How to Prove a Homogeneous System Has Infinitely Many Solutions

A homogeneous linear system has zero on the right-hand side in every equation. If it has more unknowns than equations, it must have a nonzero solution. Over the real numbers, that immediately produces infinitely many solutions: multiply the nonzero solution by any real number.

The key tool is the rank-nullity theorem. It explains why having more columns than rows forces at least one free direction in the solution space.

Write the system as a matrix equation

Suppose the system has $m$ equations and $n$ unknowns, with $n>m$. Collect its coefficients into an $m\times n$ matrix $A$, and write the unknowns as a vector $x$. The system becomes

$$Ax=0.$$

The solutions are exactly the vectors in the null space (also called the kernel) of $A$.

Apply the rank-nullity theorem

For a matrix with $n$ columns, rank-nullity says

$$\dim(\ker A)+\operatorname{rank}(A)=n.$$

Because $A$ has only $m$ rows, its rank cannot exceed $m$. Therefore,

$$\dim(\ker A)=n-\operatorname{rank}(A)\ge n-m>0.$$

So the null space has positive dimension. In particular, it contains some vector $v\ne 0$ satisfying $Av=0$.

Now take any scalar $c$. Linearity gives

$$A(cv)=c(Av)=c0=0.$$

Thus every scalar multiple $cv$ is also a solution. If the scalars are real numbers, different values of $c$ give different vectors because $v\ne0$. There are infinitely many real values of $c$, so the system has infinitely many solutions.

This proves the claim without requiring the equations to be independent. Redundant equations can lower the rank, which only increases the dimension of the solution space.

Worked example

Consider the system with two equations and three unknowns:

$$\begin{aligned} x+y+z&=0,\\ 2x-y+z&=0. \end{aligned}$$

There are $n=3$ unknowns and $m=2$ equations, so the nullity is at least $3-2=1$. Solve directly to see the free parameter. Subtract the first equation from the second:

$$x-2y=0,\qquad x=2y.$$

Put $y=t$. Substitution into the first equation gives $z=-3t$. Hence all solutions have the form

$$(x,y,z)=t(2,1,-3),\qquad t\in\mathbb{R}.$$

For instance, $t=0$ gives $(0,0,0)$, $t=1$ gives $(2,1,-3)$, and $t=-1$ gives $(-2,-1,3)$. Every real value of $t$ gives a solution, and distinct values give distinct vectors.

The same idea using row reduction

When you row-reduce an $m\times n$ coefficient matrix with $n>m$, there can be at most $m$ pivot columns. Since there are more columns than pivots, at least one variable is free. In a homogeneous system, assigning values to the free variables determines the pivot variables, and choosing a nonzero free-variable value produces a nonzero solution.

This is the computational version of the rank-nullity proof. It is useful when you want an explicit description of the solutions, rather than just a proof that nonzero solutions exist.

Common mistakes and qualifications

• Confusing a nonzero solution with infinitely many solutions. The zero vector is always a solution, but that fact alone does not show there are others. The dimension argument establishes a nonzero solution; scalar multiples then establish infinitely many over an infinite field.
• Assuming every equation is independent. Independence is unnecessary. The rank is at most the number of equations, whether or not some equations repeat or follow from others.
• Ignoring the number system. The conclusion “infinitely many” assumes the variables range over an infinite field, such as $\mathbb{R}$ or $\mathbb{C}$. Over a finite field, rank-nullity still guarantees a nonzero solution, but the total solution set is finite. For example, over the two-element field, a one-dimensional null space has just two vectors: zero and one nonzero vector.
• Forgetting the homogeneous condition. The argument uses $Ax=0$, which guarantees the zero vector is a solution and makes scalar multiples of any solution remain solutions. A nonhomogeneous system $Ax=b$ can be inconsistent, or it can have a different solution structure.

Under the usual assumption that the variables are real and unrestricted, more unknowns than equations is therefore enough to guarantee infinitely many solutions for a homogeneous system.