Assume that the number of variables N and the number of clauses K are equal. Find an algorithm that returns the number of different ways to satisfy the clauses.
I read that SAT is related to Independent Sets.
A function with N variables has a truth-table with 2^N rows. Each row corresponds to one minterm which can be either a solution or not.
A clause with N variables excludes exactly one of the minterm as part of the solutions. That is the minterm which consists of all inverted variables of the clause.
Provided, the K clauses are all different,
the number of solutions is 2^N - K
Example:
The K=3 clauses with N=3 variables:
A or B or C
!A or B or C
A or B or !C
The truth-table for three inputs:
A B C output
0 0 0 0 // excluded by A or B or C
0 0 1 0 // excluded by A or B or !C
0 1 0 1
0 1 1 1
1 0 0 0 // excluded by !A or B or C
1 0 1 1
1 1 0 1
1 1 1 1
Five of the possible eight terms remain true. Thus, the example has 2^3 - 3 = 5 solutions.
Related
I am so newbie and thank you so much in advance for advice
I want to make co-occurrence matrix, and followed link below
How to use R to create a word co-occurrence matrix
but I cannot understand why value of A-A is 10 in the matirx below
It should be 4 isn't it? because there are four A
dat <- read.table(text='film tag1 tag2 tag3
1 A A A
2 A C F
3 B D C ', header=T)
crossprod(as.matrix(mtabulate(as.data.frame(t(dat[, -1])))))
( ) A C F B D
A 10 1 1 0 0
C 1 2 1 1 1
F 1 1 1 0 0
B 0 1 0 1 1
D 0 1 0 1 1
The solution you use presumes each tag appears only once per film, which jives with the definition of a co-occurrence matrix as far as I can tell. Therefore, each A on the first line gets counted as co-occurring with itself and with the other two As, resulting in a total of ten co-occurences when factoring in the A on the second line.
i seen a post on the site about it and i didn't understand the answer, can i get explanation please:
question:
Write code to determine if a number is divisible by 3. The input to the function is a single bit, 0 or 1, and the output should be 1 if the number received so far is the binary representation of a number divisible by 3, otherwise zero.
Examples:
input "0": (0) output 1
inputs "1,0,0": (4) output 0
inputs "1,1,0,0": (6) output 1
This is based on an interview question. I ask for a drawing of logic gates but since this is stackoverflow I'll accept any coding language. Bonus points for a hardware implementation (verilog etc).
Part a (easy): First input is the MSB.
Part b (a little harder): First input is the LSB.
Part c (difficult): Which one is faster and smaller, (a) or (b)? (Not theoretically in the Big-O sense, but practically faster/smaller.) Now take the slower/bigger one and make it as fast/small as the faster/smaller one.
answer:
State table for LSB:
S I S' O
0 0 0 1
0 1 1 0
1 0 2 0
1 1 0 1
2 0 1 0
2 1 2 0
Explanation: 0 is divisible by three. 0 << 1 + 0 = 0. Repeat using S = (S << 1 + I) % 3 and O = 1 if S == 0.
State table for MSB:
S I S' O
0 0 0 1
0 1 2 0
1 0 1 0
1 1 0 1
2 0 2 0
2 1 1 0
Explanation: 0 is divisible by three. 0 >> 1 + 0 = 0. Repeat using S = (S >> 1 + I) % 3 and O = 1 if S == 0.
S' is different from above, but O works the same, since S' is 0 for the same cases (00 and 11). Since O is the same in both cases, O_LSB = O_MSB, so to make MSB as short as LSB, or vice-versa, just use the shortest of both.
thanks for the answers in advanced.
Well, I suppose the question isn't entirely off-topic, since you asked about logic design, but you'll have to do the coding yourself.
You have 3 states in the S column. These track the value of the current full input mod 3. So, S0 means the current input mod 3 is 0, and so is divisible by 0 (remember also that 0 is divisible by 3). S1 means the remainder is 1, S2 means that the remainder is 2.
The I column gives the current input (0 or 1), and S' gives the next state (in other words, the new number mod 3).
For 'LSB', the new number is the old number << 1, plus either 0 or 1. Write out the table. For starters, if the old modulo was 0, then the new modulo will be 0 if the input bit was 0, and will be 1 if the new input was 1. This gives you the first 2 rows in the first table. Filling in the rest is left as an exercise for you.
Note that the O column is just 1 if the next state is 0, as expected.
I have an algorithmic problem to solve, it looks like this:
We have sequence of length n with elements ranging from 0 to (k-1).
Example for n = 10, k = 3:
2 1 0 0 2 1 0 0 1 2
I have to sort this string, so it would look like this:
0 1 2 0 1 2 0 1 2 0 2 (sorted subsets from 0 to k-1, surplus elements may not be sorted).
But it have to be sorted by moving k adjacent elements to the end and joining the rest, example:
0 1 2 [2 0 2] 0 1
0 1 2 0 1 [2 0 2]
In solution, I can have assumption, that there will be a certain number of redundant elements.
I have already invented naive algorithm to solve this:
http://pastebin.com/Q5CR99E7
...but I have to come up with 3 algorithms and I think that there may be more sophisticated solution.
I would be grateful for your ideas.
I need to assign random papers to students of a class, but I have the constraints that:
Each student should have two papers assigned.
Each paper should be assigned to (approximately) the same number of students.
Is there an elegant way to generate a matrix that has this property? i.e. it is shuffled but the row and column sums are constant? As an illustration:
Student A 1 0 0 1 1 0 | 3
Student B 1 0 1 0 0 1 | 3
Student C 0 1 1 0 1 0 | 3
Student D 0 1 0 1 0 1 | 3
----------------
2 2 2 2 2 2
I thought of first building an "initial matrix" with the right row/column sum, then randomly permuting first the rows, then the colums, but how do I generate this initial matrix? The problem here is that I'd be choosing between (e.g.) the following alternatives, and the fact that there are two students with the same pair of papers assigned (in the left setup) won't change through row/column shuffling:
INITIAL (MA): OR (MB):
A 1 1 1 0 0 0 || 1 1 1 0 0 0
B 1 1 1 0 0 0 || 0 1 1 1 0 0
C 0 0 0 1 1 1 || 0 0 0 1 1 1
D 0 0 0 1 1 1 || 1 0 0 0 1 1
I know I could come up with something quick/dirty and just tweak where necessary but it seemed like a fun exercise.
If you want to make permutations, what about:
Chose randomly a student, say student 1
For this student, chose a random paper he has, say paper A
Chose randomly another student
For this student, chose a random paper he has, say paper B (different from A)
Give paper B to student 1 and paper A to student 2.
That way, you preserve both the number of different papers and the number of papers per student. Indeed, both students give one paper and receive one back. Moreover, no paper is created nor deleted.
In term of table, it means finding two pairs of indices(i1,i2) and (j1,j2) such that A(i1,j1) = 1, A(i2,j2)=1, A(i1,j2)=0 and A(i2,j1)=0 and changing the 0s for 1s and the 1s for 0s => The sums of the rows and columns do not change.
Remark 1: If you do not want to proceed by permutations, you can simply put in a vector all the paper (put 2 times paper A, 2 times paper B,...). Then, random shuffle the vector and attribute the k first to the first student, the k next ones to student 2, ... However, you can end with a student having several times the same paper. In this case, make some permutations starting with the surnumerary papers.
You can generate the initial matrix as follows (pseudo-Python syntax):
column_sum = [0] * n_students
for i in range(n_students):
if column_sum[i] < max_allowed:
for j in range(i + 1, n_students):
if column_sum[j] < max_allowed:
generate_row_with_ones_at(i, j)
column_sum[i] += 1
column_sum[j] += 1
if n_rows == n_wanted:
return
This is a straightforward iteration over all n choose 2 distinct rows, but with the constraint on column sums enforced as early as possible.
I'm looking for some pointers here as I don't quite know where to start researching this one.
I have a 2D matrix with 0 or 1 in each cell, such as:
1 2 3 4
A 0 1 1 0
B 1 1 1 0
C 0 1 0 0
D 1 1 0 0
And I'd like to sort it so it is as "upper triangular" as possible, like so:
4 3 1 2
B 0 1 1 1
A 0 1 0 1
D 0 0 1 1
C 0 0 0 1
The rows and columns must remain intact, i.e. elements can't be moved individually and can only be swapped "whole".
I understand that there'll probably be pathological cases where a matrix has multiple possible sorted results (i.e. same shape, but differ in the identity of the "original" rows/columns.)
So, can anyone suggest where I might find some starting points for this? An existing library/algorithm would be great, but I'll settle for knowing the name of the problem I'm trying to solve!
I doubt it's a linear algebra problem as such, and maybe there's some kind of image processing technique that's applicable.
Any other ideas aside, my initial guess is just to write a simple insertion sort on the rows, then the columns and iterate that until it stabilises (and hope that detecting the pathological cases isn't too hard.)
More details: Some more information on what I'm trying to do may help clarify. Each row represents a competitor, each column represents a challenge. Each 1 or 0 represents "success" for the competitor on a particular challenge.
By sorting the matrix so all 1s are in the top-right, I hope to then provide a ranking of the intrinsic difficulty of each challenge and a ranking of the competitors (which will take into account the difficulty of the challenges they succeeded at, not just the number of successes.)
Note on accepted answer: I've accepted Simulated Annealing as "the answer" with the caveat that this question doesn't have a right answer. It seems like a good approach, though I haven't actually managed to come up with a scoring function that works for my problem.
An Algorithm based upon simulated annealing can handle this sort of thing without too much trouble. Not great if you have small matrices which most likely hae a fixed solution, but great if your matrices get to be larger and the problem becomes more difficult.
(However, it also fails your desire that insertions can be done incrementally.)
Preliminaries
Devise a performance function that "scores" a matrix - matrices that are closer to your triangleness should get a better score than those that are less triangle-y.
Devise a set of operations that are allowed on the matrix. Your description was a little ambiguous, but if you can swap rows then one op would be SwapRows(a, b). Another could be SwapCols(a, b).
The Annealing loop
I won't give a full exposition here, but the idea is simple. You perform random transformations on the matrix using your operations. You measure how much "better" the matrix is after the operation (using the performance function before and after the operation). Then you decide whether to commit that transformation. You repeat this process a lot.
Deciding whether to commit the transform is the fun part: you need to decide whether to perform that operation or not. Toward the end of the annealing process, you only accept transformations that improved the score of the matrix. But earlier on, in a more chaotic time, you allow transformations that don't improve the score. In the beginning, the algorithm is "hot" and anything goes. Eventually, the algorithm cools and only good transforms are allowed. If you linearly cool the algorithm, then the choice of whether to accept a transformation is:
public bool ShouldAccept(double cost, double temperature, Random random) {
return Math.Exp(-cost / temperature) > random.NextDouble();
}
You should read the excellent information contained in Numerical Recipes for more information on this algorithm.
Long story short, you should learn some of these general purpose algorithms. Doing so will allow you to solve large classes of problems that are hard to solve analytically.
Scoring algorithm
This is probably the trickiest part. You will want to devise a scorer that guides the annealing process toward your goal. The scorer should be a continuous function that results in larger numbers as the matrix approaches the ideal solution.
How do you measure the "ideal solution" - triangleness? Here is a naive and easy scorer: For every point, you know whether it should be 1 or 0. Add +1 to the score if the matrix is right, -1 if it's wrong. Here's some code so I can be explicit (not tested! please review!)
int Score(Matrix m) {
var score = 0;
for (var r = 0; r < m.NumRows; r++) {
for (var c = 0; c < m.NumCols; c++) {
var val = m.At(r, c);
var shouldBe = (c >= r) ? 1 : 0;
if (val == shouldBe) {
score++;
}
else {
score--;
}
}
}
return score;
}
With this scoring algorithm, a random field of 1s and 0s will give a score of 0. An "opposite" triangle will give the most negative score, and the correct solution will give the most positive score. Diffing two scores will give you the cost.
If this scorer doesn't work for you, then you will need to "tune" it until it produces the matrices you want.
This algorithm is based on the premise that tuning this scorer is much simpler than devising the optimal algorithm for sorting the matrix.
I came up with the below algorithm, and it seems to work correctly.
Phase 1: move rows with most 1s up and columns with most 1s right.
First the rows. Sort the rows by counting their 1s. We don't care
if 2 rows have the same number of 1s.
Now the columns. Sort the cols by
counting their 1s. We don't care
if 2 cols have the same number of
1s.
Phase 2: repeat phase 1 but with extra criterions, so that we satisfy the triangular matrix morph.
Criterion for rows: if 2 rows have the same number of 1s, we move up the row that begin with fewer 0s.
Criterion for cols: if 2 cols have the same number of 1s, we move right the col that has fewer 0s at the bottom.
Example:
Phase 1
1 2 3 4 1 2 3 4 4 1 3 2
A 0 1 1 0 B 1 1 1 0 B 0 1 1 1
B 1 1 1 0 - sort rows-> A 0 1 1 0 - sort cols-> A 0 0 1 1
C 0 1 0 0 D 1 1 0 0 D 0 1 0 1
D 1 1 0 0 C 0 1 0 0 C 0 0 0 1
Phase 2
4 1 3 2 4 1 3 2
B 0 1 1 1 B 0 1 1 1
A 0 0 1 1 - sort rows-> D 0 1 0 1 - sort cols-> "completed"
D 0 1 0 1 A 0 0 1 1
C 0 0 0 1 C 0 0 0 1
Edit: it turns out that my algorithm doesn't give proper triangular matrices always.
For example:
Phase 1
1 2 3 4 1 2 3 4
A 1 0 0 0 B 0 1 1 1
B 0 1 1 1 - sort rows-> C 0 0 1 1 - sort cols-> "completed"
C 0 0 1 1 A 1 0 0 0
D 0 0 0 1 D 0 0 0 1
Phase 2
1 2 3 4 1 2 3 4 2 1 3 4
B 0 1 1 1 B 0 1 1 1 B 1 0 1 1
C 0 0 1 1 - sort rows-> C 0 0 1 1 - sort cols-> C 0 0 1 1
A 1 0 0 0 A 1 0 0 0 A 0 1 0 0
D 0 0 0 1 D 0 0 0 1 D 0 0 0 1
(no change)
(*) Perhaps a phase 3 will increase the good results. In that phase we place the rows that start with fewer 0s in the top.
Look for a 1987 paper by Anna Lubiw on "Doubly Lexical Orderings of Matrices".
There is a citation below. The ordering is not identical to what you are looking for, but is pretty close. If nothing else, you should be able to get a pretty good idea from there.
http://dl.acm.org/citation.cfm?id=33385
Here's a starting point:
Convert each row from binary bits into a number
Sort the numbers in descending order.
Then convert each row back to binary.
Basic algorithm:
Determine the row sums and store
values. Determine the column sums
and store values.
Sort the row sums in ascending order. Sort the column
sums in ascending order.
Hopefully, you should have a matrix with as close to an upper-right triangular region as possible.
Treat rows as binary numbers, with the leftmost column as the most significant bit, and sort them in descending order, top to bottom
Treat the columns as binary numbers with the bottommost row as the most significant bit and sort them in ascending order, left to right.
Repeat until you reach a fixed point. Proof that the algorithm terminates left as an excercise for the reader.