Homogeneous Linear Systems

Numerical Methods

Source inspiration: (Mathew 2000-2019).

Background

A homogeneous linear system has the form

\[ A\mathbf{x}=\mathbf{0}. \tag{1} \]

The vector \(\mathbf{x}=\mathbf{0}\) always solves (1), so every homogeneous system has at least the trivial solution.

NoteDefinition - Reduced Row Echelon Form

A matrix is in reduced row echelon form (RREF) if:

  1. Every nonzero row has leading entry \(1\).
  2. Each leading \(1\) is the only nonzero entry in its column.
  3. Leading \(1\) positions move to the right as you move down rows.
  4. Any zero rows are grouped at the bottom.
NoteDefinition - Rank

The rank of a matrix \(A\), written \(\operatorname{rank}(A)\), is the number of nonzero rows in its RREF.

Because an augmented homogeneous matrix is

\[ M=[A\mid \mathbf{0}], \tag{2} \]

the last column is always zero, so

\[ \operatorname{rank}(A)=\operatorname{rank}(M). \tag{3} \]

TipTheorem - Solution Structure for \(n\times n\) Homogeneous Systems

For \(A\in\mathbb{R}^{n\times n}\):

\[ \operatorname{rank}(A)=n \Longleftrightarrow \text{the only solution is }\mathbf{x}=\mathbf{0}, \tag{4} \]

\[ \operatorname{rank}(A)<n \Longleftrightarrow \text{there are infinitely many nontrivial solutions}. \tag{5} \]

For square matrices, this is equivalent to:

\[ \det(A)\neq 0 \Rightarrow \text{unique trivial solution}, \qquad \det(A)=0 \Rightarrow \text{infinitely many solutions}. \tag{6} \]

Algorithm

  1. Form \([A\mid\mathbf{0}]\).
  2. Row reduce to RREF.
  3. Identify pivot variables and free variables.
  4. Express pivot variables in terms of free variables.
  5. Write the parametric vector form and verify by substitution into \(A\mathbf{x}=\mathbf{0}\).

Example 1 - Unique Trivial Solution

Solve

\[ \begin{aligned} x_1+x_2-2x_3&=0,\\ 3x_1+2x_2+4x_3&=0,\\ 4x_1+3x_2+3x_3&=0. \end{aligned} \]

Step 1: Build the coefficient matrix and row reduce.

\[ A= \begin{pmatrix} 1&1&-2\\ 3&2&4\\ 4&3&3 \end{pmatrix} \longrightarrow \operatorname{RREF}(A)=I_3. \]

Step 2: Read off solution.

\[ x_1=x_2=x_3=0, \qquad \mathbf{x}=\begin{pmatrix}0\\0\\0\end{pmatrix}. \]

Verification:

Row Substitution result Target
1 \(0+0-2(0)=0\) \(0\ \checkmark\)
2 \(3(0)+2(0)+4(0)=0\) \(0\ \checkmark\)
3 \(4(0)+3(0)+3(0)=0\) \(0\ \checkmark\)

Since \(\det(A)=-1\neq0\), only the trivial solution exists.

Example 2 - One Free Variable

Solve

\[ \begin{aligned} x_1+x_2-2x_3&=0,\\ 3x_1+2x_2+4x_3&=0,\\ 4x_1+3x_2+2x_3&=0. \end{aligned} \]

Step 1: Row reduce.

\[ \operatorname{RREF}(A)= \begin{pmatrix} 1&0&8\\ 0&1&-10\\ 0&0&0 \end{pmatrix}. \]

So

\[ x_1+8x_3=0, \qquad x_2-10x_3=0. \]

Step 2: Let the free variable \(x_3=t\).

\[ x_1=-8t, \qquad x_2=10t, \qquad x_3=t. \]

Hence

\[ \mathbf{x}(t)=t \begin{pmatrix} -8\\10\\1 \end{pmatrix}, \qquad t\in\mathbb{R}. \]

Verification:

Row Substitution with \((-8t,10t,t)\) Target
1 \(-8t+10t-2t=0\) \(0\ \checkmark\)
2 \(3(-8t)+2(10t)+4t=0\) \(0\ \checkmark\)
3 \(4(-8t)+3(10t)+2t=0\) \(0\ \checkmark\)

Example 3 - Two Free Variables

Solve

\[ \begin{aligned} 4x_1+3x_2+2x_3&=0,\\ 4x_1+3x_2+2x_3&=0,\\ 4x_1+3x_2+2x_3&=0. \end{aligned} \]

Step 1: Row reduce.

\[ \operatorname{RREF}(A)= \begin{pmatrix} 1&\tfrac34&\tfrac12\\ 0&0&0\\ 0&0&0 \end{pmatrix}. \]

So

\[ x_1+\tfrac34x_2+\tfrac12x_3=0. \]

Step 2: Let \(x_2=s\) and \(x_3=t\) be free.

\[ x_1=-\tfrac34 s-\tfrac12 t, \qquad x_2=s, \qquad x_3=t. \]

Parametric vector form:

\[ \mathbf{x}(s,t)= s\begin{pmatrix}-\tfrac34\\1\\0\end{pmatrix} + t\begin{pmatrix}-\tfrac12\\0\\1\end{pmatrix}. \]

Verification:

Row Substitution result Target
1 \(4(-\tfrac34 s-\tfrac12 t)+3s+2t=0\) \(0\ \checkmark\)
2 same as row 1 \(0\ \checkmark\)
3 same as row 1 \(0\ \checkmark\)

When \(\operatorname{rank}(A)<n\), free variables are unavoidable. The solution set is a subspace (a line, plane, or higher-dimensional analogue through the origin), not a single point.

Application - Balancing Propane Combustion

Write

\[ x_1\,\mathrm{C_3H_8}+x_2\,\mathrm{O_2} \rightarrow x_3\,\mathrm{CO_2}+x_4\,\mathrm{H_2O}. \]

Atom-balance equations give the homogeneous system

\[ \begin{aligned} 3x_1-x_3&=0,\\ 8x_1-2x_4&=0,\\ 2x_2-2x_3-x_4&=0. \end{aligned} \]

From RREF,

\[ x_1=\tfrac14x_4, \qquad x_2=\tfrac54x_4, \qquad x_3=\tfrac34x_4. \]

Let \(x_4=t\):

\[ \mathbf{x}(t)= \begin{pmatrix} \tfrac14 t\\ \tfrac54 t\\ \tfrac34 t\\ t \end{pmatrix}. \]

Choose \(t=4\) for the smallest positive integers:

\[ (x_1,x_2,x_3,x_4)=(1,5,3,4). \]

Balanced equation:

\[ \mathrm{C_3H_8}+5\mathrm{O_2} \rightarrow 3\mathrm{CO_2}+4\mathrm{H_2O}. \]

Verification (atom counts):

Element Left side Right side
C \(3(1)=3\) \(1(3)=3\ \checkmark\)
H \(8(1)=8\) \(2(4)=8\ \checkmark\)
O \(2(5)=10\) \(2(3)+1(4)=10\ \checkmark\)

Summary

Homogeneous systems are classified by rank (or determinant in square cases):

  1. Full rank gives only the trivial solution.
  2. Rank deficiency creates free variables and infinitely many solutions.
  3. RREF exposes the null-space basis directly, which is useful both in algebra and in applications such as chemical equation balancing.

References

Mathew, John H. 2000-2019. Numerical Analysis - Numerical Methods Modules. https://web.archive.org/web/20190808102217/http://mathfaculty.fullerton.edu/mathews/n2003/NumericalUndergradMod.html.