ICPC 2023 · Problem I · Waterworld

47th ICPC · Luxor, Egypt

Statement

Time limit: 3 seconds

Thousands of planets outside the Solar System have been discovered in recent years. An important factor for potential life support is the availability of liquid water. Detecting water on faraway planets is not easy. For rotating planets, a brand-new technology using relativistic quantum-polarized spectroscopy can help. It works as follows (this is a simplified description as only three people on this planet understand how it really works).

Assume the telescope shows the planet such that its rotating axis is vertical and its equator is horizontal. Only the vertical line at the center of the image (the line that covers the rotating axis) is analyzed, because it provides the highest resolution of the planet’s surface.

The analysis proceeds in steps of dd degrees. In one step, data is aggregated while the planet rotates by dd degrees, so each step gives information about a slice of dd degrees of the planet’s surface. The image is split into nn segments of equal height, which are analyzed separately. So the slice of dd degrees is partitioned into nn areas A1,...,AnA_{1}, . . . , A_{n}. For each area AiA_{i}, image analysis produces a number that gives the percentage of AiA_{i} covered by water. The areas AiA_{i} for one step are highlighted in the diagram on the right.

You may assume the planet’s surface is a sphere. This means each area A2,...,An1A_{2}, . . . , A_{n- 1} is a spherical quadrilateral: it has four vertices, two sides parallel to the equator (that is, in planes parallel to the equator’s plane) and two sides on great circles through the planet’s poles, where the great circles are dd degrees apart. At either pole, two of the four vertices collapse into the pole, so A1A_{1} and AnA_{n} are spherical triangles with only one side parallel to the equator. Due to the curvature of the surface, sides that are parallel to the equator are longer if they are closer to the equator, while sides on great circles are longer if they are closer to the poles.

The above process is repeated for the next dd degrees of rotation, and so on, a total number of mm times, until the whole surface of the planet has been covered (that is, md=360md = 360 degrees). Your task is to compute the percentage of the planet’s surface covered by water from the given data.

Input

The first line of input contains the two integers nn and mm (2n,m10002 \le n, m \le 1000). Each of the following nn lines contains mm integers ai,ja_{i,j} (0ai,j1000 \le ai,j \le 100 for 1in1 \le i \le n and 1jm1 \le j \le m). Each column of this matrix describes the measurements for a single step, that is, a rotation by dd degrees. The number ai,ja_{i,j} is the percentage of area AiA_{i} that is covered by water in the jthj^{th} step.

Output

Output the percentage of the planet’s surface covered by water. Your answer should have an absolute error of at most 10610^{- 6}.

Sample Input 1

3 7
63 61 55 54 77 87 89
73 60 38 5 16 56 91
75 43 11 3 16 20 95

Sample Output 1

51.809523810

Sample Input 2

4 3
10 10 10
10 10 10
10 10 10
10 10 10

Sample Output 2

10.000000000

No official solution in the source collection.