User Guide

10 Understandin

Optimization Principles and Options

more of the parameters are limited by the specified upper or

lower bound for that parameter. In other words, the optimizer

finds a solution (if one exists) even if one or more parameters are

at their limit.

However, solving this kind of problem is intrinsically more

difficult than performing unconstrained minimization.

If one or more parameters appear to be limited during the

optimization run, and you don’t expect the final solution to have

limited parameters, you could save time by using one or both of

the following techniques:

• Use a starting point that is further from the parameter limits.

• Loosen the limits on the parameter(s) in question.

Derivatives

To perform optimization, the PSpice Optimizer computes the

matrix of partial derivatives—the Jacobian.

How the PSpice Optimizer

Estimates Derivatives

The PSpice Optimizer approximates derivatives using a finite

difference approach. In one-dimensional terms, this method

computes an approximation to the first derivative of a function

f(x) by:

where h is a small perturbation.

The optimizer organizes the simulations and evaluations to

compute the Jacobian in the most efficient way possible.

The PSpice Optimizer calculates

derivatives either:

• once when you select Update

Derivatives from the Tune

menu, or

• automatically for each

iteration when you start

optimization by selectin

Auto from the Tune menu.

See

Exploring the Effect of

Parameter and Specification

Changes on page 3-20 for more

information.

′

()

fx h

()

–

----------------------------------

≅