How to Write a Parametric Solution from RREF: Step-by-Step

Written byDeb·Published on September 2026

A parametric solution from RREF expresses every solution to a linear system in terms of one or more free parameters.

You identify the pivot and non-pivot columns, assign a parameter to each free variable, and then write every basic variable in terms of those parameters.

A nontrivial parametric family occurs when two conditions hold:

If the system is inconsistent, there is no solution to parameterize.

If the system is consistent but every variable column contains a pivot, the system has a unique solution and no free parameters are needed.

What Makes a Solution Parametric

Suppose a system has coefficient matrix A with n variables and rank r.

Then:

number of basic variables = r

and:

number of free variables = n - r

Each pivot column of the coefficient matrix corresponds to a basic variable. Each non-pivot coefficient column corresponds to a free variable.

In a consistent system, each free variable may be chosen independently. We represent these choices using parameters such as t, s, or r.

The basic variables are then determined in terms of those parameters.

One free variable gives one parameter. Two free variables give two independent parameters, and so on.

Therefore:

Parametric form is used in the second case.

Reading RREF: Pivot Columns vs Non-Pivot Columns

Before writing the parametric equations, identify the pivot and non-pivot columns.

Consider the RREF of an augmented matrix:

[A | b]

Work first with the coefficient columns, not the augmented column.

In RREF:

The corresponding variables are classified as:

Pivot column → basic variable
Non-pivot column → free variable

The augmented column does not correspond to a variable.

If the augmented column itself contains a pivot, the RREF contains a row of the form:

[ 0   0   ...   0  |  1 ]

which represents:

0 = 1

The system is inconsistent, so no parametric solution exists.

The General Rule for Writing a Parametric Solution

This is the general pattern behind every example.

Suppose one row of the RREF represents the basic variable xi and has the form:

xi + a1xf1 + a2xf2 + ... + akxfk = bi

where xf1, xf2, ..., xfk are free variables.

Assign one parameter to each free variable:

xf1 = t1
xf2 = t2
...
xfk = tk

Then isolate the basic variable:

xi = bi - a1t1 - a2t2 - ... - aktk

Do this for every basic variable.

The entire solution can then be written in vector form as:

x = xp + t1v1 + t2v2 + ... + tkvk

where:

This is the general RREF-to-parametric-form procedure.

Writing the Parametric Equations

The Main Worked Example

Consider:

x1 + x2 + 2x3 = 5
x1 - x2 + 4x3 = 3

Its augmented matrix is:

[ 1   1   2  |  5 ]
[ 1  -1   4  |  3 ]

Apply:

R2 = R2 - R1

[ 1   1   2  |  5 ]
[ 0  -2   2  | -2 ]

Then:

R2 = (-1/2)R2

[ 1   1   2  |  5 ]
[ 0   1  -1  |  1 ]

Finally:

R1 = R1 - R2

[ 1   0   3  |  4 ]
[ 0   1  -1  |  1 ]

This is RREF.

Columns 1 and 2 are pivot columns, so x1 and x2 are basic variables.

Column 3 is a non-pivot column, so x3 is free.

Set:

x3 = t, where t can be any real number.

Now read each row.

The first row gives:

x1 + 3x3 = 4

Therefore:

x1 = 4 - 3t

The second row gives:

x2 - x3 = 1

Therefore:

x2 = 1 + t

The parametric solution is:

x1 = 4 - 3t
x2 = 1 + t
x3 = t

The Sign Change Rule

A common mistake occurs when students isolate the basic variables.

From:

x1 + 3x3 = 4

you subtract 3x3 from both sides:

x1 = 4 - 3x3

Since x3 = t:

x1 = 4 - 3t

Likewise:

x2 - x3 = 1

becomes:

x2 = 1 + x3 = 1 + t

So the safest rule is not to think of coefficients as mysteriously "changing sign." Instead:

Move all free-variable terms to the other side while isolating the basic variable.

Algebraically, that means their coefficients appear with the opposite sign in the expression for the basic variable.

Writing the Solution in Vector Form

From:

x1 = 4 - 3t
x2 = 1 + t
x3 = t

collect the constant terms and the coefficients of t:

(x1, x2, x3) = (4, 1, 0) + t(-3, 1, 1)

So:

x = (4, 1, 0) + t(-3, 1, 1)

The vector:

(4, 1, 0)

is a particular solution. It is obtained by setting:

t = 0

The vector:

(-3, 1, 1)

is the direction vector associated with the free variable.

As t varies over all real numbers, the solution set traces an affine line through (4, 1, 0).

Because the original system is nonhomogeneous with b ≠ 0, the zero vector is not a solution, so this affine line does not pass through the origin.

You can verify the direction vector by checking that it satisfies the corresponding homogeneous equations:

(-3) + (1) + 2(1) = 0
(-3) - (1) + 4(1) = 0

Thus, (-3, 1, 1) lies in the null space of the coefficient matrix.

How to Build Direction Vectors Directly from RREF

For each free variable, construct one direction vector.

Suppose xfj is the free variable associated with parameter tj.

To construct its direction vector vj:

This rule works even when pivot columns are not consecutive.

For the previous example:

x1 = 4 - 3t
x2 = 1 + t
x3 = t

the coefficients of t give:

v = (-3, 1, 1)

Every such direction vector lies in the null space of the coefficient matrix.

Two Free Variables: A Second Worked Example

Consider:

x1 + 2x2 - x3 = 3

Its augmented matrix is already in RREF:

[ 1   2  -1  |  3 ]

Column 1 is a pivot column, so x1 is basic.

Columns 2 and 3 are non-pivot columns, so x2 and x3 are free.

Assign separate parameters:

x2 = s
x3 = t

From the equation:

x1 + 2x2 - x3 = 3

isolate x1:

x1 = 3 - 2x2 + x3

Substitute the parameters:

x1 = 3 - 2s + t

Therefore:

x1 = 3 - 2s + t
x2 = s
x3 = t

Now separate the constant, s, and t terms:

(x1, x2, x3) = (3, 0, 0) + s(-2, 1, 0) + t(1, 0, 1)

For the s-direction:

v1 = (-2, 1, 0)

Indeed:

-2 + 2(1) - 0 = 0

For the t-direction:

v2 = (1, 0, 1)

And:

1 + 2(0) - 1 = 0

Both direction vectors lie in the null space of the coefficient matrix.

The solution set is an affine plane in R³, anchored at:

(3, 0, 0)

and parallel to the plane spanned by:

(-2, 1, 0)

and:

(1, 0, 1)

Parametric Solutions in Homogeneous Systems

A homogeneous system has the form:

Ax = 0

It is always consistent because:

x = 0

is always a solution.

For the same coefficient matrix as the first example, consider:

x1 + x2 + 2x3 = 0
x1 - x2 + 4x3 = 0

Its RREF is:

[ 1   0   3  |  0 ]
[ 0   1  -1  |  0 ]

Again:

x3 = t

Then:

x1 = -3t
x2 = t
x3 = t

So:

(x1, x2, x3) = t(-3, 1, 1)

For the same coefficient matrix, the direction vector is unchanged. The difference is that the zero vector can now serve as the particular solution.

The solution set is:

span{(-3, 1, 1)}

which is the null space of the coefficient matrix.

More generally, every consistent nonhomogeneous solution has the form:

x = xp + xh

where xp is one particular solution and xh is a solution of:

Ax = 0

In parametric form:

x = xp + t1v1 + ... + tkvk

and the vectors v1, ..., vk are the null-space directions produced by the free variables.

If you want to verify your parametric form, enter the augmented matrix into the row reduction calculator and compare the solution readout against what you derived by hand.