You are on page 1of 12

Practice Problem

Specify the categories of the following. Functions as (strictly convex, convex, concave and strictly

Nonlinear programming (NLP)

 A nonlinear programing is similar to a linear programing in that it is composed of an
objective function, general constraints, and variable bounds. The difference is that a
nonlinear program includes at least one nonlinear function, which could be the
objective function, or some or all of the constraints.

 We often encounter problems that cannot be solved by LP algorithms.

 Solving non-linear programming models are generally much complicated than
linear programming models based on the types of non-linear models. For example:
 The objective is linear but constraints are non-linear
 Non-linear objectives and linear constraints
 Non-linear objectives and constraints

Non Linear Programming

Lagrangian multiplier

× is a vector of n decision variables (×1, . . . , ×n), f is the objective function,

and the gi, are constraint functions.
The ai, and bi are specified lower and upper bounds on the constraint functions
with ai ≤ bi and lj, uj are lower and upper bounds on the variables with lj ≤ uj
If the upper and lower limits on gi correspond to ai = −∞ and bi = +∞, the
constraint is unbounded

Similarly, with lj = uj corresponding to a variable xj whose value is fixed, and

lj = −∞ and uj = +∞ free variable.

This is nonlinear if one or more of the functions f , g1, g2, . . . , gm are


Methods of Solving NLP

Lagrange multiplier method

Direct Substitution Non Linear Programming
Quadratic programming Lagrangian multiplier

Penalty function and augmented Lagrangian method


Lagrangian Multiplier Method
Example: Consider the following constrained optimization

Non Linear Programming

Lagrangian multiplier

The gradient of the objective function ∇f (x∗) is orthogonal to the

tangent plane of the constraint at x∗. In general ∇h(x∗) is always
orthogonal to this tangent plane, hence ∇f (x∗) and ∇h(x∗) are
collinear, that is, they lie on the same line but point in opposite
directions. This means the two vectors must be multiples of each

8 Lagrangian Multiplier Method

We now introduce a new function L(x, λ) called the Lagrangian function:

Non Linear Programming

Then (NLP)
Lagrangian multiplex

so the gradient of the Lagrangian function with respect to x, evaluated

at (x∗, λ), is zero, plus the feasibility condition constitute the first-
order necessary conditions for optimality. The scalar λ is called a
Lagrange multiplier.

9 Lagrangian Multiplier Method

Using the necessary conditions to find the optimum of the above example

Non Linear Programming

Setting the first partial derivatives of L with respect to x and λ to zero, we get (NLP)
Lagrangian multiplier

10 Exercise

Non Linear Programming

Lagrangian multiplier


Maximize 𝒇 𝒙, 𝒚 = 𝒙𝒚
Subjected to 𝒉 𝒙, 𝒚 = 𝒙 + 𝒚 − 𝟏𝟐
Non Linear Programming
Lagrangian multiplier

By Using Lagrangian Multiplier Method


Comments &