1 Approximations using partial derivatives
1.1 Functions of two variables
We saw in HELM booklet 16.5 how to expand a function of a single variable in a Taylor series:
This can be written in the following alternative form (by replacing by so that ):
This expansion can be generalised to functions of two or more variables:
where, assuming and to be small, we have ignored higher-order terms involving powers of and . We define to be the change in resulting from small changes to and , denoted by and respectively. Thus:
and so . Using the notation and instead of and for small increments in and respectively we may write
Finally, using the more common notation for partial derivatives, we write
Informally, the term is referred to as the absolute error in resulting from errors in the variables and respectively. Other measures of error are used. For example, the relative error in a variable is defined as and the percentage relative error is .
Key Point 5
Measures of Error
If is the change in at resulting from small changes to and respectively, then , and
The absolute error in is
The relative error in is .
The percentage relative error in is .
Note that to determine the error numerically we need to know not only the actual values of and but also the values of and at the point of interest.
Example 12
Estimate the absolute error for the function
Solution
Then
Task!
Estimate the absolute error for at the point if and . Compare the estimate with the exact value of the error.
First find and :
Now obtain an expression for the absolute error:
Now obtain the estimated value of the absolute error at the point of interest:
. Finally compare the estimate with the exact value:
The actual error is calculated from
.
We see that there is a reasonably close correspondence between the two values.
1.2 Functions of three or more variables
If is a function of several variables the error induced in as a result of making small errors in is found by a simple generalisation of the expression for two variables given above:
Example 13
Suppose that the area of triangle is to be calculated by measuring two sides and the included angle. Call the sides and and the angle .
Then the area of the triangle is given by
Now suppose that the side is measured as 4.00 m, as 3.00 m and as . Suppose also that the measurements of the sides could be in error by as much as m and of the angle by . Calculate the likely maximum error induced in as a result of the errors in the sides and angle.
Solution
Here is a function of three variables . We calculate
Now , and , so
Here and ( must be measured in radians). Substituting these values we see that the maximum error in the calculated value of is given by the approximation
Hence the estimated value of is in error by up to about .