BEngineering Mathematics 1 (MAT 2018) July 2020

Unit 2 Partial Differentiation

2.4 Stationary Points and Lagrange Multipliers

2.4.1 Stationary Points in 2D (Review)

A stationary point of a differentiable function, 𝑦 = 𝑓(𝑥) , is that point along the function where
the rate of change or gradient is zero. It is that point where the function is neither increasing nor
decreasing.

There are three (3) types of stationary points:

- Maximum
- Minimum
- Inflection

Note There exist non-stationary points of inflection.

1
2
2.4.2 Stationary Points in 3D

Given a bivariate function 𝑧 = 𝑓(𝑥, 𝑦), stationary points, if they exist, maybe classified as:

- Maximum
- Minimum
- Saddle point

To locate the stationary points:

• Find the first-order partial derivatives, 𝑓𝑥 and 𝑓𝑦

𝑓𝑥 = 0
• Equate each to zero and solve, i.e. {
𝑓𝑦 = 0
• The critical, or function, value may be computed by substituting the values of x
and y in the original bivariate function, 𝑧 = 𝑓(𝑥, 𝑦)

Example Locate the stationary point(s) of 𝑧 = 𝑓(𝑥, 𝑦) = 𝑥 2 + 𝑥𝑦 + 𝑦 2 + 5𝑥 − 5𝑦 + 3

Solution 𝑓𝑥 = 2𝑥 + 𝑦 + 5 = 0

𝑓𝑦 = 𝑥 + 2𝑦 − 5 = 0

These two simultaneous equations may be solved using any applicable method

(elimination, substitution, graphs, determinants).

The method applied here will be substitution:

3
 Making x as the subject in the second equation gives 𝑥 = 5 − 2𝑦

Substituting this into the first equation:

𝑓𝑥 = 2𝑥 + 𝑦 + 5 = 0

2(5 − 2𝑦) + 𝑦 + 5 = 0

10 − 4𝑦 + 𝑦 + 5 = 0

−3𝑦 = −15 → 𝑦 = 5

Since 𝑦 = 5 ; then 𝑥 = 5 − 2(5) = −5

The stationary point is not fully located…the function (or z - ) value is required.

𝑧 = 𝑓(−5,5) = (−5)2 + (−5)(5) + (5)2 + 5(−5) − 5(5) + 3

= 25 − 25 + 25 − 25 − 25 + 3 = −22

The stationary point’s location is given by (−5, 5, −22)

To determine the nature/ classify the critical points:

• Obtain the second-order partial derivatives, 𝑓𝑥𝑥 , 𝑓𝑦𝑦 and 𝑓𝑥

2
• The Discriminant is defined by 𝐷 = 𝑓𝑥𝑥 𝑓𝑦𝑦 − (𝑓𝑥𝑦 )
2
• If 𝐷 < 0, then 𝑓𝑥𝑥 𝑓𝑦𝑦 < (𝑓𝑥𝑦 ) and the point is classified as Saddle point
2
• If 𝐷 > 0, then 𝑓𝑥𝑥 𝑓𝑦𝑦 > (𝑓𝑥𝑦 ) , hence 𝑓𝑥𝑥 𝑓𝑦𝑦 > 0. There are two possibilities:
o 𝑓𝑥𝑥 > 0 and 𝑓𝑦𝑦 > 0, which is a minimum
o 𝑓𝑥𝑥 < 0 and 𝑓𝑦𝑦 < 0, which is maximum
• For 𝐷 = 0, no conclusion

Example Classify the stationary point of 𝑧 = 𝑓(𝑥, 𝑦) = 𝑥 2 + 𝑥𝑦 + 𝑦 2 + 5𝑥 − 5𝑦 + 3.

Solution This function had one stationary point at (−5, 5, −22)

4
 The second-order partial derivatives will be required

The first-order partial derivatives were 𝑓𝑥 = 2𝑥 + 𝑦 + 5 and 𝑓𝑦 = 𝑥 + 2𝑦 − 5

𝑓𝑥𝑥 = 2 ; 𝑓𝑦𝑦 = 2 ; 𝑓𝑥𝑦 = 1

2
Substituting in the discriminant gives 𝐷 = 𝑓𝑥𝑥 𝑓𝑦𝑦 − (𝑓𝑥𝑦 ) = (2)(2) − (1)2 = 3

Since 𝐷 > 0 with 𝑓𝑥𝑥 > 0 and 𝑓𝑦𝑦 > 0, the stationary point is a minimum

Example December 2017, Question 4 (a)

Given the bivariate function z = x 3 + y 2 − 3x − 6 y − 1 ,

i) Find the coordinates, (x, y, z ) , of the stationary points.

ii) Classify each of these points as either relative maximum, relative minimum

or saddle point. [6 +4 marks]

Solution To find stationary points, solve:

𝑧𝑥 = 3𝑥 2 − 3 = 0 ⇒ 𝑥 = −1, 1

𝑧𝑦 = 2𝑦 − 6 = 0 ⇒ 𝑦 = 3

If 𝑥 = −1, 𝑦 = 3, 𝑧 = −1 + 9 + 3 − 18 − 1 = −8

If 𝑥 = 1, 𝑦 = 3, 𝑧 = 1 + 9 − 3 − 18 − 1 = −12

To classify the stationary points:

𝑧𝑥𝑥 = 6𝑥 𝑧𝑦𝑦 = 2 𝑧𝑥𝑦 = 0

5
 2
@(- 1, 3, -8); 𝐷 = 𝑧𝑥𝑥 𝑧𝑦𝑦 − 𝑧𝑥𝑦 = (−6)(2) − 0 = −12 saddle point

2
@(1,3,-12); 𝐷 = 𝑧𝑥𝑥 𝑧𝑦𝑦 − 𝑧𝑥𝑦 = (6)(2) − 0 = 12 relative minimium

Example December 2018, Question 4 (a)

Determine the coordinates, (𝑥, 𝑦, 𝑧), of the two stationary points of the function
𝑧 = 𝑓(𝑥, 𝑦) = 3𝑥 2 + 12𝑥 + 4𝑦 3 − 6𝑦 2 + 5.

Classify each of the stationary points as either relative maximum,

relative minimum or saddle point. [6+4 marks]

Solution
𝑧𝑥 = 6𝑥 + 12 = 0; 𝑥 = −2

𝑧𝑦 = 12𝑦 2 − 12𝑦 = 0; 12𝑦(𝑦 − 1) = 0; 𝑦 = 0, 1

𝑧 = 𝑓(−2,0) = 3(−2)2 + 12(−2) + 4(0)3 − 6(0)2 + 5 = −17; (−2, 0, −17)

𝑧 = 𝑓(−2,1) = 3(−2)2 + 12(−2) + 4(1)3 − 6(1)2 + 5 = −9; (−2, 1, −9)

Second-order partial derivatives: 𝑧𝑥𝑥 = 6; 𝑧𝑥𝑦 = 0; 𝑧𝑦𝑦 = 24𝑦 − 12

Using 𝐷 = 𝑧𝑥𝑥 𝑧𝑦𝑦 − 𝑧𝑥𝑦 2

For (−2, 0, −17); 𝐷 = 6(−12) − 0 = −72; 𝑠𝑎𝑑𝑑𝑙𝑒 𝑝𝑜𝑖𝑛𝑡

For (−2, 1, −9); 𝐷 = 6(12) − 0 = 72; 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑝𝑜𝑖𝑛𝑡

6
Example December 2019, Question 3 (b)

Determine the coordinates, (𝑥, 𝑦, 𝑧), of the two (2) stationary points of the function
𝑧 = 𝑓(𝑥, 𝑦) = 𝑦 2 − 3𝑥 + 𝑥 3 − 6𝑦 − 1.

Classify each of the stationary points as either relative maximum, relative

minimum, or saddle point. [6+4 marks]

Solution

Location of stationary point:

𝑧𝑥 = −3 + 3𝑥 2 = 0; 𝑥 = ±1 and 𝑧𝑦 = 2𝑦 − 6 = 0; 𝑦 = 3

𝑧 = 𝑓(1, 3) = 9 − 3 + 1 − 18 − 1 = −12

𝑧 = 𝑓(−1, 3) = 9 + 3 − 1 − 18 − 1 = −8

Coordinates of SPs (1, 3, −12) and (−1, 3, −8)

Classification of stationary point: 𝑧𝑥𝑥 = 6𝑥; 𝑧𝑦𝑦 = 2; 𝑧𝑥𝑦 = 0

At (1, 3, −12); 𝐷 = (6𝑥)(2) − 0 = 12; minimum point

At (−1, 3, −12); 𝐷 = (6𝑥)(2) − 0 = −12; saddle point

2.4.3 Lagrange Multipliers

Lagrange multipliers are useful in the optimization of a multivariate function subject to

constraints; that is to maximize/ minimize z = f ( x, y) subject to the constraint g ( x, y) = 0 .

Example Determine the values of x and y that minimize

𝑧 = 𝑓(𝑥, 𝑦) = 𝑥 2 − 8𝑥 + 𝑦 2 − 12𝑦 + 48 subject to the constraint 𝑥 + 𝑦 = 8;

7
 where x, y  0 .

Solution The function to be minimized if 𝑧 = 𝑓(𝑥, 𝑦) = 𝑥 2 − 8𝑥 + 𝑦 2 − 12𝑦 + 48 and the

constraint is expressed as 𝑔(𝑥, 𝑦) = 𝑥 + 𝑦 − 8 = 0

➢ Find 𝑓𝑥 , 𝑓𝑦 , 𝑔𝑥 and 𝑔𝑦

𝑓𝑥 = 2𝑥 − 8
𝑓𝑦 = 2𝑦 − 12
𝑔𝑥 = 1
𝑔𝑦 = 1

➢ Now, solve: 𝑔(𝑥, 𝑦) = 0 𝑓𝑥 = 𝜆𝑔𝑥 𝑓𝑦 = 𝜆𝑔𝑦

*no need to find the multiplier, λ
2𝑥 − 8 = 𝜆
➢ { ⇒ 2𝑥 − 8 = 2𝑦 − 12
2𝑦 − 12 = 𝜆
➢ So, 2𝑥 + 4 = 2𝑦 →→→ 𝑥+2=𝑦
➢ Substitute 𝑦 = 𝑥 + 2 in the constraint: 𝑔(𝑥, 𝑦) = 𝑥 + 𝑦 − 8 = 0

𝑥 + (𝑥 + 2) − 8 = 0
2𝑥 − 6 = 0
𝑥=3 𝑦 =3+2=5

Example Find the critical points of 𝑓(𝑥, 𝑦) = 𝑥 2 + 𝑦 2 subject to the constraint

𝑔(𝑥, 𝑦) = 𝑥 2 + 𝑦 2 + 2𝑥 − 2𝑦 + 1 = 0

Solution First, differentiate the function and the constraint with respect to each independent
variable, x and y:

𝑓𝑥 = 2𝑥 𝑓𝑦 = 2𝑦

𝑔𝑥 = 2𝑥 + 2 𝑔𝑦 = 2𝑦 − 2

Using 𝑓𝑥 = 𝜆𝑔𝑥 𝑓𝑦 = 𝜆𝑔𝑦 , solve 𝑔(𝑥, 𝑦) = 0

8
 2 1
2𝑥 = 𝜆(2𝑥 + 2) → 𝜆 = =
2𝑥+2 𝑥+1

2 1
2𝑦 = 𝜆(2𝑦 − 2) → 𝜆= =
2𝑦−2 𝑦−1

1 1
Equating these gives =
𝑥+1 𝑦−1

𝑥+1=𝑦−1 → 𝑥 =𝑦−2

Substituting in the constraint gives:

𝑔(𝑥, 𝑦) = (𝑦 − 2)2 + 𝑦 2 + 2(𝑦 − 2) − 2𝑦 + 1 = 0

𝑦 2 − 4𝑦 + 4 + 𝑦 2 + 2𝑦 − 4 − 2𝑦 + 1 = 0

2𝑦 2 − 4𝑦 + 1 = 0

Applying the quadratic formula gives:

−(−4)±√(−4)2 −4(2)(1) 4±√8 4±2√2 √2

𝑦= = = =1±
2(2) 4 4 2

√2 √2
Hence, 𝑥 = 1 ± −2=± −1
2 2

Prepared by T-A Russell ([email protected]) July 27, 2020

Revised July 26, 2021