A projection operator is a linear operator that leaves its own output unchanged. If V is a vector space, a projection on V is a linear map
P:V→V
such that
P2=P.
Equivalently,
P(P(v))=P(v)
for every v∈V. Applying the projection once moves a vector into the target subspace. Applying it again has no further effect. This property is called idempotence. Projection operators are therefore exactly idempotent linear operators.
38.1 First Example
Define
P:R3→R3
by
Pxyz=xy0.
This map sends every vector to its shadow on the xy-plane.
Apply P twice:
P2xyz=Pxy0=xy0.
Thus
P2=P.
So P is a projection operator.
The image is the xy-plane:
im(P)=⎩⎨⎧xy0:x,y∈R⎭⎬⎫.
The kernel is the z-axis:
ker(P)=⎩⎨⎧00z:z∈R⎭⎬⎫.
The projection keeps the image and removes the kernel.
38.2 Idempotence
The defining equation
P2=P
means
P(P(v))=P(v)
for every vector v∈V.
This has a direct interpretation. The first application of P sends v into im(P). Once a vector is in im(P), the projection leaves it fixed.
Indeed, if w∈im(P), then there is some v∈V such that
w=P(v).
Then
P(w)=P(P(v))=P2(v)=P(v)=w.
Thus every vector in the image is fixed by P.
Conversely, if
P(w)=w,
then w∈im(P), since w is the image of itself under P. Hence
im(P)={w∈V:P(w)=w}.
The image of a projection is exactly its fixed subspace.
38.3 Kernel and Image
For every projection P:V→V, the kernel and image describe the whole operator.
The kernel is
ker(P)={v∈V:P(v)=0}.
The image is
im(P)={P(v):v∈V}.
Every vector v∈V can be decomposed as
v=P(v)+(v−P(v)).
The first term satisfies
P(v)∈im(P).
The second term lies in the kernel, because
P(v−P(v))=P(v)−P2(v)=P(v)−P(v)=0.
Therefore
v=P(v)+(v−P(v))
is a decomposition of v into a part in the image and a part in the kernel.
Moreover, this decomposition is unique. If
u∈im(P)∩ker(P),
then u∈im(P) implies
P(u)=u.
But u∈ker(P) implies
P(u)=0.
Thus
u=0.
So
im(P)∩ker(P)={0}.
Hence
V=im(P)⊕ker(P).
Every projection gives a direct sum decomposition of the vector space.
38.4 Projection Onto a Subspace Along a Complement
Let V be a vector space, and suppose
V=U⊕W.
This means every vector v∈V has a unique decomposition
v=u+w,
where
u∈U,w∈W.
Define
P:V→V
by
P(v)=u.
That is, P keeps the U-component and discards the W-component.
Then P is linear. If
v1=u1+w1,v2=u2+w2,
then
v1+v2=(u1+u2)+(w1+w2),
so
P(v1+v2)=u1+u2=P(v1)+P(v2).
For a scalar c,
cv=cu+cw,
so
P(cv)=cu=cP(v).
Also,
P2(v)=P(u)=u=P(v).
Thus P is a projection.
Its image is U, and its kernel is W:
im(P)=U,ker(P)=W.
So a projection is the same as a choice of direct sum decomposition.
38.5 The Complementary Projection
If P is a projection on V, then
I−P
is also a projection.
Compute:
(I−P)2=I−2P+P2.
Since
P2=P,
we get
(I−P)2=I−2P+P=I−P.
Thus I−P is idempotent.
The projection I−P keeps the part that P removes. For any vector v,
v=P(v)+(I−P)(v).
The image of I−P is the kernel of P:
im(I−P)=ker(P).
The kernel of I−P is the image of P:
ker(I−P)=im(P).
Thus P and I−P are complementary projections.
38.6 Matrix Projections
A square matrix P is called a projection matrix if
P2=P.
Such a matrix defines a projection operator
x↦Px.
For example,
P=[1000]
satisfies
P2=[1000][1000]=[1000]=P.
It projects R2 onto the x-axis:
P[xy]=[x0].
A projection matrix must be square because it represents a linear operator from a vector space to itself.
38.7 Orthogonal Projections
In an inner product space, an important special case is the orthogonal projection.
An orthogonal projection onto a subspace U writes each vector v as
v=u+w,
where
u∈U
and
w∈U⊥.
Then
P(v)=u.
The vector u is the closest vector in U to v. The difference
v−P(v)
is perpendicular to U.
For example, the projection from R3 onto the xy-plane is orthogonal because the removed part lies on the z-axis, which is perpendicular to the plane.
In real coordinate spaces, an orthogonal projection matrix satisfies
P2=P
and
PT=P.
Thus it is both idempotent and symmetric. For complex spaces, symmetry is replaced by self-adjointness:
P∗=P.
Orthogonal projection matrices are therefore characterized by
P2=P=PT
in the real case, and
P2=P=P∗
in the complex case.
38.8 Projection Onto a Line
Let u∈Rn be a nonzero vector. The orthogonal projection of x∈Rn onto the line spanned by u is
proju(x)=u⋅ux⋅uu.
The scalar
u⋅ux⋅u
is the coordinate of the projection along u.
The corresponding matrix is
P=uTuuuT.
Indeed,
Px=uTuuuTx=uuTuuTx=u⋅ux⋅uu.
This matrix is symmetric and idempotent, so it is an orthogonal projection matrix.
38.9 Example: Projection Onto a Line in R2
Let
u=[12].
Then
uTu=12+22=5.
Also,
uuT=[12][12]=[1224].
Therefore the projection matrix onto the line spanned by u is
P=51[1224].
For
x=[31],
we get
Px=51[1224][31]=51[510]=[12].
The vector x projects exactly onto u. The error vector is
x−Px=[31]−[12]=[2−1].
Check orthogonality:
[2−1]⋅[12]=2−2=0.
The error is perpendicular to the line.
38.10 Projection Onto a Subspace with an Orthonormal Basis
Let U be a subspace of Rn, and let
q1,…,qk
be an orthonormal basis of U. Then the orthogonal projection of x onto U is
PUx=(x⋅q1)q1+⋯+(x⋅qk)qk.
If Q is the matrix with columns
q1,…,qk,
then
QTQ=Ik.
The projection matrix is
P=QQT.
Indeed,
QQTx=Qq1Tx⋮qkTx=(q1Tx)q1+⋯+(qkTx)qk.
The matrix QQT is symmetric and idempotent:
(QQT)T=QQT,
and
(QQT)2=Q(QTQ)QT=QIQT=QQT.
38.11 Projection Onto a Column Space
Let A be an m×k matrix with linearly independent columns. The column space of A is a k-dimensional subspace of Rm.
The orthogonal projection onto col(A) is
P=A(ATA)−1AT.
This formula generalizes the line projection formula. When A has one column u, it becomes
P=u(uTu)−1uT=uTuuuT.
The matrix ATA is invertible because the columns of A are linearly independent.
The projection Px is the vector in col(A) closest to x, and the residual
x−Px
is orthogonal to every column of A. Projection formulas of this kind are central in least squares and regression.
38.12 Derivation of the Column Space Formula
We seek a vector in col(A) closest to x. Such a vector has the form
Ac^
for some coefficient vector c^.
The residual is
r=x−Ac^.
For Ac^ to be the orthogonal projection, r must be orthogonal to every column of A. This condition is
AT(x−Ac^)=0.
Expanding gives
ATx−ATAc^=0.
So
ATAc^=ATx.
Since ATA is invertible,
c^=(ATA)−1ATx.
Therefore
Px=Ac^=A(ATA)−1ATx.
Thus
P=A(ATA)−1AT.
38.13 Oblique Projections
A projection need not be orthogonal.
Suppose
V=U⊕W.
The projection onto U along W sends
u+w
to
u.
If W=U⊥, the projection is orthogonal. If W is another complement, the projection is oblique.
An oblique projection still satisfies
P2=P.
It still has
im(P)=U,ker(P)=W.
But the removed part w may not be perpendicular to U. Oblique projections are therefore algebraically valid projections, but they do not describe nearest-point projection in the usual Euclidean metric.
38.14 Example of an Oblique Projection
In R2, let
U=span{[10]}
be the x-axis, and let
W=span{[11]}.
Every vector
[xy]
can be written uniquely as
[xy]=[a0]+t[11].
From the second coordinate,
t=y.
From the first coordinate,
a+t=x,
so
a=x−y.
Thus the projection onto U along W is
P[xy]=[x−y0].
Its matrix is
P=[10−10].
Check idempotence:
P2=[10−10][10−10]=[10−10]=P.
This projection is not orthogonal because
PT=P.
It projects onto the x-axis along diagonal lines parallel to (1,1), not along vertical lines.
38.15 Eigenvalues of a Projection
Let P be a projection. If v is an eigenvector with eigenvalue λ, then
P(v)=λv.
Apply P again:
P2(v)=P(λv)=λP(v)=λ2v.
But P2=P, so
P2(v)=P(v)=λv.
Hence
λ2v=λv.
Since v=0,
λ2=λ.
Thus
λ(λ−1)=0.
Therefore every eigenvalue of a projection is either
0
or
1.
Vectors in the kernel have eigenvalue 0. Vectors in the image have eigenvalue 1. Projection matrices have only the eigenvalues 0 and 1.
38.16 Diagonal Form
Because the minimal polynomial of a projection divides
x2−x=x(x−1),
and this polynomial has distinct roots, every projection on a finite-dimensional vector space is diagonalizable.
More concretely, choose a basis
(u1,…,ur)
for im(P), and choose a basis
(w1,…,ws)
for ker(P).
Since
V=im(P)⊕ker(P),
the combined list
(u1,…,ur,w1,…,ws)
is a basis of V.
In this basis, P has matrix
[Ir000].
Thus a projection is structurally simple: it is identity on one subspace and zero on a complementary subspace.
38.17 Trace and Rank
For a finite-dimensional projection P, the trace equals the rank.
In the basis adapted to
V=im(P)⊕ker(P),
the matrix of P is
[Ir000],
where
r=dim(im(P)).
The trace is the sum of diagonal entries:
tr(P)=r.
The rank is also
rank(P)=r.
Therefore
tr(P)=rank(P).
This fact is often useful in matrix analysis, statistics, and numerical linear algebra.
38.18 Products of Projections
The product of two projections is not necessarily a projection.
Let P and Q be projections. The product PQ is a projection if
(PQ)2=PQ.
Compute:
(PQ)2=PQPQ.
If P and Q commute, meaning
PQ=QP,
then
(PQ)2=PQPQ=PPQQ=P2Q2=PQ.
Thus, if two projections commute, their product is also a projection.
Without commutativity, the product may fail to be idempotent. Products of projections therefore require care.
38.19 Projections and Least Squares
Projection operators are the algebraic core of least squares.
Given an inconsistent system
Ax=b,
there may be no exact solution. Instead, least squares seeks a vector x^ such that
Ax^
is as close as possible to b inside the column space of A.
This means
Ax^=Pcol(A)b,
where
Pcol(A)
is the orthogonal projection onto the column space of A.
The normal equations
ATAx^=ATb
come from the orthogonality condition
b−Ax^⊥col(A).
Thus least squares is not merely an approximation trick. It is an orthogonal projection problem.
38.20 Summary
A projection operator is a linear operator
P:V→V
satisfying
P2=P.
This means applying P twice is the same as applying it once.
Every projection decomposes the vector space as
V=im(P)⊕ker(P).
It acts as the identity on its image and as zero on its kernel.
If V=U⊕W, then the projection onto U along W is the map
u+w↦u.
Orthogonal projections occur in inner product spaces when the complement is perpendicular to the target subspace. In real coordinates, an orthogonal projection matrix satisfies
P2=P=PT.
Projection matrices have eigenvalues only 0 and 1, are diagonalizable, and satisfy
tr(P)=rank(P).
Projection operators are central in geometry, least squares, regression, numerical linear algebra, and the study of decompositions of vector spaces.
← → section · ↑ ↓ slide · Space next · F fullscreen · Esc exit