Positive real functions

April 25, 2020

Motivation
After learning about Hurwitz polynomials, which in essence captures the constraints on the polynomials representing
poles or zeros, let us move onto understanding constraints on the ratio of polynomials. The first and foremost thing for
driving point function is that both numerator and denominator polynomial should be Hurwitz for the network function
to be realizable. The other constraints are captured by mathematical functions known as positive real functions which is
the topic of this notes

1 Positive Real functions

1. All physically realizable passive driving point immitances are positive real functions.
P (s)
2. A function F (s) = Q(s) is positive real if the following conditions are satisfied

• F(s) is real for real s; F (σ) is real.

• The real part of F(S) is greater than or equal to zero when the real of part of s is greater or
equal to zero
• Re[F (s)] ≥ 0 for Re(s) ≥ 0
3. If F(S) is PR, then a point σo on the positive real axis of the s-plane would correspond to or map onto a point
F (σo ), which must be on the postive real axis of F(s) plane.
4. Point si on the right half of s-plane would map onto a point F (si ) in the right half of F(s) plane.

5. Explaining in s a simple way, the right half of s-plane maps onto right of F(s) plane. The real axis of s-plane maps
onto the real axis of F(s) plane.
6. All the immittance functions of basic passive elements are PR.

Element Impedance function Admittance function

Resistor F(s) = R F(s) = G
Inductor F(s) = k1 s F(s) = k2 /s
Capacitor F(s) = k3 /s F(s) = k4 s

7. It follows that all driving point immittances of any network made of passive elements must be PR. The proof for
the same is as follows
8. Let the input impedance of the network be Z(s) and the current be I(s).

9. The first condition for PR that function Z(s) is real for real s. Z(σ) is real. This is a pure resistive network.
10. The second condition for PR which is Re[Z(s)] ≥ 0 for Re(s) ≥ 0 needs to be proved.
11. For a network having a sinusoidal excitation s = jω, the power absorbed by the network should be greater than
zero and is given by
1
Re[Z(jω)|I|2 ] ≥ 0 (1)
2

12. This implies that Re[Z(jω)] ≥ 0. This indicates that for every point of jω axis, the real part of the impedance
function Z(jω) should be non-negative.
13. Consider a network N whose driving point function is Z(s). This network N is made up of ideal passive elements
viz., resistor, inductors and capacitors.

1
14. Now let us create one more network N’ wherein each ideal energy storage is replaced by a lossy energy storage
element.
15. Ideal inductor is replaced by inductor in series with a resistor. If an inductor of impedance sLi henry in network
gets replaced by an indcutor whose impedance is sLi + i = L(s + R Li ) = L(s + α). Let α be same all inductors.
i

Each inductor in network Li is augmented by a resistor Ri such that the ratio R Li = constant = α
i

16. Similarly every Ideal capacitor Cj is replaced by capacitor in parallel with a resistor whose conductance is Gj .
Let us choose Gj in such a way that Gj = αCj . The admittance of the lossy capacitor in the new network is
Cj (s + α).

17. Let the resistors in the new network N’ be same as the network N.
18. The input impedance of new network N’ is obtained by replacing s by (s + α)
19. Z1 (s) = Z(s + α).
20. Since Re[Z(jω)] ≥ 0 and the new network N’ has only extra resistors added, it follows that Re[Z(jω + α)] ≥ 0

21. Remember α is just an arbitrary constant which is non negative. α cannot be negative because negative passive
resistors do not exist.
22. we can replace α with any other symbol. Let us replace it with σ where σ is positive.
23. This implies that Re[Z(jω + σ)] ≥ 0 for all σ ≥ 0

24. In other words Re[Z(s)] ≥ 0 for Re(s) ≥ 0. Hence proved.

2 Useful Properties of PR
1. If F(s) is PR, its inverse F 1(s) is also PR. If a driving point impedance function is PR, then the driving point
admittance function is also PR.

2. The sum of two PR functions is a PR. The difference of two PR functions need not be PR
3. The Poles and Zeros of driving point PR function cannot have positive real parts. In other words, if PR is
representing ratio of two polynomials, the both the polynomials are Hurwitz.
4. Poles and Zeros must occur in complex conjugate pairs. Otherwise it will result in imaginary coefficients.

5. Only simple poles (zeros) can with real positive residues can exist on jω axis.
• If there are multiple poles or zeros on the imaginary axis then the system is unstable.
• Let us understand what positive real residues mean. Let us consider that a function F(s) has all poles on RHS
of s-plane and a pair of conjugate poles at ±jωo .
• Let us write the function F (s) as F (s) = k
s2 +ωo2 F1 (s) where k is real.
• The partial fraction of ψ(s) = k
s2 +ωo2 can be written as

k1 k2
ψ(s) = + (2)
s − jω s − jω

• First argument we put forward is that, k1 and k2 if complex should be conjugate. Even if they complex
conjugate they should of the form k1 = ka − jkb and k2 = ka + jkb . Otherwise it results in imaginary
coefficients. Note that ka and kb are real numbers.
• With these assumptions, the function ψ(s) will become

2ka s − 2kb ωo
ψ(s) = (3)
s2 + ωo2

• For F(s) to be positive real ψ(s) should be positive real.

• However for ψ(s), there is possibility of it being negative because there is a negative term in the numerator.
The only way we can ensure that ψ(s) is PR is by making kb = 0 and ka should be positive which means
that k1 = k2 = 2ka .
2ka s
ψ(s) = 2 (4)
s + ωo2

2
 • ka is the residue that needs to be evaluated as per the standard procedure of finding residues for partial
fractions.
F (s)(s2 + ωo2 ) 

ka = (5)
2s 2
s = −ω 2

6. The highest power of the numerator and denominator may differ at most by unity. This condition prohibits multiple
poles and zeros at s = ∞.

7. The lowest power of the denominator and numerator polynomials may differ at most by unity. This condition
prohibits the possibility of multiple zeros or poles at s = 0

3 Evaluation procedure for PRs

1. Let us reiterate the definition of PR
P (s)
2. A function F (s) = Q(s) is positive real if the following conditions are satisfied

• F(s) is real for real s; F (σ) is real.

• Re[F (s)] ≥ 0 for Re(s) ≥ 0

3. The above conditions mandates the following. These become the necessary and sufficient conditions for a rational
function with real coefficients to be PR
(a) F(s) must have no poles in the right half of s-plane
(b) F(s) must have only simple poles on the jω axis with real and positive residues
(c) Re[F (jω)] ≥ 0 ∀ ω
4. Condition ’a’ requires that the denominator polynomial is Hurwitz
5. Condition ’b” requires figuring out whether the residues positive as per eqn.5
6. Condition ’c’ requires calculation of Re[F (jω)]. The next few points illustrate on the procedure to do the same.

(a) Since F(s) is rational. F(s) can be written as

P (s)
F (s) = (6)
Q(s)

(b) The numerator and denominator polynomials can be factorized into odd functions as even functions

P (s) = M1 (s) + N1 (s); Q(s) = M2 (s) + N2 (s) (7)

(c) M represents even function and N represents odd function. For the next few steps, let us remember that all
the functions that we are discussing are functions of S. We will use M for M(s) and N for N(s).
(d) The function F(s) can be written as
M 1 + N1
F (s) = (8)
M 2 + N2

(e) A simple mathematical manipulation of multiplying numerator and denominator with M2 − N2 leads to

(M1 + N1 ) (M2 − N2 )
F (s) = (9)
(M2 + N2 ) (M2 − N2 )

(M1 M2 − N1 N2 ) (M2 N1 − M1 N2 )
F (s) = 2 2 + (10)
(M2 − N2 ) (M22 − N22 )

(f) Even part of F(s)

(M1 M2 − N1 N2 )
Ev[F (s)] = (11)
(M22 − N22 )

(g) Odd part of F(s)

(M2 N1 − M1 N2 )
O[F (s)] = (12)
(M22 − N22 )

3
 (h) Now let us substitute s = jω.
(i) M22 (jω) = M22 (ω) = (M2 (ω))2 Observe the independent variable on RHS. There is a subtle difference
(j) N22 (jω) = − N22 (ω) = − (N2 (ω))2 Observe the independent variable on RHS. There is a subtle difference
(k) It follows that M22 (jω) − N22 (jω) = (M2 (ω))2 + (N2 (ω))2 ≥ 0 ∀ω
(l) The denominator is in eqn.10 is positive ∀ω
(m) Another logical argument to prove that the denominator is positive as follows
• If D(s) = M2 +N2 , M2 −N2 is nothing but D(−s). At s = jω, M22 −N22 = D(jω)∗D(−jω) = |D(jω)|2 .
• All modulus terms are positive. Hence the denominator is always positive
 
 
(n) It can be easily seen that Re[F (jω)] = Ev[F (s)]  , Im[F (jω)] = O[F (s)]
s=jω s=jω


(o) It can be concluded that in order to find out whether Re[F (jω)] ≥ 0,we need to do evaluate the Ev[F (s)] .
s=jω
Since the denominator of Ev[F (s)] is positive, we need to evaluate the numerator of Ev[F (s)]. Let it be A(s)
(p) If A(s)|s=jω is greater than equal to zero, then Re[F (jω)] ≥ 0 ∀ω.
(q) Since A(s) is an even polynomial, it can be expressed as a function of s2 as well.
(r) Evaluating A(s)|s=jω can be done by evaluating A(s2 )|s2 =−ω2 = A(ω 2 ). If A(ω 2 ) ≥ 0 ∀ω then Re[F (jω)] ≥ 0
∀ω

4 Step by Step Evaluation for determining whether a given function F(s)

is PR or not
1. Rule 1: F(s) must have no poles in the right half of s-plane
2. Check 1 : Check whether the denominator of F(s) is Hurwitz. If yes, verify Rule:2 and 3. If no, the function F(s)
is not PR
3. Rule 2: F(s) must have only simple poles on the jω axis with real and positive residues

4. Check 2: Check if the roots on jω axis are simple. If no, the function F(s) is not PR. If yes, Find out the residue
using eqn.5. If the residue is negative, the function is not PR. If the residue is positive go for Rule:3.
5. Rule 3: Re[F (jω)] ≥ 0 ∀ ω
6. Check 3:

• Separate individually, the numerator and denominator polynomials in odd and even polynomials M1 , M2 , N1 , N2
• Multiply even parts and odd parts. Subtract the products from each other M1 M2 − N1 N2 .
• The resultant polynomial is a function of s2 . Evaluate the same at s2 = − ω 2 and check whether it is non
negative for all ω. If no, the polynomial is not PR and if it is yes, the function is PR

5 Examples
s + a
1. Problem 1: What are the conditions for which, a generic driving point function F (s) = s2 + bs + c is PR.
2. Answer: Let us follow step by step procedure for identifying the conditions

(a) Check 1 : Check whether the denominator of F(s) is Hurwitz. As per given problem F (s) is driving point
function. Hence both numerator and denominator should be Hurwitz.
(b) Conclusion 1:a,b,c must be non-negative which implies a, b, c ≥ 0
(c) Check 2 : Check whether there are poles on the imaginary axis. The denominator of F(s) is s2 + bs + c. In
order to have poles on the jω axis b = 0. Is it permissible? Yes.
(d) Check 3 : Compute the function A(s2 ) = M1 M2 − N1 N2 M1 M2 = a(s2 + c) N1 N2 = bs2 ;
(e) A(s2 ) = a(s2 + c) − bs2
(f) A(s2 )|s2 = −ω 2 = (b − a)ω 2 + c
(g) For A(ω s ) to be positive b ≥ a. If b = 0 then a = 0
(h) So the conditions are a, b, c ≥ 0, b ≥ a. If b = 0 then a = 0.
s2 + a1 s + ao
3. Problem 2: What are the conditions for which, a generic driving point function F (s) = s2 + b1 s + bo is PR.

4
4. Answer: Let us follow step by step procedure for identifying the conditions
(a) Check 1 : Check whether the denominator of F(s) is Hurwitz. As per given problem F (s) is driving point
function. Hence both numerator and denominator should be Hurwitz.
(b) Conclusion 1:a1 , ao , b1 , bo must be non-negative which implies a1 , ao , b1 , bo ≥ 0
(c) Check 2 : Check whether there are poles on the imaginary axis. The denominator of F(s) is s2 + b1 s + bo .
In order to have poles on the jω axis b1 = 0. Is it permissible? Yes.
(d) Check 3 : Compute the function A(s2 ) = M1 M2 − N1 N2 M1 M2 = (s2 + a)(s2 + b) N1 N2 = a1 b1 s2 ;
(e) A(s2 ) = (s2 + a)(s2 + b) − a1 b1 s2
(f) Let ω 2 =
(g) A(s2 )|s2 = −ω 2 = (s2 )|s2 = −x = (ao − x)(bo − x) + a1 b1
2 2
√
(h) Simplifying A(s )|s2 = −x = x + ao bo + x(a1 b1 − ao − bo ). Adding and subtracting 2 ao bo x
√ √
(i) Simplifying A(s2 )|s2 = −x = (x − ao bo )2 + x(a1 b1 − ao − bo + 2 ao bo ).
√ √ √
(j) Simplifying A(s2 )|s2 = −x = (x − ao bo )2 + x(a1 b1 − ( ao − bo )2 ). The first term in this equation is
always positive. We should avoid situations where the difference may yield a negative number.
√ √
(k) For A(ω s ) to be positive, a1 b1 ≥ ( ao − bo ). If b1 = 0, it necessitates that ao = bo .
√ √
(l) So the conditions area1 , ao , b1 , bo ≥ 0 , a1 b1 ≥ ( ao − bo ). If b1 = 0, it necessitates that ao = bo .