Canonical form

Canonical Form

Canonical form writes a standard-form LP relative to a chosen basis.

It expresses the basic variables in terms of the nonbasic variables.

This is the algebraic form used inside the simplex method.

Starting Point: Standard Form

Start with:

\[\min c^Tx\]

subject to:

\[Ax=b\] \[x\ge 0\]

where:

\[A\in\mathbb{R}^{m\times n}\]

and $A$ has full row rank.

Basis

Choose $m$ linearly independent columns of $A$.

These columns form a basis matrix:

\[B=[A_{B(1)}\ A_{B(2)}\ \cdots\ A_{B(m)}]\]

The variables corresponding to these columns are the basic variables:

\[x_B\]

The remaining variables are the nonbasic variables:

\[x_N\]

Partitioned System

After reordering variables conceptually, the constraint system can be written as:

\[Bx_B+A_Nx_N=b\]

Since $B$ is invertible:

\[x_B=B^{-1}b-B^{-1}A_Nx_N\]

This is the canonical form of the constraints relative to $B$.

Basic Solution

A basic solution is obtained by setting:

\[x_N=0\]

Then:

\[x_B=B^{-1}b\]

So:

\[x=\begin{bmatrix}x_B\\x_N\end{bmatrix} = \begin{bmatrix}B^{-1}b\\0\end{bmatrix}\]

If:

\[B^{-1}b\ge 0\]

then the basic solution is a basic feasible solution.

Objective in Canonical Form

Partition the cost vector into:

\[c_B\]

and:

\[c_N\]

The objective is:

\[c^Tx=c_B^Tx_B+c_N^Tx_N\]

Substitute:

\[x_B=B^{-1}b-B^{-1}A_Nx_N\]

Then:

\[c^Tx=c_B^TB^{-1}b+(c_N^T-c_B^TB^{-1}A_N)x_N\]

The vector:

\[c_N^T-c_B^TB^{-1}A_N\]

contains the reduced costs of the nonbasic variables.

Full Reduced-Cost Form

For all variables, reduced costs are:

\[\bar{c}^T=c^T-c_B^TB^{-1}A\]

Basic variables have zero reduced cost.

This is because:

\[B^{-1}B=I\]

Optimality Connection

For a minimization problem in standard form, a basis is optimal if:

\[B^{-1}b\ge 0\]

and:

\[\bar{c}\ge 0\]

The first condition means feasibility.

The second condition means no nonbasic variable can enter and decrease the objective.

Example Structure

Suppose:

\[A=[B\ A_N]\]

and:

\[x=\begin{bmatrix}x_B\\x_N\end{bmatrix}\]

Then canonical form is:

\[x_B=B^{-1}b-B^{-1}A_Nx_N\] \[z=c_B^TB^{-1}b+\bar{c}_N^Tx_N\] \[x_B,x_N\ge 0\]

Why Canonical Form Matters

Canonical form makes simplex steps visible.

It shows:

  • current basic feasible solution
  • nonbasic variables currently set to zero
  • reduced costs
  • which variables can enter the basis
  • how increasing a nonbasic variable changes basic variables

Standard Form vs Canonical Form

Form Purpose
Standard form common problem format: $\min c^Tx$, $Ax=b$, $x\ge 0$
Canonical form same problem expressed relative to a basis

Standard form is basis-independent.

Canonical form depends on the chosen basis.

Checklist

You understand canonical form if you can:

  • choose a basis matrix $B$
  • split variables into $x_B$ and $x_N$
  • derive $x_B=B^{-1}b-B^{-1}A_Nx_N$
  • obtain the basic solution by setting $x_N=0$
  • identify feasibility using $B^{-1}b\ge 0$
  • write reduced costs
  • state the minimization optimality test $\bar{c}\ge 0$

See Also

Exam checkpoint

For standard-form questions, use the course convention $\min c^Tx$ subject to $Ax=b$, $x\ge0$. Convert max objectives, add slack to $\le$, subtract surplus from $\ge$, and split free variables.

25

25
Ready to start
Canonical form
Session: 1 | Break: Short
Today: 0 sessions
Total: 0 sessions