Tag: algorithms

Entries for tag "algorithms", ordered from most recent. Entry count: 62.

Uwaga! Informacje na tej stronie mają ponad 3 lata. Nadal je udostępniam, ale prawdopodobnie nie odzwierciedlają one mojej aktualnej wiedzy ani przekonań.

Pages: > 1 2 3 4 5 ... 8 >

Aug 2009

Calculating Linear and Quadratic Equation Coefficients

Some time ago I've written about formulas to calculate coefficients of linear and quadratic equation [pl] when having 2 or 3 given points (x,y). Yesterday I suddenly noticed that my code needs to do it every frame so I need functions to calculate these coefficients. Below you can find the code of my functions. They are templates, so they work with many types including float, double, as well as vectors of any dimmension and other types which have addition, subtraction and assignment operators, as well as multiplication and divistion by float scalar.

Read full entry > | Comments (3) | Tags: rendering math algorithms | Author: Adam Sawicki | Share

Aug 2009

Source Code and Ray Tracer

As some of you are interested in seeing source code of my 2D Software Renderer, I'll fullfill this request, although I don't think there is anything special about this code. Here is the file: SoftRender01.zip and here is a brief description of this project:

BTW, I've started coding a raytracer. Of course the first primitive I support is a sphere and first effects I've coded are shadows and reflections. Ray tracing is not magic. It's just like "foreach pixel, foreach object, render", while rasterization is the opposite: "foreach object, foreach pixel, render". But It's quite fascinating that raytracer suppors directly some of effects that require special preprocessing in standard rasterization (like preparing shadow map and enviromental map in case of my effects).


Comments (3) | Tags: productions rendering algorithms | Author: Adam Sawicki | Share

Jul 2009

Three Ways to Calculate Mean and Variance

There is a free e-book on DSP (Digital Signal Processing) called The Scientist and Engineer's Guide to Digital Signal Processing (by Steven W. Smith, Ph.D.). As I have a holiday now, I've started reading it and I've found some interesting information even in introductory chapters. For example, there are three algorithms to calculate mean and variance. Let's say we have some data in a vector of bytes and we want to calculate its statistics.

std::vector<unsigned char> Bytes;
// Fill Bytes...
uint N = Bytes.size();

Traditional algorithm looks like this:

float Mean = 0.f, Variance = 0.f;
for (uint i = 0; i < N; i++)
  Mean += (float)Bytes[i];
Mean /= (float)N;
for (uint i = 0; i < N; i++)
  Variance += ((float)Bytes[i] - Mean) * ((float)Bytes[i] - Mean);
Variance /= (float)(N-1);

Now the incremental one, which can return current mean and variance at any time while you can also post next piece of data. All we need to implement it is to keep track of current sample number, sum of samples and sum of squared samples. I've created template class for that, which can be parametrized by float, double, int or other numeric types.

template <typename T>
class IncrementalMeanAndVarianceCalc
  IncrementalMeanAndVarianceCalc() : m_Sum(T()), m_SqSum(T()), m_Count(0) { }
  void Clear() { m_Sum = m_SqSum = T(); m_Count = 0; }
  void Add(const T &v)
    m_Sum += v;
    m_SqSum += v*v;
  void Add(const T *values, uint valCount)
    for (uint i = 0; i < valCount; i++)
      m_Sum += values[i];
      m_SqSum += values[i]*values[i];
    m_Count += valCount;
  bool IsEmpty() { return m_Count == 0; }
  uint GetCount() { return m_Count; }
  T GetSum() { return m_Sum; }
  T GetSqSum() { return m_SqSum; }
  T GetMean()
    return m_Sum / m_Count;
  T GetVariance(bool varianceBiased = true)
    if (varianceBiased)
      return (m_SqSum - m_Sum*m_Sum/m_Count) / (m_Count-1);
      return (m_SqSum - m_Sum*m_Sum/m_Count) / m_Count;
  void GetMeanAndVariance(T &outMean, T &outVariance, bool varianceBiased = true)
    outMean = m_Sum / m_Count;
    if (varianceBiased)
      outVariance = (m_SqSum - m_Sum*m_Sum/m_Count) / (m_Count - 1);
      outVariance = (m_SqSum - m_Sum*m_Sum/m_Count) / m_Count;

  T m_Sum, m_SqSum;
  uint m_Count;

Finally, there is an algorithm to calculate mean and variance from histogram. It is very efficient method. If we have only 256 possible values for each sample, we don't have to do binning and calculating histogram is very simple:

uint Histogram[256] = { 0 };
for (uint i = 0; i < N; i++)

Now, the mean and variance can be calculated this way:

float Mean = 0.f, Variance = 0.f;
for (uint i = 0; i < 256; i++)
  Mean += (float)i * Histogram[i];
Mean /= (float)N;
for (uint i = 0; i < 256; i++)
  Variance += (i - Mean)*(i - Mean) * Histogram[i];
Variance /= N - 1;

Comments (2) | Tags: algorithms math | Author: Adam Sawicki | Share

Jul 2009

What is GUID and how to use it

There are many ways to identify objects in a collection. One of them is to keep direct pointers to objects. Another way is to simply index their positions in an array. We can make user see only some strange indirect "handles" without any obvious meaning, like HWND or FILE*. We can also give objects string names or some numeric identifiers. SQL databases often use integer number to identify record in a table and new identifiers are generated as subsequent numbers. Another solution is to use GUIDs.

GUID means Globally Unique Identifier and it is a kind of general purpose, 128-bit numeric identifier. GUIDs are generated from some random data. We can be sure about their uniqueness in any context because of their size, as 2^128 possible values is more than stars in the whole observable universe :) That's why they are called "globally unique".

We could just generate 128 bits of data from any good pseudorandom number generator and call it good identifiers, but to standardize this idea, they have created standard - RFC4122 (it's actually UUID, but let's ignore the difference). This standard says how to express GUIDs as strings, made of hexadecimal numbers separated by dashes (for example "f81d4fae-7dec-11d0-a765-00a0c91e6bf6"). You can find such GUIDs everywhere. For example, run "regedit.exe" and look into key HKEY_CLASSES_ROOT\CLSID or run your Visual C++ and run command Tools / Create GUID from the main menu.

The standard also defines bit structure of the UUID and algorithms to generate it. There are several possible algorithms. Some of them use system time or even network card MAC address. The simplest one is version 4 algorithm, which uses only random number generator. Some of bits have special meaning and code algorithm which have been used to generate particular GUID.

Here is my code in C++ defining GUID structure and the algorithm to generate it.

Read full entry > | Comments (2) | Tags: c++ algorithms | Author: Adam Sawicki | Share

Jul 2009

Efficient Finding of Duplicated Files

I've recently wrote a tool for finding duplicated files in a given directory. I wanted to do it efficiently and here is my algoritm.

My basic idea is to first recursively enumerate all files in the directory and its subdirectories and sort their descriptions by file size. Making it this way, we can then iterate through this list and for each file look ahead to see how many other files have same size. If two files have different sizes, they are obviously not equal in terms of their content. So if there is only one file with particular size, it can just be skipped. If there are two files with same size, we must just compare them by content.

If there are many files with same size, things become more complicated, as any possible combinations of them can turn out to be identical. My solution is to compare two files by a "header" (lets say - first 1024 bytes) and if they are equal - by MD5 checksum of their whole content. Checksum of each file is lazy evaluated, so it's calculated only once, the first time its needed.

As I iterate through the sorted file list, I mark reported files not to process them again. I can do it because I ensure that if a file is reported, all other identical files are also already reported. In my real code I do the same with files I encounter errors with, but here I wanted to simplify the code.

Read full entry > | Comments (1) | Tags: algorithms winapi | Author: Adam Sawicki | Share

Jul 2009


People at forums are usually advised to learn BSP, quadtree or octree as space partitioning data structure, but there are many more interesting structures than these three. This time my choice for home project is KD-tree. It's actually something between BSP and octree. Just like BSP it's a binary tree (each non-leaf node has two child nodes) and optimal splitting plane is estimated each time by special algorithm, but splitting planes are always aligned to one of three main axes and thus each node can be described by an AABB (axis-aligned bounding box), just like in octree.

As my tree is designed to manage objects and not geometry, nothing can be split and some of objects may intersect splitting planes. How to deal with them? I simply assign them to the parent node, so not only leaves are allowed to contain list of objects. To avoid too many small objects intersecting splitting planes to degrade performance by falling into top level nodes (it's called "sticky planes" or something like that :) I adopted "loose octree" idea to my KD-tree. It simply means I extend each node’s bounding box so that each node's children slightly overlap each other and small objects intersecting splitting plane fall into one of the children.

My KD-tree is also dynamic, which means it reorganizes itself as objects get added, removed and moved in the tree. It's actually quite simple. Each time an object is added to a node, that node can be split into two children if it's object number exceeds constant limit. Similarly node can be merged by deleting it's children each time an object is removed from one of its children, if the number of objects in that node and its children drops under constant minimum.

For additional performance, I allocate tree nodes from my own "Free List" memory pool and I keep objects connected to each node as doubly-linked list.

I also came up with an idea how to easily visualize quality of my space partitioning technique. I keep track of the number of tree nodes and objects on each depth level. This way I can tell from these several numbers whether tree is more "tall" or "wide" and whether most of objects stay in leaves instead of some top-level nodes.

Here are some screenshots and a video from my recent code:

KD-tree KD-tree KD-tree

Comments (3) | Tags: rendering video gallery algorithms math | Author: Adam Sawicki | Share

May 2009

Manager zasobów #2 - Wczytywanie w tle

Sam manager zasobów napisałem wg podobnych założeń, jakie miałem w TFQ7. Jest jeden globalny manager zasobów g_ResMngr przechowujący kolekcję wszystkich zasobów. Zasób jest obiektem klasy pochodnej od Resource implementującej odpowiednie metody wirtualne. Zasób może mieć nazwę i można wyszukiwać zasoby wg nazwy, ale może też mieć nazwę pustą. Zasoby można swobodnie tworzyć i usuwać. Obiekt klasy Resource istnieje przez cały czas życia zasobu, a wewnątrz pamięta stan - niezaładowany, w tracie ładowania, załadowany itd.

class Resource
  enum STATE {
  Resource(const string &Name);
  virtual ~Resource();
  STATE GetState() { return m_State; }
  bool IsLoaded() { return m_State == STATE_LOADED; }

  void Load(); // Żąda załadowania już teraz
  void BeginLoading(); // Rozpoczyna ładowanie w tle
  void Unload(); // Odładowuje

  virtual void OnLoad() = 0;
  virtual void GetLoadType(bool *OutUseBkg, BkgJob::TYPE *OutBkgJobType) = 0;
  virtual void OnLoadBkg() { } // Wykonywana na osobnym wątku
  virtual void OnLoadAfterBkg() { }
  virtual void OnUnload() = 0;

  string m_Name;
  STATE m_State;

class ResourceManager
  Resource * Find(const tstring &Name);
  Resource * MustFind(const tstring &Name);
  template <typename T> T * FindEx(const tstring &Name) { /*...*/ }
  template <typename T> T * MustFindEx(const tstring &Name) { /*...*/ }

extern ResourceManager * g_ResMngr;

Read full entry > | Comments (2) | Tags: algorithms c++ engine | Author: Adam Sawicki | Share

May 2009

Manager zasobów #1 - BkgJobManager

Tajemniczy "Ciekawski" prosił, żebym opisał mój asynchroniczy manager zasobów. Zanim go opiszę, muszę napisać słowo o tym, na czym się on opiera - o module do wykonywania zadań w tle.

Ogólnie chodzi o to, aby osobny wątek pracujący w tle wykonał jakieś określone zadanie. Problem w tym, że tworzenie za każdym razem nowego wątku jest powolne. Poza tym przydałoby się, żeby takie zadania były wykonywane po kolei, a nie wszystkie na raz. Dlatego napisałem globalny BkgJobManager, który ma na stałe utworzone wątki, a zadania do wykonania dodaje się do jego kolejki jako obiekty specjalnej klasy BkgJob.

class BkgJobManager {
  void Init();
  void Frame();
  void AddJob(BkgJob *Job);
extern BkgJobManager *g_BkgJobManager;

Żeby zdefiniować swoje zadanie, trzeba odziedziczyć po klasie BkgJob i zaimplementować metody: OnWork (wywoływaną na osobnym wątku) i opcjonalnie OnWorkDone (wywoływaną po zakończeniu, już na głównym wątku, w ramach wywołania BkgJobManager::Frame).

class BkgJob {
  BkgJob(TYPE Type, MODE Mode);
  virtual void OnWork() = 0;
  virtual void OnWorkDone() { }

Dodatkowo, klasa pochodna określa typ zadania jako obliczeniowe (mocno angażujące procesor) lub wejścia-wyjścia (wczytujące coś z dysku). To mój oryginalny pomysł oparty na przemyśleniu, że najoptymalniej będzie, jeśli na raz będzie się mogło wykonywać tylko tyle zadań obliczniowych, ile jest rdzeni w procesorze (więcej obniżyłoby wydajność przez częste przełączanie się procesora między wątkami) i tylko jedno zadanie wejścia-wyjścia (więcej obniżyłoby wydajność przez przeskakiwanie głowicy dysku między wieloma czytanymi na raz plikami).


BkgJobManager przechowuje kolejkę zadań do wykonania. Żeby to napisać porządnie, to pewnie powinna być jakaś struktura "lockless", ale ja póki co załatwiłem synchronizację zwykłym muteksem.

Ponadto zadanie ma swój priorytet. Na zadanie można też poczekać, np. wywołując na wątku głównym BkgJob::Join. To wywołanie zwróci sterowanie dopiero, kiedy dane zadanie się zakończy. Jeśli to zadanie czeka gdzieśtam w kolejce, to jego priorytet zostaje podbity, żeby trafiło na przód kolejki.

Ciekawym rozwiązaniem jest, że na wątku głównym należy wołać w każdej klatce (lub inaczej, ale możliwie często) BkgJobManager::Frame. Daje to okazję managerowi, aby "zebrać" wykonane w tle i zakończone zadania, wywołać im BkgJob::OnWorkDone i zwolnić je z pamięci.

Trzeba też pomyśleć, jak z wykonywanej na innym wątku funkcji BkgJob::OnWork przekazywać informację o niepowodzeniu. Ponieważ w swoim kodzie używam wyjątków, łapię wyjątek zgłoszony w BkgJob::OnWork i zachowuję jego obiekt, żeby na wątku głównym móc go potem odczytać.

Podsumowując: Oryginalnie wątek służy do tego, żeby natychmiast rozpocząć wykonywanie jakiejś pracy w tle albo żeby działać cały czas w pętli czekając na jakieś komunikaty. Mechanizm taki jak tutaj przedstawiłem pozwala zmienić koncepcję na taką, w której użytkownik może tworzyć zadania i dodawać je do kolejki, a one będą po kolei wykonywane w tle. Podobny kod - JobSwarm - umieścił na swoim blogu John Ratcliff. Mój można pobrać stąd: BackgroundJob.hpp, BackgroundJob.cpp.

Tego mojego modułu mogę teraz używać do różnych rzeczy, ale podstawowym (i póki co jedynym :) zastosowaniem jest wczytywanie w tle zasobów Direct3D. O samym managerze zasobów napiszę następnym razem...

Comments (0) | Tags: algorithms c++ engine | Author: Adam Sawicki | Share

Pages: > 1 2 3 4 5 ... 8 >

STAT NO AD [Stat] [Admin] [STAT NO AD] [pub] [Mirror] Copyright © 2004-2016 Adam Sawicki
Copyright © 2004-2016 Adam Sawicki