Splitting Vectors Into Parallel and Perpendicular Parts

Why Split a Vector?

A vector can combine motion or displacement in several directions. If one direction is singled out, it is often useful to separate the part of the vector that runs along that direction from the part that points across it. The first part is parallel to the chosen direction; the second is perpendicular to it.

This separation is useful because the two components answer different questions. The parallel component records how much of the vector lies along the chosen direction, including whether it points with or against that direction. The perpendicular component records what remains after that parallel contribution is removed.

In this tutorial, the chosen direction is represented by a nonzero vector \(\mathbf{u}\), and the vector being split is \(\mathbf{v}\). The decomposition works in two or three dimensions, and the same formulas apply in any Euclidean coordinate space.

The Parallel and Perpendicular Components

For vectors \(\mathbf{a}\) and \(\mathbf{b}\), the dot product is computed by multiplying corresponding components and adding the results. The squared length of a vector is its dot product with itself. A vector is perpendicular to a nonzero vector exactly when their dot product is zero.

Assume \(\mathbf{u}\ne\mathbf{0}\). Then \(\mathbf{u}\cdot\mathbf{u}>0\), so the denominator in the following formula is nonzero. The projection of \(\mathbf{v}\) onto \(\mathbf{u}\) is its parallel component.

The coefficient multiplying \(\mathbf{u}\) is a scalar. It tells us how far to scale \(\mathbf{u}\) to obtain the parallel component. A positive coefficient gives a component pointing in the same direction as \(\mathbf{u}\); a negative coefficient gives one pointing in the opposite direction. A zero coefficient means the original vector is already perpendicular to \(\mathbf{u}\).

The formula does not require \(\mathbf{u}\) to have length one. In fact, its denominator is \(\mathbf{u}\cdot\mathbf{u}\), not the length of \(\mathbf{u}\). This makes the decomposition independent of which nonzero scale is used to represent the chosen direction.

Why the Remainder Is Perpendicular

The key fact is that subtracting the parallel component leaves a vector whose dot product with \(\mathbf{u}\) is zero. This follows directly from the definition and the linearity of the dot product.

Because \(\mathbf{u}\ne\mathbf{0}\), a zero dot product means that \(\mathbf{v}_{\perp}\) is perpendicular to \(\mathbf{u}\). Also, \(\mathbf{v}_{\parallel}\) is a scalar multiple of \(\mathbf{u}\), so it is parallel to \(\mathbf{u}\) (or is the zero vector). Together with \(\mathbf{v}=\mathbf{v}_{\parallel}+\mathbf{v}_{\perp}\), these facts establish that the formula really does split \(\mathbf{v}\) into the required parts.

Worked Examples

The Decomposition Is Unique

The formula produces a valid split, but it is also the only possible split with a parallel part in the direction of \(\mathbf{u}\) and a perpendicular remainder. This matters: the component vectors are determined by \(\mathbf{v}\) and the chosen direction, not by an arbitrary choice in the calculation.

To prove uniqueness, suppose \(\mathbf{v}=a\mathbf{u}+\mathbf{w}\), where \(a\) is a scalar and \(\mathbf{w}\cdot\mathbf{u}=0\). Take the dot product of both sides with \(\mathbf{u}\). Since \(\mathbf{w}\cdot\mathbf{u}=0\), this gives

Because \(\mathbf{u}\ne\mathbf{0}\), its squared length \(\mathbf{u}\cdot\mathbf{u}\) is positive. We can divide by it and obtain \(a=\dfrac{\mathbf{v}\cdot\mathbf{u}}{\mathbf{u}\cdot\mathbf{u}}\). Thus the parallel part must be \(a\mathbf{u}=\mathbf{v}_{\parallel}\). The equation \(\mathbf{v}=\mathbf{p}+\mathbf{q}\) then forces \(\mathbf{q}=\mathbf{v}-\mathbf{p}=\mathbf{v}_{\perp}\). No other pair can meet the conditions, completing the proof.

Special Cases and a Common Error

If \(\mathbf{v}\cdot\mathbf{u}=0\), the projection formula gives \(\mathbf{v}_{\parallel}=\mathbf{0}\), and the entire vector is perpendicular to the chosen direction. If \(\mathbf{v}\) is itself a scalar multiple of \(\mathbf{u}\), subtracting its projection leaves \(\mathbf{v}_{\perp}=\mathbf{0}\). These are useful checks, but the general formulas work in both cases as well.

A common mistake is to use the length of \(\mathbf{u}\) as the denominator in the vector projection formula. For example, with \(\mathbf{v}=\langle 3,1,2\rangle\) and \(\mathbf{u}=\langle 1,1,0\rangle\), the dot product is \(4\) and the length of \(\mathbf{u}\) is \(\sqrt{2}\). Using \(4/\sqrt{2}=2\sqrt{2}\) as the coefficient would give the supposed parallel part \(\langle 2\sqrt{2},2\sqrt{2},0\rangle\). The resulting remainder is not perpendicular, since

The correction is to divide by \(\mathbf{u}\cdot\mathbf{u}=2\), the squared length, not by \(\|\mathbf{u}\|=\sqrt{2}\). The correct coefficient is \(4/2=2\), giving parallel component \(\langle 2,2,0\rangle\) and perpendicular remainder \(\langle 1,-1,2\rangle\), whose dot product with \(\mathbf{u}\) is zero.

How to Organize a Decomposition

For a reliable calculation, keep the two components distinct and verify their defining properties. The following sequence applies whenever the specified direction vector is nonzero.

Why the Split Matters

The decomposition isolates the contribution of a vector along a selected direction without discarding the rest of the vector. This is valuable in geometry and in later applications where a direction has a particular role. The dot product determines the signed amount along that direction; subtraction preserves the remainder, and the zero dot product confirms that this remainder is perpendicular.

It is important not to confuse the parallel component with the whole vector, or the coefficient with the parallel vector itself. The coefficient is a number, while the parallel component is that number times \(\mathbf{u}\). A complete answer to a decomposition problem gives both vector components and shows that their sum and perpendicularity condition are satisfied.