What Is A Projection Vector?
The projection vector of $\vec{a}$ onto $\vec{b}$ is the component of $\vec{a}$ that lies in the direction of $\vec{b}$ — the vector you get by dropping a perpendicular from the tip of $\vec{a}$ onto the line of $\vec{b}$. It is itself a vector: it has a magnitude and a direction (the direction of $\vec{b}$).
Two related quantities share the name "projection," and keeping them apart is half the battle:
The scalar projection (also called the component) is a single signed number — how long the shadow is.
The vector projection is that length attached to $\vec{b}$'s direction — the shadow itself, as a vector.
You get the vector projection by multiplying the scalar projection by the unit vector $\hat{b}$.
What Is The Projection Vector Formula?
The scalar projection of $\vec{a}$ onto $\vec{b}$ is:
$$\text{comp}_{\vec{b}},\vec{a} = |\vec{a}|\cos\theta = \frac{\vec{a}\cdot\vec{b}}{|\vec{b}|}$$
The vector projection is the scalar projection pointed along $\hat{b}$:
$$\text{proj}_{\vec{b}},\vec{a} = \left(\frac{\vec{a}\cdot\vec{b}}{|\vec{b}|^2}\right)\vec{b}$$
Variable glossary. $\vec{a}$ is the vector being projected; $\vec{b}$ is the direction projected onto; $\theta$ is the angle between them; $\vec{a}\cdot\vec{b}$ is the dot product; $|\vec{b}|$ is the magnitude of $\vec{b}$; $\hat{b} = \frac{\vec{b}}{|\vec{b}|}$ is the unit vector along $\vec{b}$.
Notice the two formulas differ by one factor of $|\vec{b}|$ and the direction $\vec{b}$: the scalar version divides by $|\vec{b}|$ once and stops at a number; the vector version divides by $|\vec{b}|^2$ and multiplies back by $\vec{b}$ to point the result.
Where Does The Formula Come From?
Draw $\vec{a}$ and $\vec{b}$ from a common point $O$, with angle $\theta$ between them. Drop a perpendicular from the tip of $\vec{a}$ to the line carrying $\vec{b}$, meeting it at $L$. The segment $OL$ is the length of the projection.
In right triangle $OAL$, the side $OL$ is adjacent to $\theta$:
$$\cos\theta = \frac{OL}{|\vec{a}|} \quad\Rightarrow\quad OL = |\vec{a}|\cos\theta$$
The dot product gives a way to express $|\vec{a}|\cos\theta$ without measuring the angle. Since $\vec{a}\cdot\vec{b} = |\vec{a}|,|\vec{b}|\cos\theta$, dividing both sides by $|\vec{b}|$ isolates the scalar projection:
$$OL = |\vec{a}|\cos\theta = \frac{\vec{a}\cdot\vec{b}}{|\vec{b}|}$$
To turn that length into a vector, attach it to the unit vector $\hat{b} = \frac{\vec{b}}{|\vec{b}|}$:
$$\text{proj}_{\vec{b}},\vec{a} = \left(\frac{\vec{a}\cdot\vec{b}}{|\vec{b}|}\right)\frac{\vec{b}}{|\vec{b}|} = \left(\frac{\vec{a}\cdot\vec{b}}{|\vec{b}|^2}\right)\vec{b}$$
What Does A Negative Projection Mean?
The scalar projection is signed. When the angle $\theta$ between the vectors is less than $90°$, $\cos\theta$ is positive and the projection points the same way as $\vec{b}$. When $\theta$ is greater than $90°$, $\cos\theta$ is negative, the scalar projection comes out negative, and the vector projection points opposite to $\vec{b}$. When $\theta = 90°$ exactly, the projection is zero — the vectors are perpendicular and $\vec{a}$ casts no shadow along $\vec{b}$. (For the full angle story, see angle between vectors.)
Examples of Projection Vector
Example 1
Find the scalar projection of $\vec{a} = 3,\hat{i} + 4,\hat{j}$ onto $\vec{b} = \hat{i}$.
$$\vec{a}\cdot\vec{b} = (3)(1) + (4)(0) = 3, \qquad |\vec{b}| = 1$$ $$\text{comp}_{\vec{b}},\vec{a} = \frac{3}{1} = 3$$
Final answer: $3$. Projecting onto the $x$-axis just reads off the $x$-component, as expected.
Example 2
Find the vector projection of $\vec{a} = 4,\hat{i} + \hat{j}$ onto $\vec{b} = 2,\hat{i} + 2,\hat{j}$.
Wrong attempt. A student divides by $|\vec{b}|$ once and forgets the second factor, writing $\text{proj} = \frac{\vec{a}\cdot\vec{b}}{|\vec{b}|},\vec{b} = \frac{10}{\sqrt{8}}(2,\hat{i} + 2,\hat{j})$. That over-scales the result — it multiplies a length by a full vector instead of a unit vector, so the answer is $\sqrt{8}$ times too long.
Correct. Use $|\vec{b}|^2$ in the denominator, which pairs with the $\vec{b}$ on the outside.
$$\vec{a}\cdot\vec{b} = (4)(2) + (1)(2) = 10$$ $$|\vec{b}|^2 = 2^2 + 2^2 = 8$$ $$\text{proj}{\vec{b}},\vec{a} = \frac{10}{8}(2,\hat{i} + 2,\hat{j}) = \frac{5}{4}(2,\hat{i} + 2,\hat{j})$$ $$\text{proj}{\vec{b}},\vec{a} = \frac{5}{2},\hat{i} + \frac{5}{2},\hat{j}$$
Final answer: $\dfrac{5}{2},\hat{i} + \dfrac{5}{2},\hat{j}$.
Example 3
Find the scalar projection of $\vec{a} = 4,\hat{i} + 2,\hat{j} + \hat{k}$ onto $\vec{b} = 5,\hat{i} - 3,\hat{j} + 3,\hat{k}$.
$$\vec{a}\cdot\vec{b} = (4)(5) + (2)(-3) + (1)(3) = 20 - 6 + 3 = 17$$ $$|\vec{b}| = \sqrt{5^2 + (-3)^2 + 3^2} = \sqrt{43}$$ $$\text{comp}_{\vec{b}},\vec{a} = \frac{17}{\sqrt{43}}$$
Final answer: $\dfrac{17}{\sqrt{43}} \approx 2.59$.
Example 4
Find the scalar projection of $\vec{a} = \hat{i} + 2,\hat{j}$ onto $\vec{b} = -2,\hat{i} - \hat{j}$.
$$\vec{a}\cdot\vec{b} = (1)(-2) + (2)(-1) = -2 - 2 = -4$$ $$|\vec{b}| = \sqrt{(-2)^2 + (-1)^2} = \sqrt{5}$$ $$\text{comp}_{\vec{b}},\vec{a} = \frac{-4}{\sqrt{5}} \approx -1.79$$
The result is negative, so $\vec{a}$ has a component pointing against $\vec{b}$ — the angle between them is obtuse.
Final answer: $\dfrac{-4}{\sqrt{5}} \approx -1.79$.
Example 5
Project $\vec{a} = 3,\hat{i} + 4,\hat{j}$ onto $\vec{b} = -4,\hat{i} + 3,\hat{j}$.
$$\vec{a}\cdot\vec{b} = (3)(-4) + (4)(3) = -12 + 12 = 0$$ $$\text{proj}_{\vec{b}},\vec{a} = \frac{0}{|\vec{b}|^2},\vec{b} = \vec{0}$$
The dot product is zero, so the vectors are perpendicular and the projection is the zero vector.
Final answer: $\vec{0}$ — no shadow, because $\vec{a} \perp \vec{b}$.
Example 6
A $20\ \text{N}$ force pulls a sled along a rope at $60°$ above the ground. How much of the force acts horizontally (along the ground)?
The horizontal direction is the projection direction, and $\cos 60° = \frac{1}{2}$.
$$F_{\text{horizontal}} = |\vec{F}|\cos\theta = 20\cos 60°$$ $$F_{\text{horizontal}} = 20 \times \frac{1}{2} = 10\ \text{N}$$
Final answer: $10\ \text{N}$ pulls the sled forward; the rest lifts it.
Why Projection Vectors Matter: "Aplitting A Force Into The Part That Counts"
The projection vector exists to answer one recurring question — how much of this vector acts in that direction? — and the answer drives a surprising amount of applied math. Pulling a sled by an angled rope (Example 6), only the horizontal projection does useful work; the vertical part just lifts. A block on a ramp feels gravity split into a component down the slope (which slides it) and a component into the slope (which presses it). In machine learning, projecting data onto a chosen direction is the core move behind dimensionality reduction.
The deeper payoff is decomposition: every vector $\vec{a}$ splits cleanly into the part along $\vec{b}$ (the projection) and the part perpendicular to it. Subtract the projection from $\vec{a}$ and what remains is exactly the perpendicular component. That split is the reverse of vector addition — instead of combining two vectors into a resultant, you're pulling one vector apart into two useful directions.
What Are The most Common Mistakes With Projection Vectors?
Mistake 1: Using $|\vec{b}|$ instead of $|\vec{b}|^2$ for the vector projection
Where it slips in: computing the vector projection (not the scalar one).
Don't do this: writing $\text{proj}_{\vec{b}},\vec{a} = \frac{\vec{a}\cdot\vec{b}}{|\vec{b}|},\vec{b}$. The first instinct is to copy the scalar-projection denominator and tack on $\vec{b}$ — but that multiplies a length by a full-length vector and over-scales the result by a factor of $|\vec{b}|$.
The correct way: the vector projection divides by $|\vec{b}|^2$. Deriving it once as "scalar projection times the unit vector $\hat{b}$" — rather than memorising a denominator — means you never lose track of which power of $|\vec{b}|$ belongs where.
Mistake 2: Projecting onto the wrong vector
Where it slips in: the phrase "projection of $\vec{a}$ onto $\vec{b}$."
Don't do this: swapping the roles and computing the projection of $\vec{b}$ onto $\vec{a}$. The projection is not symmetric — $\text{proj}{\vec{b}},\vec{a} \neq \text{proj}{\vec{a}},\vec{b}$ in general — and the memorizer who knows the formula but not which vector is "the direction" lands on the wrong one.
The correct way: the vector after the word "onto" is the direction $\vec{b}$ — it goes in the denominator and supplies the final direction. The vector being projected, $\vec{a}$, is the one whose shadow you want.
Mistake 3: Dropping the sign of a negative projection
Where it slips in: obtuse-angle cases where $\vec{a}\cdot\vec{b}$ is negative.
Don't do this: taking an absolute value out of habit and reporting the projection as positive. A negative scalar projection is information — it says the component points opposite to $\vec{b}$.
The correct way: keep the sign. As Example 4 shows, a negative result means the angle is obtuse and the projection runs against $\vec{b}$.
The everyday version of Mistake 3 is the ramp problem in physics: ignore the sign or direction of the component and you'll predict a block sliding up a slope that's actually pulling it down. The math is the same as the inclined-plane force split — and getting the direction wrong there is how a calculation that looks tidy sends a load the wrong way.
Conclusion
The projection vector of $\vec{a}$ onto $\vec{b}$ is $\text{proj}_{\vec{b}},\vec{a} = \left(\dfrac{\vec{a}\cdot\vec{b}}{|\vec{b}|^2}\right)\vec{b}$.
The scalar projection $\dfrac{\vec{a}\cdot\vec{b}}{|\vec{b}|}$ is a signed number; the vector projection attaches that number to $\hat{b}$.
A negative projection means an obtuse angle — the component points opposite to $\vec{b}$; a zero projection means the vectors are perpendicular.
Projection is not symmetric: projecting $\vec{a}$ onto $\vec{b}$ differs from projecting $\vec{b}$ onto $\vec{a}$.
The most common mistake is using $|\vec{b}|$ instead of $|\vec{b}|^2$ in the vector-projection formula.
A Practical Next Step
Practice these problems to solidify your understanding: pick two vectors, compute both the scalar and the vector projection, and check the sign against the angle between them. If the vector-projection denominator trips you up, return to Example 2 and the derivation. From here, the angle between vectors and the vectors overview round out the toolkit.
Want a live Bhanzu trainer to walk through more projection-vector problems? Book a free demo class.
Was this article helpful?
Your feedback helps us write better content