Chapter 4: Applications of Differentiation

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement, not replace the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome to the deep dive, where we dig into the sources and notes to pull out the really important stuff.

Today, we're looking at something that, well, it sounds abstract, but it's actually behind so much efficiency and even beauty in the world.

That's right.

We're talking about optimization through calculus.

Yeah, the power of finding the best.

Exactly.

The mathematician, Leonard Euler, he had this great quote, something like,

nothing at all takes place in the universe in which some rule of maximum or minimum does not appear.

That's pretty profound when you think about it.

Everything trying to find some kind of best state.

It really is.

And calculus,

specifically differentiation, well, that's the lens we use to figure out how and why.

So like real world questions.

Think about bees.

How do they build their honeycombs to use the least amount of wax?

That's minimization.

Ah, okay.

Or like how blood vessels branch to make it easiest for the heart to pump blood.

Precisely.

Minimizing the heart's energy expenditure, or even something really specific, like what's the best way for your windpipe to contract during a cough to expel air fastest?

Wow, okay.

And you mentioned rainbows too.

Even rainbows.

It tangles the colors.

It connects back to principles of light taking paths that minimize or maximize travel time.

It's all about finding those peaks and valleys of functions.

So for this deep dive, we're pulling the core ideas straight from chapter four, applications of differentiation in Stuart, Clegg, and Watson's calculus,

early transcendentals.

Right, the goal is to give you a clear, engaging summary.

Break down these sometimes tricky ideas into steps that make sense.

With intuitive explanations and examples, connecting it all back to why calculus is so useful for finding the best way to do things.

Let's start with the basics then.

How do we even talk about these best or worst outcomes mathematically?

It starts with maximum and minimum values, extreme values.

Okay, so if I'm looking at a graph, say over a certain range, how do I tell the difference between just a high point and the highest point overall?

Good question.

We need to distinguish between absolute extremes and local extremes.

An absolute maximum is the single highest point the function reaches across its entire domain that we're considering.

The absolute minimum is the lowest, think global.

The overall winner.

Exactly.

Then you have local maximums and minimums.

These are like neighborhood peaks and valleys.

They're the highest or lowest points only compared to points immediately nearby.

So you could have a bunch of local hills, but only one Mount Everest, the absolute highest.

Perfect analogy.

An important detail, a local maximum or minimum usually happens where the function smoothly turns around.

It generally can't be right at the edge, an endpoint, unless that endpoint also happens to be the absolute highest or lowest point nearby too.

Okay, that makes sense.

Now you mentioned finding these is crucial for optimization.

There's a theorem for this, right?

The extreme value theorem.

Why is that such a big deal?

It is a big deal.

The EVT, the extreme value theorem.

It sounds simple, but it's powerful, it says.

If your function is continuous, no breaks, no jumps on a closed interval, meaning from point A to point B, including the ends.

Then what?

Then you are guaranteed to find both an absolute maximum value and an absolute minimum value somewhere in that interval.

Ah, so it guarantees they exist.

You're not hunting for something that might not even be there.

Precisely.

That guarantee is huge for optimization problems.

If the conditions hold, a best and worst case exists.

So if we know they exist, how do we actually find them?

Okay, that brings us to Fermat's theorem.

This theorem tells us something really useful.

If a function has a local max or min at some point C, and if the function is differentiable there.

Meaning it has a defined slope.

Right.

Then the derivative at that point, if C, must be zero.

Which visually means the tangent line is flat, horizontal at that peak or valley.

Exactly, the slope is zero.

This leads us straight to the idea of critical numbers.

Okay, critical numbers, what are they?

A critical number C is just a number in the functions domain where one of two things happens.

Either the derivative FC is zero or the derivative SD doesn't exist.

So places with horizontal tangents or sharp corners, cusps, places where the slope is undefined.

You got it.

And Fermat's theorem basically says,

if you're looking for a local maximum or minimum, it has to occur at one of these critical numbers.

They're your prime suspects.

So we hunt down the critical numbers, but how does that help find the absolute max and min on a closed interval?

The ones the EVT guaranteed.

That's where we use the closed interval method.

It's a really straightforward three -step recipe.

First, find all the critical numbers inside your interval.

Calculate the functions value at each of those critical numbers.

Okay, step one, check the suspects inside the interval.

Step two, calculate the functions value at the two end points of the interval.

Don't forget the edges.

Right, the max or min could be right at the start or finish.

Exactly.

Step three,

just compare all the values you calculated in steps one and two.

The biggest value you found, that's your absolute maximum, the smallest value, that's your absolute minimum.

Done.

That's very systematic.

Find the suspects, check the boundaries, compare the results.

And it always works if the EVT conditions, continuous function, closed interval are met.

The textbook has this great example applying it to the NASA space shuttle launch.

Oh yeah, how does that work?

Well, they have a function describing the shuttle's velocity over time.

To find the acceleration, you take the derivative of velocity.

Right, acceleration is the rate of change of velocity.

So they get this acceleration function and they wanna know the maximum and minimum acceleration during the first couple of minutes, say from time T zero to T one of five, 26 seconds.

That's a closed interval.

Ah, so they apply the closed interval method to the acceleration function.

Exactly, find the critical numbers of the acceleration function, check the acceleration at the start, T zero and end, T one, and compare all those values.

It lets them pinpoint the moments of peak stress, basically.

Real world application.

That's cool, it's not just about graphs, but physical forces.

Okay, so finding peaks and valleys is one major application.

What else can derivatives tell us?

There's another big theorem, right?

The mean value theorem.

Yes, the MVT.

But actually, it helps to quickly mention Rolle's theorem first, because MVT is like a tilted version of Rolle's.

Okay, Rolle's theorem.

Rolle's is specific.

It says, if you have a function that's continuous, differentiable, and it starts and ends at the exact same height over an interval,

so FA equals FB.

Starts and ends level.

Then there must be at least one point C somewhere in between where the derivative is zero.

So at least one horizontal tangent.

Makes sense, if you drive up a mountain and then back down to the same starting elevation and your path was smooth, you had to be level at the peak, at least for an instant.

Good analogy.

Now the mean value theorem generalizes this.

It drops the requirement that FA has to equal FB.

It just says, if F is continuous on the closed interval AB and differentiable on the open interval AB.

So smooth path between the endpoints.

Then there's at least one point C inside the interval where the instantaneous rate of change, FC, is exactly equal to the average rate of change over the whole interval.

Wait, say that last part again.

Instantaneous equals average.

Yeah, the derivative at C, FC, equals the slope of the line connecting the endpoints,

FB, FA, BA.

Okay, so geometrically that means the tangent line at C is parallel to the second line connecting the start and endpoints.

Precisely, imagine that second line.

The MVT guarantees that somewhere along the curve, the tangent line will have that exact same slope.

It's parallel.

Okay, I think I see it.

What about a more physical feel for it?

The driving example is perfect here.

Let's say you drive 180 kilometers in two hours.

What's your average speed?

90 kilometers per hour.

Right, the mean value theorem guarantees that at some point during those two hours, your speedometer, your instantaneous speed, must have read exactly 90 kilometers.

Maybe you went faster sometimes, slower other times, but you definitely hit 90 at least once.

Ah, okay, that clicks.

The instantaneous rate had to match the overall average rate at some point.

That's actually really intuitive when you put it that way.

It is, and it has some important consequences.

For example, if we know a function's derivative is zero everywhere across an interval.

Meaning its slope is always zero.

Then the function must be constant on that interval.

It can't be going up or down if its slope is always zero.

That seems obvious, but the MVT actually proves it rigorously.

Yes, and that's why when we find anti -derivatives later, we always add that plus C, that constant of integration, because any two functions with the same derivative can only differ by a constant.

Got it.

Okay, so MVT connects average and instantaneous rates.

Let's shift back to analyzing the shape of graphs more.

Beyond just finding maximum points, what else do derivatives reveal?

A lot.

The first derivative tells us about increasing and decreasing behavior.

It's simple, if fx is positive, the function fx is going uphill, increasing.

If fx is negative, it's going downhill, decreasing.

Slope up, function up.

Slip down, function down, makes sense.

And this leads directly to the first derivative test for finding local max and mines.

If the derivative f changes sign from positive to negative at a critical number C.

So going uphill, then downhill.

You've hit a local maximum at C, like the peak of a hill.

If f changes from negative to positive at C.

From uphill to uphill.

You found a local minimum at C, the bottom of a valley.

And if the sign doesn't change.

Like a flat spot on a hill.

Yeah, or like an fxx cubed at x zero, then there's no local extremum there.

Okay, so the first derivative handles increasing, decreasing, and local peaks, valleys.

What about the way the curve bends?

Ah, that's concavity.

And that's the job of the second derivative, fx.

A function is concave upward, cu.

On an interval, if its graph looks like a cup holding water, it lies above its tangent lines.

Like the bottom part of a smile.

Sort of, yeah.

And it's concave downward, cd.

If it looks like an upside down cup spilling water, the graph lies below its tangent lines.

Like a frown.

So how does the second derivative tell us this?

It's the concavity test.

If fx is positive on an interval, the function f is concave upward there.

If fx is negative, f is concave downward.

Second derivative positive, concave up.

Second derivative negative, concave down.

Got it, what happens when it changes?

That's an inflection point.

It's a point on the curve where the concavity changes.

Maybe from up to down or down to up.

And the function has to be continuous there.

So the curve switches its bending direction.

Is that important?

It can be very important.

Think about population growth like in the honeybee example of the book.

The growth often starts slow, then speeds up, then slows down again as it approaches some limit.

Like an S -shaped curve.

Exactly.

The inflection point on that S -curve is where the population is growing fastest.

The rate of growth hits its maximum right there before it starts to level off.

That's a really significant point in understanding the dynamics.

Wow, okay.

So inflection points mark changes in the rate of change.

You nailed it.

And the second derivative also gives us another way to test for local maxvins called the second derivative test.

How does that one work?

It's sometimes quicker than the first derivative test.

If you have a critical point C where Fc equals zero, so a horizontal tangent.

Okay.

Then you check the sign of the second derivative there.

If Fe is positive.

Positive second derivative means concave up.

Right, so if you have a flat spot on a curve that's concave up, you must be at a local minimum.

Like the bottom of a bowl.

Makes sense.

And if Fdc is negative.

Then you must be at a local maximum, the top of a hill.

But what if the second derivative is zero there too?

Then the test is inconclusive, it tells you nothing.

You have to go back to the first derivative test in that case.

So the first derivative test is kind of more robust,

but the second can be faster when it works.

Okay, so we have tools for increasing, decreasing, local maxmin, concavity, inflection points.

Putting it all together helps us sketch graphs accurately.

Exactly.

Section 4 .5 in the book gives guidelines for curve sketching that combine all these things with intercepts, asymptotes, domain.

It's the whole toolkit for understanding a functions graph.

Cool.

Now sometimes when we're evaluating limits, we run into problems, right?

Like try and divide zero by zero.

Yes, exactly.

Those are called indeterminate forms.

The most common are zero, zero and infinity, infinity.

If you just plug in the limit value and get one of those, you don't actually know what the limit is.

Could be anything.

So we need another tool.

And that tool is L Hospital's Rule.

It's spelled L apostrophe, H -O -P -I -T -A -L, but often pronounced lopital.

It's a fantastic trick for handling these indeterminate ratios.

What does it let us do?

Basically, if you have the limit of a fraction,

fx, that results in zero, zero or infinity, infinity, L Hospital's Rule says you can instead take the limit of the ratio of their derivatives, fdhs.

So you differentiate the top, differentiate the bottom separately, and then try the limit again.

Exactly.

And often that new limit is much easier to evaluate.

Sometimes you might even need to apply the rule more than once if the result is still indeterminate.

Like in the example limit of x by two, as x goes to infinity, both blow up.

Right, apply all hospitals once, you get x to x, still infinity over infinity, apply it again, you get x two.

Now as x goes to infinity, the limit is clearly infinity.

Wow, that's powerful.

Does it work for other weird forms?

Like zero times infinity?

Yes, but you have to transform them first.

For zero times infinity, you rewrite the product as a fraction like f1g or g1f to get it into the zero, zero or infinity, infinity form.

Then you can use the rule.

Clever.

What about infinity minus infinity?

Usually involves finding a common denominator or factoring to turn the difference into a fraction.

And those weird power ones, like zero to the zero or one to the infinity?

For those, the trick is usually use logarithms.

You take the natural log of the expression, evaluate the limit of the log, which often becomes a zero, zero or infinity, infinity type after some algebra, and then exponentiate the result to get the original limit.

Okay, so all hospitals rule with some algebraic manipulation gives us a systematic way to crack a lot of tough limits.

It really does, it's a lifesaver sometimes.

And all these techniques, finding extremes, analyzing shape, handling limits, they all come together when we tackle optimization problems, right?

This is where we really use calculus to find the best.

Section 4 .7 is all about this.

It's applying everything we've learned to practical scenarios.

There's a general strategy.

What's the gist?

First,

really understand the problem.

What quantity are you trying to maximize or minimize?

Then draw a picture, it almost always helps, introduce variables.

Okay, visualize and label.

Then the key step,

write an equation for the quantity you wanna optimize.

Try to get it in terms of just one variable.

This might involve using some constraint or relationship given in the problem.

Get it down to a function of a single variable, say fx.

Exactly, then you use calculus, find the critical numbers of fx, use the first or second derivative test, maybe the closed interval method if applicable, to find the maximum or minimum value.

And then double check it makes sense in the context of the problem.

Right, make sure you actually found a max if you wanted a max or min if you wanted a min and that the answer is physically reasonable.

The book has some classic examples, like the farmer fencing a rectangular field next to a river.

Yeah, maximizing the area with a fixed amount of fence

or the optimal can design problem.

Yes, minimizing the amount of metal needed for a cylindrical can holding, say, one liter.

That one's fascinating.

If you only care about minimizing surface area, the math tells you the most efficient shape is when the height equals the diameter, so h equals two r.

But most cans aren't shaped like that, are they?

Soda cans are taller and thinner.

Exactly, and that's where the real world bumps up against pure optimization.

Maybe taller cans are easier to hold or fit better in vending machines or are more stable on shelves.

So the actual design is a compromise between minimizing material cost and other practical factors.

Calculus gives the ideal baseline.

That's a great point.

It shows the power but also the context.

What other kinds of problems?

Oh, finding the point on a curve closest to a given point or that river crossing problem.

A woman wants to row across a river and then run along the bank to reach a point downstream.

Where should she land her boat to minimize her total travel time?

Right, trading off slower rowing speed for faster running speed.

Exactly, you model the total time as a function of where she lands and then find the minimum, or maximizing the area of a rectangle inscribed in a semicircle.

Lots of geometry applications.

And of course, business and economics.

Huge applications there.

Minimizing average cost, maximizing revenue, maximizing profit.

You model cost CX, demand PX, revenue RX equals XPX and profit PX RICX.

Then you use derivatives to find optimal production levels or pricing strategies.

So these tools are fundamental across science, engineering, economics,

everywhere you wanna find the best way.

Pretty much.

Now the final section in this chapter, 4 .9, kind of flips things around.

It introduces anti -derivatives.

So instead of finding the derivative, we're finding the function whose derivative is known, like going backward.

Exactly, if you know FX, you're looking for a function FX, such that FX equals FX, we call FX an anti -derivative of FX.

It's the reverse process.

Like undoing differentiation.

Precisely.

And the key thing here, connected back to that MVT consequence, is that if you find one anti -derivative SX, then any function of the form FX plus C, where C is just some constant number, is also an anti -derivative.

Because the derivative of a constant is zero?

Right, so there isn't just one anti -derivative, there's a whole family of them, all different by a constant.

We write the general anti -derivative as FX plus C.

Okay, so the anti -derivative of, say, cos X would be sin X plus C.

Or the anti -derivative of X is XN plus one, N plus one plus C, as long as N isn't negative one.

What about N negative one?

That's one X.

Ah, the anti -derivative of one X is LNX plus C.

We need the absolute value, because the domain of one X includes negative numbers, but LNX is only defined for positive X.

Got it.

So what's the point of this S plus C?

Can we ever figure out what C is?

Yes.

That's what initial value problems are about.

If you're given the derivative, say, X, and you also know the value of the original function FX at one specific point, like F zero I or something.

The initial condition.

Right.

Then you can find the general anti -derivative, FX plus C, plug in the known point, and solve for the specific value of C that makes the condition true.

That gives you the particular anti -derivative that satisfies the initial condition.

So you nail down the exact function from the whole family of curves.

And you might even have problems where you're given the second derivative in two initial conditions,

maybe the value of the function and the value of the first derivative at certain points.

You just anti -differentiate twice, solving for a constant each time.

Okay, and this clearly connects to motion, right?

Acceleration, velocity, position.

Absolutely fundamental there.

We know velocity VLA is the derivative of position ST, and acceleration AT is the derivative of velocity VT.

So ST is VT and VT.

Anti -differentiation lets us go backward.

If you know the acceleration, like gravity maybe minus 9 .8 meters all on us, you can find the velocity function by anti -differentiating using the initial velocity to find the constant.

And then anti -differentiate velocity to get position.

Exactly, use the initial position to find the second constant.

So if you know the acceleration and the initial velocity and position, you can completely determine the object's position function for all time.

Like the example of throwing a ball off a cliff.

Given initial height and velocity, you can find the function for its height at any time, Tid?

Yes, and then you can use that function to answer questions like, when does it reach its maximum height?

That's when its velocity is zero.

When does it hit the ground?

That's when its height ST is zero.

It ties everything together.

Wow, okay.

So wrapping this up, chapter four really shows how derivatives move beyond just being a calculation.

They become tools for deeply understanding how things behave.

That's the core takeaway.

Peaks, valleys, rates of change, concavity, how functions bend.

Derivatives unlock all of that.

And that understanding lets us optimize things.

Finding the best, the most efficient, the maximum, the minimum, and everything from engineering and physics to biology and economics.

It connects the abstract math to tangible real -world outcomes.

It's about finding those optimal solutions that nature often hints at or that engineers strive for.

It really does make you look at the world a bit differently.

Hopefully.

So maybe the next time you see, I don't know, a bridge structure, or even just observe something in nature, you can pause and wonder.

What's being optimized here?

What are the trade -offs?

Are there mathematical principles shaping this?

A great thought to leave with, how calculus might be subtly shaping the world around us.

Well, from the entire Deep Dive team, thank you for joining us on this exploration of the applications of differentiation.

We hope you feel a bit more clued into the practical power hiding within calculus.