Date: 07/26/05     Keywords: php, mysql, sql, java

Сабж.Для создание многопользовательской игры на flash.
Труд будем оплачивать. Можно удалённо. Знание Java и Php. Также очень желательно MySQL.

Source: http://www.livejournal.com/community/algorithms/61549.html

Date: 07/26/05     Keywords: web

I'm trying to help a friend in the electrical engineering field out. Their application needs to compute a minimum spanning tree with arbitrary limits on the degree of particular nodes. Plus they insist on certain edges being a part of the solution. I've looked around on the web a bit, sounds like it's an NP problem. If I'm wrong, do let me know! So... any good thoughts on heuristics etc?

My thought was a backtracking algorithm based on Kruskal's (the greedy one), with pruning. Basically you go from shortest to longest. At any edge that could be part of the solution (not create a loop or overload a node), you decide whether to use it. If you say yes all the way, it's effectively Kruskal's, but with the limitations I mentioned. If you remove the degree limitation it *is* Kruskal's.

Anyway, the pruning would be based on estimating the minimum cost of a total solution given the choices you've already made. One way of doing this would be to multiply the number of remaining edges you need by the cost of the current edge, and add on the existing solution vector. If it's greater than your best so far, backtrack.

Oh yeah, any idea how you'd get a tight minimum and maximum limit on the solution cost? A minimum would be Kruskal's without restriction. A maximum... nothing obvious comes to mind. Obviously ANY solution is a maximum, but a tight limit would make for better pruning.

Source: http://www.livejournal.com/community/algorithms/61283.html

Date: 07/20/05     Keywords: no keywords

I don't know much about machine learning, so here's what may be a basic question:

I have a program that generates a set of 8 numerical scores, and then needs to separate them into a low and a high group. Now what's low in one set may not be in another, so I'm using k-means right now (k=2) to separate them. However, this does not always separate the way I want it to, especially if the standard dev of the scores is not very high. So I'm wondering if some sort of Bayesian learning would help. Assuming I have a training dataset of about 60 score sets, would this be enough to construct a decision procedure for eight scores? And any other ideas on how to improve the classification? Thanks!

Source: http://www.livejournal.com/community/algorithms/60989.html

Date: 07/12/05     Keywords: software, technology, ebay

The benefits of carpooling are obvious, and I would like to highlight that we have the technology to enhance participation.

I see two fundamental challenges with carpooling:
1) Trust: How do I know this guy will actually pick me up, be on time, etc?
2) Knowledge of other local commuters

Implementing #1:
Before E-bay, it would be considered inconcievable that consumers could rate each other and provide services directly without mediation.
Yes, e-bay adds some mediation but only minimally to resolve financial disputes. (Ebay essentially provides the sidewalk in the global yard sale.) Allowing fellow commuters to rate each others' availability and reliability well suites the trust issue.
The Open Source community could easily build such a rating system.

Implementing #2:
Users provide frequent routes, and custom software matches possible local participants.
I know this software exists in algorithm communities, since students frequently write network traversal simulation software.

I theorize that this can ONLY be done Open Source since as privitazation means paying somewhere.
People will not feel comfortable providing home addresses etc to a private body since that equates to selling your identity.
(Regardless of what any Privacy/End User agreement says).

Thoughts?

Source: http://www.livejournal.com/community/algorithms/60498.html

Date: 07/12/05     Keywords: software

I'm looking for an algorithm to fetch dominant (no redunant) colors from 24-bit color photos.

The only thing i've found is such algorithm for monochrome pictures. Its author said that the algorithm for photos would be much harder.

Free or commercial Unix console software that implements such algorithms are also pleased.

Source: http://www.livejournal.com/community/algorithms/60210.html

Date: 07/08/05     Keywords: no keywords

I am having difficulty on how to do analysis to show the relationship between 2 algorithms, basically I am not sure what type of statistic, data to produce.  The problem is on the "movement" of agent-like system (Ant System, a bit diff from ACOhere ),  one algorithm is 'dependent'  movement and another is 'independent' movement. 

Dependent movement is at stage s,  agent(i) decision to move to a new place is based on the information obtained from agent(0:i-1)  .  (i.e when ants move, they lay pheremone and such).  Independent movement is agent(i) decision to move to a new place is independent from other agents, the decision is based on information on previous stage s-1. 

Imagine 2 ants at different places X , Y , try to decide if they should go to Z,  dependent movement is ant 1 makes decision first, it might move to Z, and thus ant 2 will be more likely to also choose Z since Z now has a stronger pheromone attaction.  Whereas independent movement is both ants will make the decision without regarding others's decision, so if ant 1 moves to Z that doesn't encourage ant 2 moves to Z since ant 2 is not aware of ant 1's decision. 

So clearly 2 algorithms are very different.  In my experiments, both of these algorithms give Equivalent results (i.e the solutions (on average) are equivalent). But I need to do a more detailed and formal analysis on Why 2 different behaviors yet producing same results (which is my goal actually, to show both algorithms will work).   I am not sure what kind of testing should I do, what statistic, numbers etc I want to obtain to show, also this is a heuristic kind of problem so things are generated randomly.   Any advice will be greatly appreciated.  Thanks in advance.

Source: http://www.livejournal.com/community/algorithms/59939.html

Date: 07/08/05     Keywords: no keywords

It seems there's no distributed/concurrent computing community on LJ ? Anyone interests in the area ? if so probably we should create one community for it. Also how do you post a message to multiple communities at once ?

Source: http://www.livejournal.com/community/algorithms/59445.html

Date: 07/08/05     Keywords: no keywords

How do I do weighted cross correlation of images using Fourier transforms? (That is, each pixel being weighted, by something like a mask.) I don't want to do from-scratch coding -- I'd preferably like to work using existing FFT functions.

Source: http://www.livejournal.com/community/algorithms/58997.html

Date: 07/08/05     Keywords: no keywords

This feels like something I should know, but... I don’t.

How does one map Cartesian planar points to a 2-d curvilinear coordinate space? Or the reverse?

My specific problem is this: Say I have an image with translational symmetry, but along a pair of open curves. Once I identify the curves, I want to ‘unwarp’ the image so the translation is along scalar orthogonal coordinates. (So I have to map the matrix that represents the original image into one that represents the new image. I'm working with MATLAB -- in case anyone knows of some quick tricks in it.)

Any suggestions or pointers will be appreciated. Thanks.

Source: http://www.livejournal.com/community/algorithms/58634.html

Date: 07/05/05     Keywords: no keywords

Hi,

Does anyone know a program that can convert a picture into a fractal?

Source: http://www.livejournal.com/community/algorithms/58596.html

Date: 06/27/05     Keywords: no keywords

So I am writing a program that will calculate the number of days since 1/1/1600 till the date passed to the function.
I used this calculator to check my test cases so that I don't have to do them all by hand.
However of the ~50 cases I put in all work, except for 09/01/6912. My code gives me 1940413. I get that by hand too. (365*5312 + 13*97 [97 leap years in block of 400)+ 25 [25 in the first hundred years of the block of 400] + 3 [3 blocks of 4] + 243 [number of days from jan-aug, counting feb as 28] + 1 + 1 [adding th e leap day since it is after febuary on a leap year] - 1 [don't count in the end points]) The calculator gives 1940412. Arrrrr.

Am I just being dumb, and calculating it wrong by hand too? or is that calculator messing up?

(leap years being years divisble by 4, unless they are divisible by 100, in which case they have to be divisible by 400.)

Edit: caught it. Fixed it. Thanks. =)

Source: http://www.livejournal.com/community/algorithms/57925.html

Date: 06/26/05     Keywords: software, google

Well, without preludes :)
I need a software synchronous detection algorithm. I have been searching Google for already about a week but found nothing but mentions of using synchronous detection here and there; almost in all cases it was analog and implemented in hardware.
Please someone point me to the right place where to find any information about this!

Source: http://www.livejournal.com/community/algorithms/57677.html

Date: 06/25/05     Keywords: software

PREFACE:
By popular demand, here is the first of the Design of the Day series.

INTENT:
1) Much like "Word of the Day" only related to Algorithms, Advanced Datastructures, or Patterns.
2) Expand knowledge of useful design principles, much like expanding your vocabulary.

FINALLY:
I belive it would be mutually beneficial to evaluate wikipedia definitions and make corrections/additions when approved unanimously here in this community.

//---

Without further Ado: The Abstract Factory (wikipedia).

Summary:
"Abstract Factory Pattern provides a way to encapsulate a group of individual factories that have a common theme. In normal usage, the client software would create a concrete implementation of the abstract factory and then use the generic interfaces to create the concrete objects that are part of the theme. The client does not know (nor care) about which concrete objects it gets from each of these internal factories since it uses only the generic interfaces of their products. This pattern separates the details of implementation of a set of objects from its general usage"

Example:
"An example of this would be an abstract factory class documentCreator that provides interfaces to create a number of products (eg. createLetter() and createResume()). The system would have any number of derived concrete versions of the documentCreator class like fancyDocumentCreator or modernDocumentCreator, each with a different implementation of createLetter() and createResume() that would create a corresponding object like fancyLetter or modernResume."



UML Diagram

Source: http://www.livejournal.com/community/algorithms/57445.html

Date: 06/24/05     Keywords: no keywords

I need to get a quantitative indication of how similar two given images are to each other.

The existing algorithm I'm working with takes the sum of squared differences for each pixel. But this doesn't work too well if the proportion of whitespace is high -- for instance, if we have just a thin vertical line on a large white background, it would give a good correlation score with the 90 degree rotaion of itself, since most of the pixels would agree after the rotation. But to a human, the two images would register a low correlation.

This is because, presumably, we filter out the background when we make a mental comparison. So my idea is to somehow have the program separate background hue-range from foreground, so it can ignore a large near-monochrome background. So my first question is: does this intuition make sense? If not, any ideas on how to better go about it?

Edit: Sorry I wasn't more specific. What I'm looking at is the correlation of an image under rotation or reflection with itself. So, mathematically, testing for symmetries, where a good correlation denotes the existence of a symmetry about that transformation. And the images in question do often tend to be photographs, so edge identification is a little more difficult.

Source: http://www.livejournal.com/community/algorithms/57326.html

Date: 06/24/05     Keywords: programming

If you are like me, you get your word of the day to expand your vocabulary, and maybe a QOTD for something funny, insightful, etc.

I was thinking how much more valuable an Algorithm or Design Pattern would be.
Is that something that would be of interest to everyone?

For instance: We post Creational Patterns - the Abstract Factory and then discuss its uses, abuses (anti-patterns) etc.
Maybe reference a code sample, etc.

The topics don't need to be uber-advanced, just practical.
Much like adding words to our vocabulary, I think it would be useful to add STRATEGIES to our programming.

Thoughts?

Source: http://www.livejournal.com/community/algorithms/56972.html

Date: 06/23/05     Keywords: no keywords

Here is my first attempt at a PULL based Master-Slave replication which also allows
* Lazy Loading
* Localized Caching
* DOS prevention (At least limits resource consumption)

Questions:
1) Is this design valid and effecient?
2) Given the problem domain, can anyone propose a non-locking strategy for the Process Fetch Latest ACL List From Master?


I should have written this in UML, but I like to brainstorm in flowcharts:

Source: http://www.livejournal.com/community/algorithms/56644.html

Date: 06/12/05     Keywords: no keywords

given two disjoint convex geometries, what is the fastest way for finding a seperating plane?

Source: http://www.livejournal.com/community/algorithms/56357.html

Date: 06/07/05     Keywords: no keywords

So what is the best way to store really huge numbers(>1000 digits) in c++?

Also how do I get 1/c to be as precise as I want it?

Source: http://www.livejournal.com/community/algorithms/56303.html

Date: 06/07/05     Keywords: java, web

Dear Group Member,


The Percepts and Concepts Laboratory at Indiana University would like to to invite you to participate in our interactive, on-line group experiments. The games are entertaining and should be intriguing to those interested in the sciences, gaming etc. Some of the games that we are currently working on include:

Forager: In this experiment, we are interested in how groups of people allocate themselves to resources. Your goal in a four-minute experiment is to pick up as many resources as you can by moving your icon with the arrow keys.

Spread of Innovations: This experiment explores how innovative ideas spread in a group. As you try to solve a puzzle, you see the solutions proposed by your neighbors, and they see your solutions.

Opinion Sharing: You rate things (e.g. faces, music, or paintings) and can see the ratings given by others. Over several rounds of ratings, does your group converge in its opinion or not?

In all of these experiment, you play with other humans when they are available or with artificial intelligence bots when no other humans are currently on-line. By playing our games, our GEE group benefits because we collect data on how collective behavior in human groups emerges, and you benefit by 1) learning about current research questions and methods in the social sciences, 2) sharing an engrossing, communal experience with others people (or bots!), 3) playing games that have deeper significance and more interesting dynamics than solitaire, tetris, or the typical fodder of on-line games. The experiments will run on any computer that has Java 1.4 or later, and can be found at:

http://groups.psych.indiana.edu/

If you like the experiments, please tell your friends or favorite news group, blog, or forum. The more people who are in our virtual world at any time, the more fun it is (and the more data we collect). Also, please reply to this email is you would like to be placed on our mailing list, or if you're willing to place a link to our site on your web page. If you find any bugs or have suggestions for improvements, please email Andy Jones or Allen Lee , and CC Rob Goldstone (rgoldsto@indiana.edu).



Thank you for being part of the global brain,

The Group Experiment Environments (GEE) team

Robert Goldstone - Director

Rowan Johnston - PR/Marketing/Research

Andy Jones - Programmer/Researcher

Allen Lee - Programmer/Researcher

Zoran Rilak - Programmer/Researcher

Michael Roberts - Artificial Inteligence and Computational Modeler

Source: http://www.livejournal.com/community/algorithms/55983.html

Date: 06/04/05     Keywords: no keywords

Hey everyone! This is a shameless plug for my new automated freebie finder. I'm interested in people's comments and thoughts about my technique. I scrape other sites and look for recurrent links to find relevant freebies. It's completely automated. If anyone wants to spread the word about my new site, I'd be indebted.

http://absurdlycool.com/

Thoughts?

Source: http://www.livejournal.com/community/algorithms/55574.html