Back to the on-screen lesson ·

Standard form

Slacks, surpluses, split free variables and the change of sense that put any program into one shape.

Paper packet. Every task here also exists on screen, where it is checked automatically; answers written on paper are not assessed by Nydus. When you are back at a device, enter your answers there.

1. What you will learn

By the end of this lesson you will be able to convert any linear program to standard form by adding a slack to each less-than constraint, subtracting a surplus from each greater-than constraint, splitting each free variable into a difference of two non-negative ones and negating the objective to change its sense, and to say how many variables and equations the result has.

2. What you already have

You can write a model as $A$, $b$ and $c$, and you can rearrange an inequality. This lesson is a short list of mechanical rewrites that put any model into the one shape the simplex method is written for.

3. Slack, surplus and free

A slack is a non-negative variable added to a $\le$ constraint to make it an equation. A surplus is subtracted from a $\ge$ constraint for the same reason. A free variable is one the problem allows to be negative. Standard form is a program whose constraints are all equations and whose variables are all non-negative.

4. One shape for every program

The simplex method is written for programs of one shape: minimise (or maximise) $c^{T}x$ subject to $Ax = b$ and $x \ge 0$. Any linear program reaches that shape by four rewrites, none of which changes which plans are feasible or which is best.

A $\le$ row gains a slack: $a^{T}x \le b$ becomes $a^{T}x + s = b$, $s \ge 0$. A $\ge$ row gives up a surplus: $a^{T}x - s = b$. A free variable splits: $y = u - v$ with $u, v \ge 0$. A maximisation becomes a minimisation by negating $c$; the same $x$ wins, and the optimal values differ only in sign.

What none of them can do is impose integrality. That is not an oversight in the list — it is the line between this subject and unit 5.

Another way: steps

  1. Split every free variable.
  2. Add a slack to every $\le$ row.
  3. Subtract a surplus from every $\ge$ row.
  4. Negate $c$ if the sense is wrong.

Another way: example

$\max 30x_1 + 50x_2$ with $2x_1 + 5x_2 \le 100$ and $3x_1 + 2x_2 \le 60$ becomes $2x_1 + 5x_2 + s_1 = 100$, $3x_1 + 2x_2 + s_2 = 60$, with all four variables non-negative: two equations in four unknowns.

5. Where this usually goes wrong

The surplus is the trap. A $\ge$ row needs a variable subtracted, and adding one instead quietly reverses the constraint. The second trap is a free variable handled by adding $y \ge 0$, which forbids answers the original problem allows — the rewrite has to preserve the feasible set exactly, or the answer is to a different question.

6. A diet problem reaching standard form

  1. $\min 2x_1 + 3x_2$ with $x_1 + 2x_2 \ge 8$ and $3x_1 + x_2 \ge 10$: two $\ge$ rows, so two surpluses.

    Requirements point the other way.

  2. $x_1 + 2x_2 - s_1 = 8$ and $3x_1 + x_2 - s_2 = 10$, with every variable non-negative.

    Subtract, do not add.

  3. Setting $x = 0$ now gives $s_1 = -8$, which is not allowed — so the slack columns are not a feasible starting basis here. That is what phase one exists for.

    Standard form does not guarantee a start.

7. Your turn: put $\max 4x - y$ subject to $x + y \le 6$, $y$ free, into standard form

  1. Write $y = u - v$ with $u, v \ge 0$, so the objective is $4x - u + v$ and the constraint is $x + u - v \le 6$.

    Split first.

  2. Your turn: work this step out. Its working is at the end of the packet.

    Then add a slack: $x + u - v + s = 6$, with $x, u, v, s \ge 0$.

8. Guided practice

A program has $5$ constraints of the form $\le$, $4$ of the form $\ge$ and $1$ equalities. Fill in how many new variables each kind needs.

New variables
From the $\le$ constraints
From the $\ge$ constraints
From the equalities
New variables in total

9. Guided practice

A program has $2$ decision variables, $5$ constraints of the form $\le$ and $2$ of the form $\ge$. How many variables does its standard form have?

Answer:

10. Guided practice

The wood constraint is $4x + 6y \le 44$. Write its slack $s$ in terms of $x$ and $y$.

Answer:

11. Practice

A program with $8$ variables has one of them, $y$, free: it may come out positive or negative. How does standard form carry it?

12. Practice

In standard form, $7x + 2y + s_1 = k$ and $x + 7y + s_2 = m$. Write the coefficient matrix, columns in the order $x$, $y$, $s_1$, $s_2$.

This task has no paper form; do it on a device.

13. Somewhere new

A minimisation has $5$ variables, of which $1$ are free, and $5$ constraints of the form $\le$ and $1$ of the form $\ge$. Complete the count for its standard form.

After the splits the program has a original variables; adding slacks and surpluses brings it to b variables in c equations.

14. Lesson test

Lesson test: one question per skill, one attempt each, no hints. Your answers are checked when you submit.

15. Test question

A program has $2$ constraints of the form $\le$, $3$ of the form $\ge$ and $2$ equalities. Fill in how many new variables each kind needs.

New variables
From the $\le$ constraints
From the $\ge$ constraints
From the equalities
New variables in total

16. What you can do now

You can put any linear program into standard form and count the variables and equations it then has. Say in your own words why a greater-than constraint needs its new variable subtracted rather than added.

Working for the steps left to you

7. Your turn: put $\max 4x - y$ subject to $x + y \le 6$, $y$ free, into standard form, step 2