limit and address challenges in the proof process.

Discussion Character

  • Exploratory
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • One participant seeks assistance in proving the limit \(\lim_{(x,y) \to (1,1)}\frac{xy}{x+y} = \frac{1}{2}\) using an epsilon/delta approach.
  • Another participant suggests a formulation of the epsilon/delta definition for the limit, emphasizing the need to establish a relationship between \(\epsilon\) and \(\delta\).
  • A participant proposes a proof strategy involving the triangle inequality and manipulations of the limit expression, but expresses uncertainty about certain steps in their reasoning.
  • Concerns are raised regarding the validity of the argument that relates the distance in the limit to the square and circle geometries, with a specific counterexample provided.
  • A later reply acknowledges the previous correction and attempts to refine the proof by suggesting that if a point is chosen within a circle of radius \(\epsilon\), it will also lie within a square of side \(2\epsilon\), thus proposing a new condition for \(\delta\).

Areas of Agreement / Disagreement

Participants do not reach a consensus on the validity of the proof steps. There is an ongoing debate regarding the correct application of the epsilon/delta definition and the geometric interpretations involved.

Contextual Notes

Participants express uncertainty about the implications of their chosen geometric shapes (circle vs. square) and how these relate to the epsilon/delta conditions. There are unresolved questions about the sufficiency of the proposed conditions for \(\delta\) in relation to \(\epsilon\).

pureza
Messages
3
Reaction score
0
Hi,

I'm trying to wrap my head around epsilon/delta proofs for multivariable limits and it turns out I became stuck on an easy one!

The limit is:

[itex]\lim_{(x,y) \to (1,1)}\frac{xy}{x+y}[/itex]

Obviously, the result is [itex]1/2[/itex], but I'm unable to prove it!

Any hints?

Thank you!
 
Physics news on Phys.org
Well, what have you done? I assume you want to prove that

"Given [itex]\epsilon> 0[/itex] there exist [itex]\delta> 0[/itex] such that if [itex]\sqrt{(x-1)^2+ (y-1)^2}<\delta[/itex] then
[tex]\sqrt{\frac{xy}{x+y}- 1}<\epsilon[/tex]"
 
Actually I think I might have solved it:

I want to prove that

Given [itex]\epsilon> 0[/itex] there exist [itex]\delta> 0[/itex] such that if [itex]\sqrt{(x-1)^2+ (y-1)^2}<\delta[/itex] then
[tex]\left|\frac{xy}{x+y}- \frac{1}{2}\right|<\epsilon[/tex]

Now,

[tex]\left|\frac{xy}{x+y}- \frac{1}{2}\right|=\left|\frac{2xy-(x+y)}{2(x+y)}\right|=\left|\frac{xy + xy-x-y)}{2(x+y)}\right|=\left|\frac{x(y-1)+y(x-1)}{2(x+y)}\right|\leq\left|\frac{x(y-1)}{2(x+y)}\right|+\left|\frac{y(x-1)}{2(x+y)}\right|[/tex] by the triangle inequality.

Now, I know that since [itex]{(x,y) \to (1,1)}[/itex] I can make [itex]x[/itex] and [itex]y[/itex] positive so that [itex]\left|\frac{x}{x+y}\right|\leq1[/itex] and [itex]\left|\frac{y}{x+y}\right|\leq1[/itex]. [I'm no sure about this argument]

This way,

[tex]\left|\frac{x(y-1)}{2(x+y)}\right|+\left|\frac{y(x-1)}{2(x+y)}\right|\leq|y-1|+|x-1|[/tex]

But this is just a square with side 2 around the point [itex](x, y)[/itex]. This square is entirely contained by the circle with center [itex](x, y)[/itex] and radius [itex]\sqrt{2}[/itex].

So, when given some [itex]ε>0[/itex] if I set [itex]δ=min(1, ε\sqrt{2})[/itex] and I choose a point [itex](x, y)[/itex] such that [itex]\sqrt{(x-1)^2+ (y-1)^2}<ε\sqrt{2}[/itex], then certainly [itex]|y-1|+|x-1|<ε[/itex] and, by extension,

[tex]\left|\frac{xy}{x+y}- \frac{1}{2}\right|<ε[/tex]

What's wrong with my rationale? :-)
 
Last edited:
pureza said:
So, when given some [itex]ε>0[/itex] if I set [itex]δ=min(1, ε\sqrt{2})[/itex] and I choose a point [itex](x, y)[/itex] such that [itex]\sqrt{(x-1)^2+ (y-1)^2}<ε\sqrt{2}[/itex], then certainly [itex]|y-1|+|x-1|<ε[/itex] and, by extension,

[tex]\left|\frac{xy}{x+y}- \frac{1}{2}\right|<ε[/tex]

I did not read anything other than this line so I cannot comment on anything else you wrote, but this is wrong. Take (x,y) such that ε < [(x-1)2+(y-1)2]1/2 ≤ ε21/2. Then notice that [(x-1)2+(y-1)2]1/2 ≤ |x-1|+|y-1| to find (x,y) with [(x-1)2+(y-1)2]1/2 ≤ ε21/2 but ε < |x-1|+|y-1|.
 
Obviously, you are right (thank you). Take [itex]ε=1[/itex]. I can choose a point in the circle with radius [itex]\sqrt{2}[/itex] that is not inside the square [itex]|x-1|+|y-1|<1[/itex].

How about this: if I choose a point inside the circle with radius ε, then that point is also inside the square with side 2ε, because the circle is contained within the square.

Therefore, if I set [itex]δ\leq ε[/itex], then I know that any point inside the circle with radius δ verifies [itex]|x-1|+|y-1|<ε[/itex], which, if the first part of my derivation is correct, implies that

[tex]\left|\frac{xy}{x+y}- \frac{1}{2}\right|<ε[/tex]

Is this correct?
 

Similar threads

  • · Replies 20 ·
Replies
20
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
Replies
11
Views
2K
  • · Replies 19 ·
Replies
19
Views
6K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 25 ·
Replies
25
Views
4K
  • · Replies 3 ·
Replies
3
Views
30K