By RAD, 7 years ago, translation, , 278B - New Problem

The total number of different strings of 2 letters is 262 = 676, but the total length of the input strings is no more than 600. It means that the length of answer is no more than 2. So just check all the strings of length 1 and 2.

277A - Learning Languages

Build bipartite graph with n nodes for employees and m nodes for languages. If an employee initially knows a language, than there will be an edge between corresponding nodes. Now the problem is simple: add the minimal number of edges in such a way, that all the n employees will be in the same connected component. Obviously, this number equals to the number of initially connected components, containing at least one employee, minus one. But there is one exception (pretest #4): if initially everyone knows no languages, we'll have to add n edges, because we can't add the edges between employees (remember that the graph is bipartite).

277B - Set of Points

For m = 3, n = 5 and m = 3, n = 6 there is no solution.

Let's learn how to construct the solution for n = 2m, where m ≥ 5 and is odd. Set up m points on a circle of sufficiently large radius. This will be the inner polygon. The outer polygon will be the inner polygon multiplied by 2. More precisely (1 ≤ i ≤ m):    If m is even, construct the solution for m + 1 and then delete one point from each polygon. If n < 2m, delete 2m - n points from the inner polygon.

Unfortunately, this solution doesn't work for m = 4, n = 7 and m = 4, n = 8.

Another approach is to set up m points on a convex function (for example, y = x 2 + 107), and set up the rest n - m points on a concave function (for example, y =  - x 2 - 107). Take a look at rng_58's solution — 3210150.

277C - Game

At first, notice that horizontal and vertical cuts are independent. Consider a single horizontal line. It contains m unit segments. And in any game state it's always possible to decrease the number of uncut units as the player wants. Imagine, that she starts growing a segment from a border, increasing it's length by 1 at a time. Each time the total uncut length decreases by either 0 or 1. In the end it obviously reaches 0.

The same holds for vertical lines as well. So if there are no initial cuts, the game is a nim with n - 1 piles of m stones and m - 1 piles of n stones. Could be solved with simple formula.

Initial k cuts should be just a technical difficulty. For any vertical/horizontal line, which contains at least one of the cuts, it's pile size should be decreased by the total length of all segments on this line.

How to make a first move in nim: let res is the result of state (grundy function), and a i is the size of the i-th pile. Then the result of the game without i-th pile is . We want to replace a i with some x, so that . Obviously, the only possible . The resulting solution: find a pile for which , and decrease it downto .

Suppose we have fixed set of inputs that we have to solve. Let's learn how to determine the optimal order. Obviously, Small inputs (and Large inputs with probFail = 0) won't fail in any case. It means that our penalty time is no less than submission time of last such safe'' inputs. So we will solve such inputs before all the others. Inputs with probFail = 1 are just a waste of time, we won't solve such inputs. Now we have only inputs with 0 < probFail < 1. Let i and j be two problems that we are going to solve consecutively at some moment. Let's check, if it is optimal to solve them in order i, j, or in reversed order. We can discard all the other inputs, because they don't affect on the relative order of these two.

(timeLarge i + timeLarge j)(1 - probFail j) + timeLarge i(1 - probFail i)probFail j < (timeLarge i + timeLarge j)(1 - probFail i) + timeLarge j(1 - probFail j)probFail i

- probFail j·timeLarge j - timeLarge i·probFail j·probFail i <  - probFail i·timeLarge i - timeLarge j·probFail i·probFail j

timeLarge i·probFail i(1 - probFail j) < timeLarge j·probFail j(1 - probFail i)

timeLarge i·probFail i / (1 - probFail i) < timeLarge j·probFail j / (1 - probFail j)

Now we've got a comparator for sort, which will give us the optimal order. Note, that inputs with probFail = 0, 1 will be sorted by the comparator correctly as well, so it's not a corner case.

Let's return to the initial problem. First of all, sort problems with the optimal comparator (it's clear that any other order won't be optimal by time, and the score doesn't depend on the order). Calculate the DP: z[i][j] = pair of maximal expected total score and minimal expected penalty time with this score, if we've already decided what to do with the first i problems, and we've spent j real minutes from the contest's start. There are 3 options for the i-the problem:

1. skip: update z[i + 1][j] with the same expected values

2. solve the Small input: update z[i + 1][j + timeSmall i], the expected total score increases by scoreSmall i, and the expected penalty time increases by timeSmall i (we assume that this input is solved in the very beggining of the contest)

3. solve both inputs: update z[i + 1][j + timeSmall i + timeLarge i], the expected total score increases by scoreSmall i + (1 - probFail i)scoreLarge i, and the expected penalty time becomes timeSmall i + (1 - probFail i)(j + timeLarge i) + probFail i·penaltyTime(z[i][j]), where penaltyTime(z[i][j]) is the expected penalty time from DP

The resulting answer is the best of z[n][i], (0 ≤ i ≤ t).

The expected total score could be a number around 1012 with 6 digits after decimal point. So it can't be precisely stored in double. And any (even small) error in calculating score may lead to completely wrong expected time (pretest #7). For example, you can multiply all the probabilities by 106 and store the expected score as integer number to avoid this error.

277E - Binary Tree on Plane

If there is no "binary" restriction, the solution is simple greedy. Each node of the tree (except the root) must have exactly 1 parent, and each node could be parent for any number of nodes.

Let's assign for each node i (except the root) such a node p i as a parent, so that y p i > y i and distance between i and p i is minimal possible. Renumerate all the nodes in order of non-increasing of y. Now it's clear that p i < i (2 ≤ i ≤ n). So we've just built a directed tree with all the arcs going downwards. And it has minimal possible length.

Let's recall the "binary" restriction. And realize that it doesn't really change anything: greedy transforms to min-cost-max-flow on the same distance matrix as edge's costs, but each node must have no more than 2 incoming flow units. Tutorial of Codeforces Round #170 (Div. 1) Tutorial of Codeforces Round #170 (Div. 2)  Comments (22)
 » There should be said a word about overlapping in the editorial on 278B. it s not obvious that it doesnt matter
 » although there could be several different solutions for a particular problem, I guess using disjoint set data structure(including some modifications) would be an easier solution for 277A, isn't it?
•  » » The tutorial only talks about the idea, involving connected component, and of course you can use disjoint set to deal with it, while bfs,Floyd-Warshall,etc. also works.
 » Please explain me, why in 277B there is no solution for n=6, m=3?{(0,0), (10,0), (0,10),(1,1),(3,2),(6,3)}
•  » » 7 years ago, # ^ | ← Rev. 2 →   (0,0),(1,1),(3,2),(6,3),(10,0) makes a convex polygon of 5 vertices, which is more than m=3
•  » » » Thank you! :) I read it wrong...:(
 » Could someone explain 227E — Binary Tree: how will the flow graph look like? Solution codes don't give me a hint :(
•  » » For each vertex of the tree you need to create two vertices, the first one having an incoming edge of capacity equal to 2 and cost equal to zero from the source, the second one must have an outcoming edge to the sink with unit capacity and, again, zero cost. Also, you should add edges from all vertices of the first cathegory to all vertices of the second cathegory (unit capacity and cost equal to the distance). If your maxflow equals to n — 1, then your answer is correct, else there is no suitable binary tree.
 » in 278-B if the range in bigger like n<=10^6 and length of each string be atmost 1000 or 2000 then what will be the approach?
•  » » i think may be the use of hashing can be good.
•  » » 7 years ago, # ^ | ← Rev. 2 →   There exists a very easy solution using suffix structures such as suffix tree, suffix automaton and suffix array. In the first two cases you need just to find a shortest and lexicografically smallest path, which is impossible to follow, in the last case you can build an LCP (largest common prefix) array and binsearch for the length of the answer (you can check whether the answer exists just by counting the number of different substrings of some length in linear time).UPD. I'd advise you to read yourself about these structures (you should start with suffix array).
•  » » » 7 years ago, # ^ | ← Rev. 2 →   A solution involving a suffix structure is by definition not "very easy" ;)But consider all strings of length <= 5. There are over 11 million of them, and only around under 5 million of these can be substrings of any string from a collection with total size under 1000000. So just mark off the used substrings, and then pick the smallest unused one.
•  » » » » By the phrase "very easy" I meant: you don't need to think a lot to guess such a solution.But your solution is great and "very easy" in all means:).
 » Hi,i'm solved Div 1. B too. But i cant understand how this problem check program work?..i'm very intrested of this..can sb tell me ?
 » can anyone give me some proof according to problem 277B — Set of Points about rng_58's solution ?
 » ah, maybe it is a little silly...I have a little problem to compute the expected with penalty. Denote the corresponding var with ti, pi, then we have 4 situations:  probability penalty i WA, j WA pi*pj 0 i AC, j WA (1-pi)pj ti i WA, j AC pi(1-pj) ti+tj i AC, j AC (1-pi)(1-pj) ti+(ti+tj) then sum it up, it is ti(1-pi)+(ti+tj)(1-pj), not ti(1-pi)pj+(ti+tj)(1-pi). where am I wrong?
•  » » Read the problem statement: "By the Google Code Jam rules the time penalty is the time when the last correct solution was submitted."
•  » » 7 years ago, # ^ | ← Rev. 5 →   Read problem statement more carefully: By the Google Code Jam rules the time penalty is the time when the last correct solution was submitted.so, the last line i AC, j AC (1-pi)(1-pj) ti+tj
•  » » ahhh, I thought it would be for each problem solved. many thx to all.
 » https://ideone.com/e.js/xhi3fi why my code is getting runtime error for problem 277A
 » Neat and Easy disjoint set union using solution 277A — Learning Languageshttps://github.com/joy-mollick/Problem-Solving-Solutions/blob/master/Codeforces-277A%20-%20Learning%20Languages.cpp
 » Why in 277D GOOGLE CODE JAM, we are not doing time penalty calculation in an integer as doing the calculation in double may lead to precision error. Is it because the time is only 1560?