Intuition The one core ideaTo find the top of a hill or the bottom of a valley on a smooth landscape, you look for the single spot where the ground is perfectly flat in every direction . The whole topic is just the machinery for measuring "how tilted is the ground here?" — and that measuring tool is the gradient , which must equal zero at any interior peak or pit.
Before you can read the parent note, you must be able to read its symbols . Below is every piece of notation and every idea it silently assumes, built up one at a time — each one earned before the next uses it. Nothing here assumes you have seen calculus notation before.
f f rule that eats numbers and spits out one number . We write f ( x ) x f
Picture a landscape. You stand somewhere, and the machine tells you your height above sea level. In one variable, the input is your position along a path; in two variables, your position on a map.
The parent note needs f means finding the input that makes this height as small as possible.
Definition The number line and its bigger cousins
R all real numbers — every point on an infinite number line, including fractions and decimals.R n n n R 2 x , y R 3
The arrow notation f : R n → R f n
Intuition Why the topic writes it this wayThe parent statement must cover a valley in any number of dimensions at once. R n → R
Reveal check:
f : R 2 → R a point ( x , y )
Definition The star means "the answer"A plain x x ∗ specific, distinguished input — the location of the optimum we are trying to describe. The star is just a name-tag saying "this one is special."
In the picture of a valley, x ∗ floor of the valley: the horizontal position directly under the lowest point.
Definition Three location words
Local minimum : a point that is the lowest compared only to its nearby neighbours , not necessarily the whole world. A dimple on a hillside counts.Interior point : a point with room to move in every direction — you can step a little in + − Boundary point : a point on the edge of the allowed region , where at least one direction is blocked.
Intuition Why the topic cares so muchThe whole proof relies on being able to step both ways . That is only possible at an interior point. This is exactly why the parent's "I forgot the boundary" warning exists — the flat-ground argument silently assumes you are not pinned against a wall.
Reveal check:
Why must x ∗ You must be free to move in both the + −
Definition The small steph h tiny nudge away from a point: from x ∗ x ∗ + h h
Definition What a limit is, in wordslim h → 0 ( expression ) h at zero (which might be undefined).
Intuition Why a tool for shrinking is neededThe slope of a curve keeps changing. To pin down the slope exactly at one point , we look at the slope of a tiny segment and shrink the segment to nothing. The limit is the machine that performs that shrink cleanly.
The little superscript matters:
h → 0 + h from the positive side " (approach from the right).h → 0 − from the negative side " (approach from the left).
The parent's proof approaches from both sides separately, so this + / −
Definition Derivative = slopef ′ ( x ) f x f ′ ( x ) = lim h → 0 h f ( x + h ) − f ( x ) .
Break the fraction down, because every piece is a picture:
this tool and not another?
We want to know "which way is downhill and how steep?" The derivative is precisely the answer: a positive f ′ negative means downhill-to-the-right, and zero means flat. That is exactly the flatness the topic is chasing.
Sign meaning (all three cases):
f ′ ( x ) > 0 f ′ ( x ) < 0 f ′ ( x ) = 0 x 3
Definition Freeze all but one inputWhen f ∂ x i ∂ f x i ∂ d
Intuition Why we need several of themOn a 2-D landscape "the slope" is ambiguous — steep north-to-south, maybe flat east-to-west. Splitting into one partial per axis lets us measure the tilt one clean direction at a time, then reassemble.
Reveal check:
What does the curly ∂ d That other input variables are being held constant while we differentiate along one axis.
Definition A vector, plainlyA vector is an arrow with a length and a direction . In coordinates we stack its parts in a column, e.g. ( 3 − 2 )
The gradient ∇ f f f ∇ f = ∂ x 1 ∂ f ⋮ ∂ x n ∂ f . arrow pointing in the direction of steepest increase , and its length says how steep that steepest direction is.
Intuition Why the gradient is the star of the whole topicIf any direction is uphill, the gradient arrow points that way and has non-zero length — so you can improve by walking against it. The only way no direction is uphill is if the arrow has zero length : ∇ f = 0
The bold zero 0 vector == — an arrow of length zero, pointing nowhere — as opposed to the plain number 0 ∇ f ( x ∗ ) = 0 = 0
See Gradient and directional derivatives for the full construction.
A unit vector u length exactly 1 . It carries a pure direction with no magnitude, so it answers "which way?" without also answering "how far?"
The dot product a ⋅ b how much they point the same way . Same direction → large positive; perpendicular → zero; opposite → negative.
Definition Directional derivativeD u f ( x ∗ ) = ∇ f ( x ∗ ) ⋅ u u
Intuition Why this gives the deepest proofPick the direction u against the gradient: u = − ∇ f / ∥ ∇ f ∥ ∥ ⋅ ∥ length of the vector . Then D u f = − ∥ ∇ f ∥
Reveal check:
D u f ( x ∗ ) ∇ f u The dot product ∇ f ( x ∗ ) ⋅ u
Definition Stationary (critical) pointAny input where ∇ f = 0 candidate for a max, min, or saddle — flat ground, cause unknown.
Definition Hessian (named here, studied later)The Hessian is the table of second derivatives — the "curvature" of the surface. It tells max from min from saddle after the gradient has shortlisted a flat point. Full detail in Hessian matrix and second-order conditions ; the trap it resolves lives in Saddle points .
Function f: input to height
Derivative f prime: slope in 1D
Partial derivative: slope along one axis
Vector: arrow with length and direction
Gradient: all slopes in one arrow
Unit vector: pure direction
Interior point: room both ways
Gradient = zero condition
Hessian: classify the candidate
Each foundation above is a prerequisite arrow feeding into the parent: the topic note .
Read every line as a question; only unfold to the parent note once you can answer all of them.
I can say what f : R n → R f n
I know the difference between 0 0 0 0 vector — an arrow of length zero.
I can describe an interior point and why the topic needs it. A point with room to step in every direction; the proof must move both + h − h
I can read lim h → 0 + lim h → 0 − Approach zero from the positive (right) side vs the negative (left) side.
I can state what a derivative measures. The slope of the curve at a point — rise over run as the run shrinks to zero.
I know what the curly ∂ A partial derivative — slope along one axis while other variables are held fixed.
I can say what the gradient ∇ f The arrow pointing in the direction of steepest increase; its length is that steepness.
I can define the dot product in one sentence. A number measuring how much two vectors point the same way.
I can write the directional derivative as a formula. D u f ( x ∗ ) = ∇ f ( x ∗ ) ⋅ u
I know what a stationary point is and that it is only a candidate. A point with ∇ f = 0
