Vector geometry, projections and oriented area
Introduced| English | Meaning |
|---|---|
| projection/prəˈdʒekʃn/ | The component of a vector along a specified nonzero direction. |
| cross product/krɒs ˈprɒdʌkt/ | In three dimensions, a vector perpendicular to both inputs, with length equal to their parallelogram area and orientation set by their order. |
| normal/ˈnɔːml/ | A nonzero vector perpendicular to a plane or, at a regular surface point, to its tangent plane. |
How far along, and how far away?
- A drone's displacement may point partly along a flight path and partly away from it. A vector projection 投影 separates these parts; a dot product measures alignment, while a cross product 叉积 measures oriented area in three dimensions.
- We will connect angles, nearest points, triangle areas, planes and volumes. Keep track of what each answer is: a scalar, a vector, a length or an equation.
Dot products and nonzero directions
- In Euclidean coordinates, $\mathbf u\cdot\mathbf v=\sum_i u_i v_i$ is a scalar. For nonzero vectors it equals $|\mathbf u||\mathbf v|\cos\theta$, with $0\leq\theta\leq\pi$. The zero vector has no direction and no defined angle.
- Positive, zero and negative products mean $\theta<90^\circ$, $\theta=90^\circ$ and $\theta>90^\circ$. Acute and obtuse exclude the endpoints: same-direction vectors have angle $0^\circ$, and opposite vectors have angle $180^\circ$.
The zero vector makes a right angle with every nonzero vector because their dot product is zero.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
The zero vector has no direction and no defined angle. The angle formula requires both lengths to be nonzero.
Work the sign and the angle
- For $\mathbf u=(1,2,-1)$ and $\mathbf v=(3,-1,1)$, the product is $3-2-1=0$. Both vectors are nonzero, so they are perpendicular. Zero product does not mean either vector must be zero.
- For $(1,1,0)$ and $(1,0,1)$, the product is 1 and both lengths are $\sqrt2$. Thus $\cos\theta=1/2$ and $\theta=\pi/3$. Divide by both lengths to find the angle, not just its type.
Which angle type is defined?
Use nonzero conditions and the dot product sign.
$(1, 2, -1)$ and $(3, -1, 1)$ are perpendicular.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Both vectors are nonzero and their dot product is 3-2-1=0, so they are perpendicular.
Derive a vector projection
- Project $\mathbf v$ onto the line through a nonzero vector $\mathbf u$. The projected vector has form $\mathbf p=\lambda\mathbf u$. Require the remainder $\mathbf r=\mathbf v-\mathbf p$ to satisfy $\mathbf r\cdot\mathbf u=0$.
- Solving gives $\lambda=(\mathbf v\cdot\mathbf u)/(\mathbf u\cdot\mathbf u)$, hence $\mathbf p=\dfrac{\mathbf v\cdot\mathbf u}{\mathbf u\cdot\mathbf u}\mathbf u$. The denominator is positive because $\mathbf u\ne\mathbf0$.
Vector, signed component and length
- For $\mathbf v=(2,3)$ onto $\mathbf u=(1,1)$, $\lambda=5/2$, so $\mathbf p=(5/2,5/2)$ and $\mathbf r=(-1/2,1/2)$. Check $\mathbf r\cdot\mathbf u=0$: the remainder is perpendicular to the target line.
- The signed scalar component is $(\mathbf v\cdot\mathbf u)/|\mathbf u|=5/\sqrt2$; the projection's length is its absolute value. For $\mathbf v=(-2,1)$ onto $(1,0)$ these are $-2$ and 2, while the vector projection is $(-2,0)$.
The vector projection of $(2, 3)$ onto $(1, 1)$ is
Difficulty 1/5 · 100 coins + a small bonus on first completion today
The coefficient is (2+3)/(1+1)=5/2. Multiply the target vector (1,1) by it to obtain (5/2,5/2); the remainder (-1/2,1/2) is perpendicular to (1,1).
Find the signed scalar component of $(-2,1)$ in the direction $(1,0)$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
Divide the dot product -2 by the target length 1. The signed component is -2; its absolute value 2 is the projection length.
Projection finds the nearest point
- The original $\mathbf w=(3,4)$ projected onto $(1,0)$ gives $\mathbf p=(3,0)$ and remainder $\mathbf r=(0,4)$. Every point $\mathbf p+s\mathbf u$ on the target line has squared distance $|\mathbf r|^2+s^2|\mathbf u|^2$ from $\mathbf w$, because the cross term is zero. The minimum is $|\mathbf r|$ at $s=0$.
- For an offset line $L=\{\mathbf a+t\mathbf u\}$, project the displacement $\mathbf q-\mathbf a$ first, then add $\mathbf a$ back. Replacing $\mathbf u$ by any nonzero multiple leaves the projected vector and nearest point unchanged.
Find the nearest point to $(4,3)$ on the line $(1,1)+t(1,1)$.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Project displacement (3,2) onto (1,1): coefficient 5/2. Add (1,1) back to obtain (7/2,7/2). The residual (1/2,-1/2) is perpendicular to the line.
Project $\mathbf v=(4,1)$ onto $\mathbf u=(1,2)$. Give the vector projection, perpendicular remainder, signed scalar component and distance from $\mathbf v$ to the line spanned by $\mathbf u$.
Compare your quantities, nonzero conditions and checks with the model.
Compute a cross product in three dimensions
- For $\mathbf u,\mathbf v\in\mathbb R^3$, write $\mathbf u\times\mathbf v=(c_1,c_2,c_3)$, where $c_1=u_2v_3-u_3v_2$, $c_2=u_3v_1-u_1v_3$ and $c_3=u_1v_2-u_2v_1$. Check the middle component's sign carefully.
- The result is a vector, perpendicular to both inputs. Its length is $|\mathbf u||\mathbf v|\sin\theta$ for nonzero inputs. A zero cross product means linear dependence; for two nonzero vectors this means parallel or opposite directions.
Order sets the perpendicular direction
- In a right-handed coordinate system, $\mathbf e_1\times\mathbf e_2=\mathbf e_3$. Curl the right-hand fingers from the first vector toward the second through the smaller angle; the thumb gives the cross-product direction when it is nonzero.
- Swapping the inputs reverses the vector: $\mathbf v\times\mathbf u=-\mathbf u\times\mathbf v$. Its length stays the same. A nonzero cross product selects an orientation; parallel inputs give zero, so no perpendicular direction is selected by that product.
If $\mathbf u\times\mathbf v=(2,-1,4)$, what is $\mathbf v\times\mathbf u$?
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Swapping the order reverses every component of the vector; its length stays the same.
From a parallelogram to a triangle
- The parallelogram spanned by $\mathbf u,\mathbf v$ has area $|\mathbf u\times\mathbf v|$. A triangle with vertices $A,B,C$ has half that area, using two edges from the same vertex, such as $B-A$ and $C-A$.
- For $(0,0,0)$, $(2,0,0)$ and $(0,3,0)$, the edge cross product is $(0,0,6)$ and the triangle area is 3. The original diagram also shows four axis directions: their dot products will matter later.
Area does not depend on position or order
- For the original vertices $A=(0,0,0)$, $B=(1,2,0)$ and $C=(3,1,0)$, the edge cross product is $(0,0,-5)$. The area is $5/2$, never $-5/2$.
- Translating every vertex by the same vector leaves the edge differences unchanged. Swapping $B,C$ reverses the cross product but keeps its length. Three collinear vertices give zero area.
Find the area of the triangle with vertices $(0,0,0)$, $(1,2,0)$ and $(3,1,0)$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
Use edges (1,2,0) and (3,1,0) from the origin. Their cross product is (0,0,-5); its length is 5, so the triangle area is 5/2.
For $A=(1,1,1)$, $B=(2,3,1)$ and $C=(4,2,1)$, find an oriented normal using $(B-A)\times(C-A)$ and the triangle area. What changes if you swap $B,C$?
Compare your quantities, nonzero conditions and checks with the model.
A plane needs a nonzero normal
- A normal 法向量 to a plane is a nonzero vector perpendicular to every direction in that plane. Through $\mathbf a$ with normal $\mathbf n$, the equation is $\mathbf n\cdot(\mathbf x-\mathbf a)=0$.
- Equivalently, $\mathbf n\cdot\mathbf x=d$ with $d=\mathbf n\cdot\mathbf a$. Multiplying both sides by the same nonzero scalar describes the same plane. If $\mathbf n=\mathbf0$, the equation gives all space when $d=0$, or no points when $d\ne0$.
The equation $0x+0y+0z=0$ describes one plane with normal $(0,0,0)$.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Every point in space satisfies this equation. A plane equation needs a nonzero normal.
Point-to-plane distance and foot
- The signed component of $\mathbf q-\mathbf a$ along $\mathbf n$ is $(\mathbf n\cdot\mathbf q-d)/|\mathbf n|$. Its absolute value is the distance. The nearest point is $\mathbf q-\dfrac{\mathbf n\cdot\mathbf q-d}{|\mathbf n|^2}\mathbf n$; it lies on the plane and the connecting vector is normal to it.
- For $2x-y+2z=6$, the normal has length 3, so the origin is distance 2 away. For $(1,1,1)$ and $x+2y+2z=14$, the signed numerator is $-9$, distance 3, and the nearest point is $(2,3,3)$.
Find the distance from $(1, 1, 1)$ to the plane $x + 2y + 2z = 14$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
The nonzero normal is (1,2,2), of length 3. The signed numerator is 1+2+2-14=-9; take its absolute value and divide by 3. The nearest point is (2,3,3).
Build a plane from three points
- Take $A=(1,0,0)$, $B=(2,1,0)$ and $C=(1,1,1)$. The edges $(1,1,0)$ and $(0,1,1)$ have cross product $(1,-1,1)$. Thus the plane is $x-y+z=1$; substitute all three points to check it.
- Three noncollinear points give a nonzero normal and one plane. Collinear points have dependent edges and zero cross product; they lie in many planes, so this construction cannot choose a unique plane.
For the plane $x-y+z=1$ and point $Q=(2,0,2)$, find the distance and nearest point. Show that multiplying the plane equation by $-2$ changes neither answer.
Compare your quantities, nonzero conditions and checks with the model.
Triple product: height times base area
- The scalar triple product $S=\mathbf u\cdot(\mathbf v\times\mathbf w)$ is a scalar. The cross product gives an oriented base normal with length equal to base area; dotting with $\mathbf u$ multiplies by the signed height. Parallelepiped volume is $|S|$.
- $S=0$ exactly when the three vectors in $\mathbb R^3$ are linearly dependent, including cases with a zero vector. Their span then lies in some plane through the origin, possibly a line or a single point.
For $\mathbf u,\mathbf v,\mathbf w\in\mathbb R^3$, if $\mathbf u\cdot(\mathbf v\times\mathbf w)=0$, the three vectors are linearly ____.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
The scalar triple product is the determinant of the three vectors. It vanishes exactly when they are linearly dependent; the spanned parallelepiped has zero volume.
Signed volume and ordinary volume
- For $\mathbf u=(1,0,0)$, $\mathbf v=(0,2,0)$ and $\mathbf w=(1,1,3)$, $\mathbf v\times\mathbf w=(6,0,-2)$, so $S=6$. The volume is 6. Swapping two inputs makes $S=-6$ but leaves the volume 6.
- A cyclic move preserves $S$: both $\mathbf v\cdot(\mathbf w\times\mathbf u)$ and $\mathbf w\cdot(\mathbf u\times\mathbf v)$ equal $S$. For a tetrahedron with the three edge vectors from one vertex, volume is $|S|/6$; in this example it is 1.
What kind of answer is this?
Distinguish vector results from signed and nonnegative scalars.
A tetrahedron has three edges $(1,0,0)$, $(0,2,0)$ and $(1,1,3)$ from one vertex. Find its volume.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
The scalar triple product is 6. A tetrahedron has one sixth of the corresponding parallelepiped volume, so its volume is |6|/6=1.
Three pairwise obtuse directions can fit
- In a plane, $(2,0)$, $(-1,2)$ and $(-1,-2)$ have pairwise dot products $-2,-2,-3$. Thus all three smaller angles are obtuse. The directions have cyclic gaps about $117^\circ,127^\circ,117^\circ$, summing to $360^\circ$ before rounding.
- Normalize each nonzero vector to a unit direction when considering angles: positive rescaling changes dot-product size but keeps its sign. The restriction below concerns vectors in one common plane, not all of three-dimensional space.
Why four pairwise negative products cannot fit
- Suppose four nonzero planar vectors have every pairwise product negative. Their directions are distinct; order them around a circle. Each successive cyclic gap $g$ has $\cos g<0$, hence $90^\circ<g<270^\circ$. All four gaps exceed $90^\circ$, contradicting their sum $360^\circ$.
- Strict negativity matters. The original $\mathbf e_1,-\mathbf e_1,\mathbf e_2,-\mathbf e_2$ have two products $-1$ and four zeros, so every product is nonpositive. They meet the weaker condition but not the strictly negative one.
Four nonzero vectors in a common plane can have all six pairwise dot products strictly negative.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Order the four distinct directions around a circle. Every successive gap would exceed 90 degrees, so the four gaps would sum to more than 360 degrees.
Give three nonzero planar vectors with all pairwise dot products negative, and explain why four cannot exist in a common plane. Do four vectors in $\mathbb R^3$ obey the same restriction?
Compare your quantities, nonzero conditions and checks with the model.
Recap: choose the right quantity
- Check nonzero directions before finding an angle or projection. Use a perpendicular remainder for nearest-point questions, edge differences for triangle area, a nonzero normal for a plane and an absolute triple product for volume.
- projection: The component of a vector along a specified nonzero direction.
- cross product: In three dimensions, a vector perpendicular to both inputs, with length equal to their parallelogram area and orientation set by their order.
- normal: A nonzero vector perpendicular to a plane or, at a regular surface point, to its tangent plane.