**Ambiguous answers**

In case of numeric solutions, we are bound by the precision of the hardware processing unit (or software implementation). We typically get exact answers only within the rational number set and only when stored in the form of a numerator and denominator - decimal (or binary) representation of fractions would need to be cut of at some point.

Moving on to real numbers, only an approximation representation is possible.

Inaccuracies may also appear in different types of data. Text may differ slightly, while still preserving the same meaning, e.g. the letter case and white spaces may be just insignificant. Representations of scientific data, such as DNA sequences, are also extremely prone to errors in encoding.

In this example we show how we can accept approximated answers for a numerical problem.

**Sample challenge**

The input of the program consists of a certain number of datasets. Each dataset consists of three lines. The first line contains a positive integer n<10, representing the degree of a polynomial. The second line contains a list of n+1 coefficients a_0, a_1, ..., a_{n} of the polynomial

f(x) = a_nx^n + a_{n-1}x^{n-1} + ... + a_1x + a_0

The last line contains two integers c and d satisfying the property f(c) f(d) < 0. Your task is to find an arbitrary root (zero) x_0 of polynomial f with precision \varepsilon = 0.001, i.e., print any point x_0 satisfying the condition f(x_0)| \le \varepsilon.

For instance, the polynomial x^5 - x - 1 has exactly one real root in the vicinity of point 1.167236. Interestingly enough, this point cannot be represented precisely through rational numbers of their roots of any order.

##### Example

**input**
3
4 3 2 1
-2 -1
2
-2 0 1
1 2
5
-1 -1 0 0 0 1
1 2
**output**
-1.650635
1.414062
1.167236