Keywords

1 Introduction

Let \(f\left( w \right) = \sum\nolimits_{j = 0}^{m} {c_{j} w^{j} }\) be a polynomial of degree \(m\) and let

$${\text{Max}}\left( {f,r} \right) = \mathop {\max }\limits_{\left| w \right| = r} \left| {f\left( w \right)} \right|.$$

If \(f\left( w \right)\) has no zero in \(\left| w \right| < \rho ,\) \(\rho \ge 1\), Malik [8] proved that

$${\text{Max}}\left( {f^{\prime},1} \right) \le \frac{m}{1 + \rho }{\text{Max}}\left( {f,1} \right),$$
(1)

for which the equality holds for the polynomial \(\left( {w + \rho } \right)^{m}\).

A natural question that arise is whether there exists an analogous inequality of (1) for \(f\left( w \right)\) having no zero in \(\left| w \right| < \rho\), \(\rho \le 1\). In this regard, Govil [5, 6] proved the following two results.

Theorem 1

[5] If \(f\left( w \right)\) is a polynomial of degree \(m\) having no zero in \(\left| w \right| < \rho\), \(\rho \le 1\), then

$${\text{Max}}\left( {f^{\prime},1} \right) \le \frac{m}{{1 + \rho^{m} }}{\text{Max}}\left( {f,1} \right),$$
(2)

provided that \(\left| {f^{\prime}\left( w \right)} \right|\) and \(\left| {F^{\prime}\left( w \right)} \right|\) attain their maxima at the same point on the circle \(\left| w \right| = 1\), where

$$F\left( w \right) = w^{m} \overline{{f\left( {\frac{1}{{\overline{w}}}} \right)}} .$$

Theorem 2

[6] If \(f\left( w \right)\) is a polynomial of degree \(m\) having all its zeros on \(\left| w \right| = \rho ,\rho \le 1,\)

$${\text{Max}}\left( {f^{\prime},1} \right) \le \frac{m}{{\rho^{m} + \rho^{m - 1} }}{\text{Max}}\left( {f,1} \right).$$
(3)

In literature we find refinements, generalizations and extensions of Theorems 1 and 2 by involving some coefficients of the polynomial \(f\left( w \right)\) (see [2,3,4]).

Definition 1

Let \(f\left( w \right)\) be a polynomial of degree \(m\) and let \(\beta\) be any complex number. The polar derivative of \(f\left( w \right)\) with respect to the point \(\beta\) , denoted by \(D_{\beta } f\left( w \right)\) , is defined as

$$D_{\beta } f\left( w \right) = mf\left( w \right) + \left( {\beta - w} \right)f^{\prime}\left( w \right).$$

\(D_{\beta } f\left( w \right)\) is a polynomial of degree at most \(m - 1\). It can be considered as a generalized form of the ordinary derivative of \(f\left( w \right)\) with respect to \(w\) due to the fact that

$$\mathop {\lim }\limits_{\beta \to \infty } \frac{{D_{\beta } f\left( w \right)}}{\beta } = f^{\prime}\left( w \right).$$

Polar derivative extension of (1) was proved by Aziz [1], who under the same hypothesis on \(f\left( w \right)\) proved that

$${\text{Max}}\left( {D_{\beta } f\left( w \right),1} \right) \le m\left( {\frac{\rho + \left| \beta \right|}{{1 + \rho }}} \right){\text{Max}}\left( {f,1} \right), \quad {\text{where}}\, \left| \beta \right| \ge 1.$$
(4)

2 Lemmas

We need the following results to prove our results.

The following lemma is due to Govil and Rahman [7].

Lemma 1

[7] If \(f\left( w \right)\) is a polynomial of degree \(m\) having all its zeros on \(\left| w \right| = \rho\), \(\rho \le 1\), then on \(\left| w \right| = 1\).

$$\left| {f^{\prime}\left( w \right)} \right| + \left| {F^{\prime}\left( w \right)} \right| \le m \;{\text{Max}}\left( {f,1} \right).$$
(5)

The next two lemmas are due to Barchand et al. [3].

Lemma 2

[3] If \(f\left( w \right) = \sum\nolimits_{j = 0}^{n} {c_{j} } w^{j}\) is a polynomial of degree \(m\) having all its zeros in \(\left| w \right| \le \rho ,\) \(\rho \le 1\), then

$$\frac{1}{\rho m}\left( {\frac{{c_{m - j} }}{{c_{m} }}} \right) \le 1.$$
(6)

Lemma 3

[3] If \(f\left( w \right) = \sum\nolimits_{j = 0}^{n} {c_{j} } w^{j}\) is a polynomial of degree \(m\) having all its zeros on \(\left| w \right| = \rho ,\) \(\rho \le 1\), then on \(\left| w \right| = 1\).

$${\text{Max}}\left( {f^{\prime},1} \right) \le \frac{m}{{\rho^{m} + \rho^{m - 1} }}E_{\rho } \;{\text{Max}}\left( {f,1} \right),$$
(7)

where

$$E_{\rho } = \frac{{\left( {1 + \left| t \right|} \right)\left( {\rho^{2} + \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}}{{\left( {1 - \left| t \right|} \right)\left( {1 - \rho + \rho^{2} + \rho \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}},$$
$$t = \frac{1}{\rho m}\left( {\frac{{\overline{c}_{m - 1} }}{{\overline{c}_{m} }}} \right)$$
$$s = \frac{2}{{\rho^{2} m\left( {m - 1} \right)}}\left( {\frac{{\overline{c}_{m - 2} }}{{\overline{c}_{m} }}} \right).$$

Lemma 3 is, in fact, a refinement of Theorem 2 due to Govil [6] (see Remark 3).

Remark 1

Under the hypothesis of Lemma 3, we have (see [3, Lemma 2.6]).

$$\left| t \right| = \frac{1}{\rho m}\left| {\frac{{\overline{c}_{m - 1} }}{{\overline{c}_{m} }}} \right| \le 1.$$

The following lemma was proved by Malik [8].

Lemma 4

[8] If \(f\left( w \right)\) is a polynomial of degree \(m\) having no zero in \(\left| w \right| < \rho\), \(\rho \ge 1\), then

$$\rho \left| {f^{\prime}\left( w \right)} \right| \le \left| {F^{\prime}\left( w \right)} \right|,$$
(8)

where \(F\left( w \right) = w^{m} \overline{{f\left( {\frac{1}{{\overline{w}}}} \right)}} .\)

Lemma 5

If \(f\left( w \right)\) is a polynomial of degree \(m\) having all its zeros on \(\left| w \right| = \rho ,\) \(\rho \le 1\), then on \(\left| w \right| = 1\).

$$\left| {F^{\prime}\left( w \right)} \right| \le \rho \left| {f^{\prime}\left( w \right)} \right|.$$
(9)

Proof

Since \(f\left( w \right)\) has all its zeros on \(\left| w \right| = \rho\), \(\rho \le 1\), then \(F\left( w \right)\) has all its zeros on \(\left| w \right| = \frac{1}{\rho }\), \(\frac{1}{\rho } \ge 1\). This implies that \(F\left( w \right)\) has no zeros in \(\left| w \right| < \frac{1}{\rho }\), \(\frac{1}{\rho } \ge 1.\) Thus, applying Lemma 1 to the polynomial \(F\left( w \right)\) we get, on \(\left| w \right| = 1\).

$${ }\frac{1}{\rho }\left| {F^{\prime}\left( w \right)} \right| \le \left| {f^{\prime}\left( w \right)} \right|.$$
$$\therefore \left| {F^{\prime}\left( w \right)} \right| \le \rho \left| {f^{\prime}\left( w \right)} \right|.$$

3 Main Results

In this paper, we prove polar extensions of Theorem 2. Precisely, we prove the following result.

Theorem 3

If \(f\left( w \right) = \mathop \sum \limits_{j = 0}^{n} c_{j} w^{j}\) is a polynomial of degree \(m\) having all its zeros on \(\left| w \right| = \rho ,\) \(\rho \le 1\), and \(\beta \in {\mathbb{C}}\) with \(\left| \beta \right| \ge 1\), then on \(\left| w \right| = 1\).

  1. (a)
    $$\left| {D_{\beta } f\left( w \right)} \right| \le m\left( {1 + \frac{\left| \beta \right| - 1}{{\rho^{m} + \rho^{m - 1} }}E_{\rho } } \right){\text{Max}}\left( {f,1} \right).$$
    (10)
  2. (b)
    $$\left| {D_{\beta } f\left( w \right)} \right| \le m\left( {\frac{\left| \beta \right| + \rho }{{\rho^{m} + \rho^{m - 1} }}E_{\rho } } \right){\text{Max}}\left( {f,1} \right).$$
    (11)

where

$$E_{\rho } = \frac{{\left( {1 - \left| t \right|} \right)\left( {\rho^{2} + \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}}{{\left( {1 - \left| t \right|} \right)\left( {1 - \rho + \rho^{2} + \rho \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}},$$
(12)
$$t = \frac{1}{\rho m}\left( {\frac{{\overline{c}_{m - 1} }}{{\overline{c}_{m} }}} \right)$$
(13)
$$s = \frac{2}{{\rho^{2} m\left( {m - 1} \right)}}\left( {\frac{{\overline{c}_{m - 2} }}{{\overline{c}_{m} }}} \right).$$
(14)

Proof

Let \(F\left( w \right) = w^{m} \overline{{f\left( {\frac{1}{{\overline{w}}}} \right)}}\). Then on \(\left| w \right| = 1\).

$$\left| {F^{\prime}\left( w \right)} \right| = \left| {mf\left( w \right) - wf^{\prime}\left( w \right)} \right|.$$
(15)

For \(\beta \in {\mathbb{C}}\), polar derivative of \(f\left( w \right)\) with respect to \(\beta\) is

$$D_{\beta } f\left( w \right) = mf\left( w \right) + \left( {\beta - w} \right)f^{\prime}\left( w \right).$$

Therefore,

$$\left| {D_{\beta } f\left( w \right)} \right| = \left| {mf\left( w \right) + \left( {\beta - w} \right)f^{\prime}\left( w \right)} \right|$$
$$\le \left| {mf\left( w \right) - wf^{\prime}\left( w \right)} \right| + \left| \beta \right|\left| {f^{\prime}\left( w \right)} \right|$$
(16)
$$= \left| {F^{\prime}\left( w \right)} \right| + \left| \beta \right|\left| {f^{\prime}\left( w \right)} \right|$$
(17)
$$= \left| {F^{\prime}\left( w \right)} \right| + \left| {f^{\prime}\left( w \right)} \right| + \left( {\left| \beta \right| - 1} \right)\left| {f^{\prime}\left( w \right)} \right|.$$
(18)

Using (5) of Lemma 1 in (18) we have on \(\left| w \right| = 1\)

$$\left| {D_{\beta } f\left( w \right)} \right| \le m{\text{Max}}\left( {f,1} \right) + \left( {\left| \beta \right| - 1} \right){\text{Max}}\left( {f^{\prime},1} \right).$$
(19)

Now using (7) of Lemma 3 in (19) we have for \(\left| w \right| = 1\)

$$\left| {D_{\beta } f\left( w \right)} \right| \le m\left( {1 + \frac{\left| \beta \right| - 1}{{\rho^{m} + \rho^{m - 1} }}E_{\rho } } \right){\text{Max}}\left( {f,1} \right),$$

which proves (a).

We now prove (b).

Using (9) of Lemma 5 in (17), we get for \(\left| w \right| = 1\)

$$\left| {D_{\beta } f\left( w \right)} \right|$$
$$\le \rho \left| {f^{\prime}\left( w \right)} \right| + \left| \beta \right|\left| {f^{\prime}\left( w \right)} \right|$$
(20)
$$\le \left( {\left| \beta \right| + \rho } \right){\text{Max}}\left( {f^{\prime},1} \right).$$
(21)

Using (7) of Lemma 3, we obtain for \(\left| w \right| = 1\).

$$\left| {D_{\beta } f\left( w \right)} \right| \le m\left( {\frac{\left| \beta \right| + \rho }{{\rho^{m} + \rho^{m - 1} }}E_{\rho } } \right){\text{Max}}\left( {f,1} \right),$$

which proves (b).

Remark 2

Dividing inequalities (10) and (11) by \(\left| \beta \right|\) and taking \(\left| \beta \right| \to \infty\), both reduce to

$${\text{Max}}\left( {f^{\prime},1} \right) \le \frac{m}{{\rho^{m} + \rho^{m - 1} }}E_{\rho } {\text{Max}}\left( {f,1} \right),$$

which is the conclusion of Lemma 3, where \(E_{\rho }\) is given by (12).

Remark 3

Inequalities (10) and (11) are improved extensions of inequality (3) to polar derivative. In other words, the ordinary form of (10) and (11) obtained in Remark 2 is an improvement of (3). To see this, it is sufficient to show that

$$E_{\rho } \le 1.$$

That is,

$$\frac{{\left( {1 - \left| t \right|} \right)\left( {\rho^{2} + \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}}{{\left( {1 - \left| t \right|} \right)\left( {1 - \rho + \rho^{2} + \rho \left| t \right|} \right) + \rho \left( {m - 1} \right)\left| {s - t^{2} } \right|}} \le 1.$$

i.e., \(\rho^{2} + \left| t \right| \le 1 - \rho + \rho^{2} + \rho \left| t \right|\)

which holds as \(\rho \le 1\) and \(\left| t \right| \le 1\) (by Remark 1).