Implicit circles
Excerpt: Students are asked to find the gradient of three points on the graph of a circle, thinking about different ways to solve this problem. The specific problem is then generalised as a way of introducing implicit differentiation.
Problem
Find the gradient of the curve at the following three points.
What is the derivative, , at a general point on the curve?
Next step
What is the derivative at a general point on this curve?
Can you use the same methods as above?
Things you might have done
There are different ways we could think about finding the gradient of a circle at a point.
Geometric approach
| The gradient of the curve is the same as the gradient of the tangent at that point, and we know that the tangent to a circle is perpendicular to the radius. The radius through is part of the line . So the tangent is the line . We don’t need to find as we just want the gradient, which is . | |
|---|---|
| We don’t need to find the equations of the radius and tangent, we can just think about their gradients. At the point the gradients are |
Differentiation
We could differentiate the equation of our curve to find the gradient at .
The difficulty is that we do not have an equation in the form . Can we rearrange it?
This final line causes problems as there are two options for the square root. How might we proceed from here?
Given that we have a specific point, , we know we want the negative square root as our value of is negative. Therefore we could differentiate using the chain rule to get At this gives us a gradient of .
Notice that our expression for is undefined for and for . But here we are only interested in .
What is the derivative at a general point on the curve?
Again, we can tackle this in a number of different ways.
Geometric approach
We see that the gradient of the radius at is .
As the tangent is perpendicular to the radius, the gradient of the curve at that point will be or .
Would it matter if our point was in a different quadrant?
Rearranging the equation
If we rearrange the equation, we get as above. The two cases make this slightly awkward, but we can treat them separately.
If then using the chain rule we have
Whereas, if we have
So for each case we get the same result as above.
Notice that something strange happens in between the two cases, at .
Differentiating directly
What happens if instead we differentiate the original equation directly?
Differentiating both sides we get which is the same as and therefore
This in itself is not very helpful, but if we define we can then use the chain rule. This is useful because now we have an expression involving the gradient of the curve.
This works because, even though we could not write as a function of , if you look at just the part of the curve near a given point, and are functions of and the gradients do have sensible meanings.
We can differentiate the and the , but how do we differentiate the term with respect to ? Although we can't write explicitly in the form , is still related to . This is what is called an . We can think of as a function of that is unknown. That means the term can be thought of as some function of squared, i.e. . This is a composite function in that can be differentiated by the chain rule. Differentiating using the chain rule gives us . If we use the notation of we would write the derivative as . Can you see that these two different ways of writing the derivative are the same?
What we have done is called implicit differentiation and means that when we differentiate we get
We can now rearrange to find , the same as we had from the other approaches.
You might notice that the gradient function is in terms of both and . This makes sense because for each value of , there may be two points on the curve, each of which has a different and different gradient.
You might check this result by substituting the coordinates given earlier to make sure we get the same gradients.
What happens when ? What does this mean geometrically on the circle?
In this case, we had a choice of three methods, but for many other curves the first two approaches do not work and implicit differentiation becomes a valuable tool.
The first approach above relied on the geometry of the circle. It used the fact that a tangent to a circle is at right-angles to the radius. Is this true for an ellipse? Imagine your ellipse as a circle that has been stretched in one direction. What happens to the angle between the tangent and the radius?
Here we will differentiate implicitly.
We can rearrange to find the gradient.