GEG 311 Calculus of Several Variables

Limits and Continuity; Partial Derivatives of First and Higher Orders;

Total Differential of a Function; Jacobians; Higher-Order Partial and
Total Differentials with Applications to Mechanics; Introduction to
Vector Fields – Divergence and Curl; Generalized Taylor Series: the
extremum of a function of several variables; Differentiation Under
the Integral Sign; Calculus of Variations; Line Integral with
Applications on Computation of Areas and Volumes; Functions of
Complex Variables; Cauchy-Riemann Equations; Analytical
Functions; Mapping by Elementary Functions

Dr. Patrick S. Olayiwola

Dr. John Ogbemhe
Engr. Chidi Onyedikan
PARTIAL DERIVATIVES
The partial derivative of z with respect to x measures the
instantaneous rate of change of z with respect to x while y is
held constant. Likewise, the partial derivative of z with
respect to y measures the rate of change of z with respect to
y while x is held constant.

𝜕𝑧 f x + ∆x, y − f(𝑥, 𝑦)
= lim
𝜕𝑥 ∆x→0 ∆x

𝜕𝑧 f x, y + ∆y − f(𝑥, 𝑦)
= lim
𝜕𝑦 ∆y→0 ∆y
PARTIAL DERIVATIVES
Find the partial derivative of the multivariable function such
that,
𝑧 = [3𝑦 2 ]. 𝑥 2
𝜕𝑧 2
𝑑 2
= 𝑧𝑥 = 3𝑦 . 𝑥
𝜕𝑥 𝑑𝑥
= 6𝑥𝑦 2
Similarly,
𝜕𝑧 2
𝑑 3
= 𝑧𝑦 = 3𝑥 . 𝑦
𝜕𝑦 𝑑𝑦
= 9𝑥 2 𝑦 2
PARTIAL DERIVATIVES
Exercises: Find the partial derivatives for the following;

(i) 𝑧 = 5𝑥 3 − 3𝑥 2 𝑦 2 + 7𝑦 5
(ii) z= 2𝑥 3 𝑦 2 𝑐𝑜𝑠𝑥𝑦
(iii) z=6𝑒 𝑥𝑦
SECOND-ORDER PARTIAL DERIVATIVES
Meaning that the function has been partially differentiated
w.r.t one of the independent variable twice while the other
is held constant. For, 𝑧 = 𝑓(𝑥, 𝑦)
𝜕 𝜕𝑧 𝜕2𝑧
𝑓𝑥𝑥 = (𝑓𝑥 )𝑥 = ( )= 2
𝜕𝑥 𝜕𝑥 𝜕𝑥
Similarly, 𝜕 𝜕𝑧 𝜕2𝑧
𝑓𝑦𝑦 = (𝑓𝑦 )𝑦 = ( )= 2
𝜕𝑦 𝜕𝑦 𝜕𝑦
and for cross (or mixed) partial derivatives,
𝜕 𝜕𝑧 𝜕2𝑧
𝑓𝑥𝑦 = (𝑓𝑥 )𝑦 = ( )=
𝜕𝑦 𝜕𝑥 𝜕y𝜕𝑥
and
𝜕 𝜕𝑧 𝜕2𝑧
𝑓𝑦𝑥 = (𝑓𝑦 )𝑥 = ( )=
𝜕𝑥 𝜕𝑦 𝜕x𝜕𝑦
SECOND-ORDER PARTIAL DERIVATIVES
Examples: Find (a) first, (b) second, and (c) cross partial
derivatives for;
𝑖 𝑧 = 7𝑥 3 − 9𝑥 2 𝑦 2 + 2𝑦 5
(ii) z= 3𝑥 2 𝑦 3 ,
and evaluate their values at x=4,y=1
Solution:
𝜕𝑧
(i) (a) = 21𝑥 2 − 18𝑥𝑦 2 ; 𝜕𝑧 = −18𝑦𝑥 2 + 10𝑦 4
𝜕𝑥 𝜕𝑦
(b) 𝜕2 𝑧 𝜕2 𝑧
= 42x-18 𝑦2; = -18 𝑥 2 + 40𝑦 3 ;
𝜕𝑥 2 𝜕𝑦 2
𝜕2 𝑧 𝜕
(c) = (21𝑥 2 − 18𝑥𝑦 2 )=-36xy;
𝜕𝑥𝜕𝑦 𝜕𝑦
𝜕2 𝑧 𝜕
= (−18𝑦𝑥 2 + 10𝑦 4 )=-36xy;
𝜕𝑦𝜕𝑥 𝜕𝑥
SECOND-ORDER PARTIAL DERIVATIVES
Solution continues:
(ii)(a) (𝜕𝑧 )(4,1) = [6𝑥𝑦 3 ](4,1) = 24;
𝜕𝑥

𝜕𝑧
( )(4,1) = [9𝑥 2 𝑦 2 ](4,1) = 144
𝜕𝑦

𝜕2 𝑧
(b) ( 2 )(4,1) = [6𝑦 3 ](4,1) = 6;
𝜕𝑥
𝜕2 𝑧
( 2 )(4,1) = [18𝑥 2 𝑦 2 ](4,1) = 288;
𝜕𝑦
𝜕 𝜕𝑧
(c) ( )(4,1) = [18𝑥𝑦 2 ](4,1) = 72;
𝜕𝑦 𝜕𝑥
𝜕 𝜕𝑧
( ) = [18𝑥𝑦 2 ](4,1) = 72;
𝜕𝑥 𝜕𝑦 (4,1)
SECOND-ORDER PARTIAL DERIVATIVES
Exercises: Find (a) first, (b) second, and (c) cross partial
derivatives for;
𝑖 𝑧 = (2𝑥 3 +9𝑥 2 𝑦 2 )5
(ii) z= 3𝑥 0.3 𝑦 0.5
(3𝑥+11𝑦)3
(iii) z=
2𝑥}6𝑦
TOTAL AND PARTIAL DIFFERENTIAL
Differential of a function: For a function given by
y = 4𝑥 3 + 5𝑥 2 − 7
𝑑𝑦
Then, = 12𝑥 2 + 10𝑥;
𝑑𝑥
Therefore, the differential can be written as,
dy= 12𝑥 2 + 10𝑥 dx
Total and Partial Differentials
If 𝐳 = f x, y , the total differential dz is expressed as,
dz= 𝑧𝑥 dx + 𝑧𝑦 dy
𝑥−𝑦
Example: Given z =
𝑥+1
TOTAL AND PARTIAL DIFFERENTIAL
𝑥−𝑦
Solution: z =
𝑥+1

𝑥+1 1 −(𝑥−𝑦)(1) 𝑦+1

𝑧𝑥 = =
𝑥+1 2 𝑥+1 2

𝑥+1 −1 −(𝑥−𝑦)(0) −1(𝑥+1) −1

𝑧𝑦 = = =
𝑥+1 2 𝑥+1 2 𝑥+1

The total differential is

𝑦+1 1 𝑑𝑦
dz= 2 𝑑𝑥 −
𝑥+1 𝑥+1
TOTAL AND PARTIAL DIFFERENTIAL
Total Derivatives
If 𝐳 = f x, y , the total differential dz is expressed as
Given a case where z = f(x, y) and y = g(x), that is, when
x and y are not independent, a change in x will affect z
directly through the function f and indirectly through
the function g. To measure the effect of a change in x
on z when x and y are not independent, the total
derivative must be found. The total derivative measures
𝜕𝑧
the direct effect of x on z, , plus the indirect effect of
𝜕𝑥
𝜕𝑧 𝜕𝑦
x on z through y, .
𝜕𝑦 𝜕𝑥
𝑑𝑧 𝑑𝑦
𝑖. 𝑒, = 𝑧𝑥 + 𝑧𝑦
𝑑𝑥 𝑑𝑥
TOTAL AND PARTIAL DIFFERENTIAL
Total Derivatives: Examples
1. Given z = f(x, y) =6 𝑥 3 + 7𝑦
where y=g(x)=4 𝑥 2 + 3𝑥 + 8.
𝑑𝑧 𝑑𝑦
= 𝑧𝑥 + 𝑧𝑦
𝑑𝑥 𝑑𝑥
2 𝑑𝑦
𝑧𝑥 =18 𝑥 , 𝑧𝑦 =7, and =8x+3
𝑑𝑥
𝑑𝑧
=18 𝑥 2 +7(8x+3)= 18 𝑥 2 +56x+21
𝑑𝑥

2. Given z = 8 𝑥 2 + 3𝑦 2 , x=4t, y=5t

𝑑𝑧 𝑑𝑥 𝑑𝑦 𝑑𝑥 𝑑𝑦
= 𝑧𝑥 + 𝑧𝑦 ; 𝑧𝑥 =16x, 𝑧𝑦 = 6y, and =4, =5
𝑑𝑥 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝑧 𝑑𝑧
=16x(4)+6y(5)=64x+30y; = 64 4𝑡 + 30 5𝑡 = 406𝑡
𝑑𝑡 𝑑𝑡
THE JACOBIAN
The Jacobian determinant tests for functional dependence
of both linear and nonlinear systems. A Jacobian
determinant 𝑱 is composed of all the first-order partial
derivatives of a system of equations, arranged in ordered
sequence. Given
𝑦1=𝑓1(𝑥1, 𝑥2, 𝑥3)
𝑦2=𝑓2(𝑥1, 𝑥2, 𝑥3)
𝑦3=𝑓3(𝑥1, 𝑥2, 𝑥3)
𝜕𝑦1 𝜕𝑦1 𝜕𝑦1
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
𝜕𝑦1, 𝜕𝑦2, 𝜕𝑦3 𝜕𝑦2 𝜕𝑦2 𝜕𝑦2
𝑱 = =
𝜕𝑥1, 𝜕𝑥2, 𝜕𝑥3 𝜕𝑥1 𝜕𝑥1 𝜕𝑥3
𝜕𝑦3 𝜕𝑦2 𝜕𝑦3
𝜕𝑥1 𝜕𝑥2 𝜕𝑥3
THE JACOBIAN
Example: Use Jacobian to test for the functional dependence
of the following systems;
𝑦1=5𝑥1+3 𝑥2
𝑦2=25𝑥12+30 𝑥1𝑥2+9𝑥22
𝜕𝑦1 𝜕𝑦1
Using
𝜕𝑦1, 𝜕𝑦2
𝑱 = = 𝜕𝑥1 𝜕𝑥2
𝜕𝑥1, 𝜕𝑥2 𝜕𝑦2 𝜕𝑦2
𝜕𝑥1 𝜕𝑥2
5 3
𝑱 =
50𝑥1 + 30𝑥2 30𝑥1 + 18𝑥2
𝑱 = 5(30𝑥1 + 18𝑥2)-3(50𝑥1 + 30𝑥2)=0
𝑱 =0, hence there is functional dependence between the equations
THE JACOBIAN
Exercises: Use the Jacobian to test for functional
dependence
THE HESSIAN
Given that the first-order conditions 𝑧𝑥 = 𝑧𝑦 = 0 are met, a
sufficient condition for a multivariable function z = f(x, y) to
be at an optimum is
𝑧𝑥𝑥, 𝑧𝑦𝑦 >0 for a minimum
𝑧𝑥𝑥, 𝑧𝑦𝑦<0 for a maximum
𝑧𝑥𝑥 𝑧𝑦𝑦 >(𝑧𝑥𝑦)2
A Hessian 𝐇 is a determinant composed of all the second-
order partial derivatives, with the second-order direct
partials on the diagonal and the second-order cross partials
of the principal diagonal. Thus,
𝑧𝑥𝑥 𝑧𝑥𝑦
𝐇=𝑧 𝑧𝑦𝑦
𝑦𝑥
THE HESSIAN
If the first element on the principal diagonal, the first
principal minor, 𝐻1 = 𝑧𝑥𝑥 is positive and the second
principal minor,
𝑧𝑥𝑥 𝑧𝑥𝒚 2 >0
𝐻2 = 𝑧 = 𝑧 𝑧 -(𝑧 )
𝑥𝒚 𝑧𝑦𝑦
𝑥𝑥 𝑦𝑦 𝑥𝑦

The second-order conditions for a minimum are met. When

𝐻1 >0 and 𝐻2 >0, the Hessian 𝐇 is called positive-
definite.
But 𝐻1 <0 and 𝐻2 >0, second-order conditions for
maximum are met. The Hessian 𝐇 is called negative-
definite.
THE HESSIAN
Example: 𝑧 = 3𝑥2 − 𝑥𝑦 + 2𝑦2 − 4𝑥 − 7𝑦 + 12
6 −1
𝐇= = 23 > 0
−1 4
𝐻1 =6> 0, 𝐻2 =23 > 0, 𝐇 is positive-definite, and z is
minimized at the critical values.
THE DISCRIMINANT
The determinant of a quadratic form is called a discriminant.
Given the quadratic form,
𝑧 = 𝑎𝑥2 + 𝑏𝑥𝑦 + 𝑐𝑦2
𝑎 𝑏/2
𝐃=
𝑏/2 𝑐
𝑏2
𝐷1 =a, 𝐷2 =𝑎𝑐 − . If 𝐷1 , 𝐷2 > 0, 𝐃 is positive-
4
definite, and z is positive for all nonzero values of x and y.
If 𝐷1 < 0, 𝐷2 > 0, 𝐃 is negative-definite, and z is
negative for all nonzero values of x and y. If 𝐷2 𝑖𝑠 𝑛𝑜𝑡 > 0,
z is not sign definite and z may assume both +ve and –ve
values.
THE DISCRIMINANT
Example: Test for sign definiteness for the given the
quadratic form,
𝑧 = 2𝑥2 + 5𝑥𝑦 + 8𝑦2
2 2.5
𝐃 =
2.5 8
The principal minors are
52
𝐷1 =2, 𝐷2 =2 8 − = 9.75 > 0
4
i. e. z is positive definite
HIGHER-ORDER HESSIANS
Given that y = f x1, x2, x3 , the third-order Hessian is
𝑦11 𝑦12 𝑦13
𝐇 = 𝑦21 𝑦22 𝑦23
𝑦31 𝑦32 𝑦33
𝜕2𝑦 𝜕2𝑦 𝜕2𝑦
where y11= 2, y12= ,y = etc.
𝜕𝑥1 𝜕𝑥2𝜕𝑥1 23 𝜕𝑥3𝜕𝑥2

The conditions for relative minimum or maximum depend

on the sign of the first, second, and third principal minors.,
respectively
𝑦11 𝑦12
𝐻1 =y11 >0, 𝐻2 = 𝑦 𝑦 >0, 𝐻2 = 𝐇 >0
21 22

𝑡ℎ𝑎𝑡 𝑖𝑠, 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑒 − 𝑚𝑖𝑛𝑖𝑚𝑢𝑚

HIGHER-ORDER HESSIANS
𝑦11 𝑦12
𝐻1 =y11 <0, 𝐻2 = 𝑦 𝑦22 >0, 𝐻3 = 𝐇 <0
21

𝑡ℎ𝑎𝑡 𝑖𝑠, 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑒 − 𝑚𝑎𝑥𝑖𝑚𝑢𝑚

Example: Optimize the function

𝑦 = −5𝑥12 + 10𝑥1 + 𝑥1𝑥3 − 2𝑥22 + 4𝑥2 + 2𝑥2𝑥3 − 4𝑥32

Solution: First − order conditions
𝜕𝑦
=y =-10x1+10+x3=0
𝜕𝑥1 1
𝜕𝑦
=y =-4x2+2x3+4=0
𝜕𝑥2 2
𝜕𝑦
=y =x +2x2-8x3=0
𝜕𝑥3 3 1
HIGHER-ORDER HESSIANS
That is,

−10 0 1 𝑥1 −10
0 −4 2 𝑥2 = −4
1 2 −8 𝑥3 0
Solving,
𝑥ҧ 1 1.04
𝑥ҧ 2 = 1.22
𝑥ҧ 3 0.43
Now, the second partial derivatives − the Hessian - is
−10 0 1
𝐇 = 0 −4 2
1 2 −8
𝑇ℎ𝑖𝑠 𝑖𝑠 ℎ𝑎𝑣𝑖𝑛𝑔 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠𝑎𝑠 𝑖𝑛 𝑡ℎ𝑒 𝐽𝑎𝑐𝑏𝑖𝑎𝑛.
HIGHER-ORDER HESSIANS
Finally, applying the Hessian test, i.e.

𝐻1 =−10 <0,

−10 0
𝐻2 = >0
0 −4
𝐻3 = 𝐇 = −276<0
Since the principal minors alternate correctly in sign, the Hessian
is negative definite and the function is maximized at 𝑥ҧ 1= 1.04,ഥ𝑥2=
1.22, andഥ𝑥3= 0.43.
CONSTRAINED OPTIMIZTION - BORDERED HESSIANS
To optimize a function f(x, y) subject to a constraint g(x, y), we form
a new function F x, y, 𝜇 = f x, y + 𝜇[𝑘 − 𝑔 𝑥, 𝑦 ],where the first-
order conditions are 𝐹𝑥 = 𝐹𝑦 = 𝐹𝜇 = 0.
The second-order conditions can be expressed in terms of bordered
Hessian 𝐇ഥ,

𝐹𝑥𝑥 𝐹𝑥𝑦 𝑔𝑥 0 𝑔𝑥 𝑔𝑦
ഥ = 𝐹𝑦𝑥
𝐇 𝐹𝑦𝑦 𝑔𝑦 or 𝑔𝑥 𝐹𝑥𝑥 𝐹𝑥𝑦
𝑔𝑥 𝑔𝑦 0 𝑔𝑦 𝐹𝑦𝑥 𝐹𝑦𝑦
𝐹𝑥𝑥 𝐹𝑥𝑦
which is simply the plain Hessian
𝐹𝑦𝑥 𝐹𝑦𝑦
bordered by the 1st-derivatives of the constraint with zero
on the principal diagonal.
CONSTRAINED OPTIMIZTION - BORDERED HESSIANS
For a function of 𝑛 variables 𝑓(𝑥1, 𝑥2, … , 𝑥𝑛), subject to
g 𝑥1, 𝑥2, … , 𝑥𝑛 , 𝐹 𝐹 𝑔
11 12 ⋯ 1
𝐹21 𝐹22 ⋯ 𝑔2
ഥ = −−
𝐇 −− −−
𝐹𝑛1 𝐹𝑛2 ⋯ 𝑔𝑛
𝑔1 𝑔2 ⋯ 0
where 𝐇 ഥ = 𝐻 ഥ 𝑛 , because of the 𝑛 × 𝑛 principal minor being
bordered.
ഥ2 , 𝐻
If all the principal minors are negative, i.e., if 𝐻 ഥ 3 , . . ., 𝐻ഥ 𝑛 <0,
the bordered Hessian is positive definite, and satisfies the sufficient
condition for a relative minimum. If the principal minors alternate
ഥ 2 >0, 𝐻
consistently in sign from positive to negative, i.e., if 𝐻 ഥ 3 <0,
𝐻ഥ 4 >0, etc., the bordered Hessian is negative definite, and satisfies
the sufficient condition for a relative maximum.