I am writing some CUDA code for finding the 3 parameters of a circle (centre X,Y & radius) from many (m) measurements of positions around the perimeter.
As m > 3 I am (successfully) using Singular Value Decomposition (SVD) for this purpose (using the cuSolver library). Effectively I am solving m simulaneous equations with 3 unknowns.
However, not all of my perimeter positions are valid (say q of them), and so I have to go through my initial set of m measurements and remove the q invalid ones. This involves moving the size m data array from the card to the host, processing linearly to remove the q invalid entries and then re loading the smaller (m-q) array back onto the card...
My question is; if I were to set all terms on both sides of the q invalid equations to zero, could I just run the m equations (including the zeros) through my SVD analysis (without the data transfer etc) or would this cause other problems?
My instinct tells me that this is a bit like applying weights to the data but instinct and SVD are not terms that sit well together in my experience...
I am hesitant just to try this as I don't know if it will work in some cases and not in others...
I have tested the idea by inserting rows of zeros into my matrix. The solution that I am getting is not significantly affected by this.
So I am answering my own question with a non-rigorous Yes it is OK do do this.
If anybody has a more rigorous or more considered answer I would very much like to hear it.
This question already has answers here:
Is it possible to program in binary?
(6 answers)
Closed 2 years ago.
Is that possible? Can this be done using just 1 and 0 (true/false, on/off ...)?
If so, how would this code look?
If this example is too complex i am open to all other kinds of examples, but would like to have an operation included, because i have no idea how such operations get encoded (i guess they also are just an entry in a conversion chart)
The reason why i ask this, is that i want to give people a concrete example why datatypes and functions/operations are a practical abstraction (easier to read). Im writing a tutorial.
In a 1-bit wide integer = boolean value, carry-out has nowhere to go, so addition simplifies to just XOR.
Fun fact: XOR is add-without-carry. It's part of implementing a single-bit adder out of logic gates, e.g. a "half adder" that has 2 inputs (no carry-in) and produces a sum and carry-out. (sum = a xor b, carry = a AND b). A simple 32-bit adder could be build out of a half adder and 31 "full adders". Or more adders in parallel with tricks to optimize it for lower latency than a simple ripple-carry binary adders.
Carryless multiplication is a thing in some crypo, where summing partial products is done with XOR instead of normal binary addition.
See also What is the best way to add two numbers without using the + operator? for a software use of the same idea.
I suck at math, so I can't figure this out: how many combinations of k neighboring pixels are there in an image? Combinations of k pixels out of n * n total pixels in the image, but with the restriction that they must be neighbors, for each k from 2 to n * n. I need the sum for all values of k for a program that must take into account that many elements in a set that it's reasoning about.
Neighbors are 4-connected and do not wrap-around.
Once you get the number of distinct shapes for a blob of pixels of size k (here's a reference) then it comes down to two things:
How many ways on your image can you place this blob?
How many of these are the same so that you don't double-count (because of symmetries)?
Getting an exact answer is a huge computational job (you're looking at more than 10^30 distinct shapes for k=56 -- imagine if k = 10,000) but you may be able to get good enough for what you need by fitting for the first 50 values of k.
(Note: the reference in the wikipedia article takes care of duplicates with their definition of A_k.)
It seems that you are working on a problem that can be mapped to Markovian Walks.
If I understand your question, you are trying to count paths of length k like this:
Start (end)-> any pixel after visiting k neighbours
* - - - - -*
| |
| |
- - - -
in a structure that is similar to a chess board, and you want to connect only vertical and horizontal neighbours.
I think that you want the paths to be self avoiding, meaning that a pixel should not be traversed twice in a walk (meaning no loops). This condition lead to a classical problem called SAWs (Self Avoiding Walks).
Well, now the bad news: The problem is open! No one solved it yet.
You can find a nice intro to the problem here, starting at page 54 (or page 16, the counting is confusing because the page numbers are repeating in the doc). But the whole paper is very interesting and easy to read. It manages to explain the mathematical background, the historical anecdotes and the scientific importance of markovian chains in a few slides.
Hope this helps ... to avoid the problem.
If you were planning to iterate over all possible polyominos, I'm afraid you'll be waiting a long time. From the wikipedia site about polyominos, it's going to be at least O(4.0626^n) and probably closer to O(8^n). By the time n=14, the count will be over 5 billion and too big to fit into an int. By time n=30, the count will be more than 17 quintillion and you won't be able to fit it into a long. If all the world governments pooled together their resources to iterate through all polyominos in a 32 x 32 icon, they would not be able to do it before the sun goes supernova.
Now that doesn't mean what you want to do is intractable. It is likely almost all the work you do on one polyominal was done in part on others. It may be a fun task make an exponential speedup using dynamic programming. What is it you're trying to accomplish?
I have to do some work with Q Learning, about a guy that has to move furniture around a house (it's basically that). If the house is small enough, I can just have a matrix that represents actions/rewards, but as the house size grows bigger that will not be enough. So I have to use some kind of generalization function for it, instead. My teacher suggests I use not just one, but several ones, so I could compare them and so. What you guys recommend?
I heard that for this situation people are using Support Vector Machines, also Neural Networks. I'm not really inside the field so I can't tell. I had in the past some experience with Neural Networks, but SVM seem a lot harder subject to grasp. Are there any other methods that I should look for? I know there must be like a zillion of them, but I need something just to start.
Thanks
Just as a refresher of terminology, in Q-learning, you are trying to learn the Q-functions, which depend on the state and action:
Q(S,A) = ????
The standard version of Q-learning as taught in most classes tells you that you for each S and A, you need to learn a separate value in a table and tells you how to perform Bellman updates in order to converge to the optimal values.
Now, lets say that instead of table you use a different function approximator. For example, lets try linear functions. Take your (S,A) pair and think of a bunch of features you can extract from them. One example of a feature is "Am I next to a wall," another is "Will the action place the object next to a wall," etc. Number these features f1(S,A), f2(S,A), ...
Now, try to learn the Q function as a linear function of those features
Q(S,A) = w1 * f1(S,A) + w2*f2(S,A) ... + wN*fN(S,A)
How should you learn the weights w? Well, since this is a homework, I'll let you think about it on your own.
However, as a hint, lets say that you have K possible states and M possible actions in each state. Lets say you define K*M features, each of which is an indicator of whether you are in a particular state and are going to take a particular action. So
Q(S,A) = w11 * (S==1 && A == 1) + w12 * (S == 1 && A == 2) + w21 * (S==2 && A==3) ...
Now, notice that for any state/action pair, only one feature will be 1 and the rest will be 0, so Q(S,A) will be equal to the corresponding w and you are essentially learning a table. So, you can think of the standard, table Q-learning as a special case of learning with these linear functions. So, think of what the normal Q-learning algorithm does, and what you should do.
Hopefully you can find a small basis of features, much fewer than K*M, that will allow you to represent your space well.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 2 years ago.
Improve this question
Although I know the basic concepts of binary representation, I have never really written any code that uses binary arithmetic and operations.
I want to know
What are the basic concepts any
programmer should know about binary
numbers and arithmetic ? , and
In what "practical" ways can binary
operations be used in programming. I
have seen some "cool" uses of shift
operators and XOR etc. but are there
some typical problems where using binary
operations is an obvious choice.
Please give pointers to some good reference material.
If you are developing lower-level code, it is critical that you understand the binary representation of various types. You will find this particularly useful if you are developing embedded applications or if you are dealing with low-level transmission or storage of data.
That being said, I also believe that understanding how things work at a low level is useful even if you are working at much higher levels of abstraction. I have found, for example, that my ability to develop efficient code is improved by understanding how things are represented and manipulated at a low level. I have also found such understanding useful in working with debuggers.
Here is a short-list of binary representation topics for study:
numbering systems (binary, hex, octal, decimal, ...)
binary data organization (bits, nibbles, bytes, words, ...)
binary arithmetic
other binary operations (AND,OR,XOR,NOT,SHL,SHR,ROL,ROR,...)
type representation (boolean,integer,float,struct,...)
bit fields and packed data
Finally...here is a nice set of Bit Twiddling Hacks you might find useful.
Unless you're working with lower level stuff, or are trying to be smart, you never really get to play with binary stuff.
I've been through a computer science degree, and I've never used any of the binary arithmetic stuff we learned since my course ended.
Have a squizz here: http://www.swarthmore.edu/NatSci/echeeve1/Ref/BinaryMath/BinaryMath.html
You must understand bit masks.
Many languages and situations require the use of bit masks, for example flags in arguments or configs.
PHP has its error level which you control with bit masks:
error_reporting = E_ALL & ~E_NOTICE
Or simply checking if an int is odd or even:
isOdd = myInt & 1
I believe basic know-hows on binary operations line AND, OR, XOR, NOT would be handy as most of the programming languages support these operations in the form of bit-wise operators.
These operations are also used in image processing and other areas in graphics.
One important use of XOR operation which I can think of is Parity check. Check this http://www.cs.umd.edu/class/sum2003/cmsc311/Notes/BitOp/xor.html
cheers
The following are things I regularly appreciate knowing in my quite conventional programming work:
Know the powers of 2 up to 2^16, and know that 2^32 is about 4.3 billion. Know them well enough so that if you see the number 2147204921 pop up somewhere your first thought is "hmm, that looks pretty close to 2^31" -- that's a very effective module for your bug radar.
Be able to do simple arithmetic; e.g. convert a hexadecimal digit to a nybble and back.
Have some vague idea of how floating-point numbers are represented in binary.
Understand standard conventions that you might encounter in other people's code related to bit twiddling (flags get ORed together to make composite values and AND checks if one's set, shift operators pack and unpack numbers into different bytes, XOR something twice and you get the same something back, that kind of thing.)
Further knowledge is mostly gravy unless you work with significant performance constraints or do other less common work.
At the absolute bare minimum you should be able to implement a bit mask solution. The tasks associated with bit mask operations should ensure that you at least understand binary at a superficial level.
From the top of my head, here are some examples of where I've used bitwise operators to do useful stuff.
A piece of javascript that needed one of those "check all" boxes was something along these lines:
var check = true;
for(var i = 0; i < elements.length; i++)
check &= elements[i].checked;
checkAll.checked = check;
Calculate the corner points of a cube.
Vec3f m_Corners[8];
void corners(float a_Size){
for(size_t i = 0; i < 8; i++){
m_Corners[i] = a_Size * Vec3f(axis(i, Vec3f::X), axis(i, Vec3f::Y), axis(i, Vec3f::Z));
}
}
float axis(size_t a_Corner, int a_Axis) const{
return ((a_Corner >> a_Axis) & 1) == 1
? -.5f
: +.5f;
}
Draw a Sierpinski triangle
for(int y = 0; y < 512; y++)
for(int x = 0; x < 512; x++)
if(x & y) pixels[x + y * w] = someColor;
else pixels[x + y * w] = someOtherColor;
Finding the next power of two
int next = 1 << ((int)(log(number) / log(2));
Checking if a number is a power of two
bool powerOfTwo = number & (number - 1);
The list can go on and on, but for me these are (except for Sierpinksi) everyday examples. Once you'll understand and work with it though, you'll encounter it in more and more places such as the corners of a cube.
You don't specifically mention (nor rule out!-) floating point binary numbers and arithmetic, so I won't miss the opportunity to flog one of my favorite articles ever (seriously: I sometimes wish I could make passing a strict quiz on it a pre-req of working as a programmer...;-).
The most important thing every programmer should know about binary numbers and arithmetic is : Every number in a computer is represented in some kind of binary encoding, and all arithmetic on a computer is binary arithmetic.
The consequences of this are many:
Floating point "bugs" when doing math with IEEE floating point binary numbers (Which is all numbers in javascript, and quite a few in JAVA, and C)
The upper and lower bounds of representable numbers for each type
The performance cost of multiplication/division/square root etc operations (for embedded systems
Precision loss, and accumulation errors
and more. This is stuff you need to know even if you never do a bitwise xor, or not, or whatever in your life. You'll still run into these things.
This really depends on the language you're using. Recent languages such as C# and Java abstract the binary representation from you -- this makes working with binary difficult and is not usually the best way to do things anyway in these languages.
Middle and low level languages like C and C++, however, require you to understand quite a bit about how the numbers are stored underneath -- especially regarding endianness.
Binary knowledge is also useful when implementing a cross platform protcol of some sort .... for example, on x86 machines, byte order is little endian. but most network protocols want big endian numbers. Therefore you have to realize you need to do the conversion for things to go smoothly. Many RFCs, such as this one -> https://www.rfc-editor.org/rfc/rfc4648 require binary knowledge to understand.
In short, it's completely dependent on what you're trying to do.
Billy3
It's handy to know the numbers 256 and 65536. It's handy to know how two's complement negative numbers work.
Maybe you won't run into a lot of binary. I still use it pretty often, but maybe out of habit.
A good familiarity with bitwise operations should make you more facile with boolean algebra, and I think that's important for every programmer--you want to be able to quickly simplify complex logic expressions.
Absolute minimum is, that "2" is not a binary digit and 10b is smaller than 3.
If you never do low-level programming (like C in embedded systems), never have to use a debugger, and never have to work with real numbers, then I suppose you could get by without knowing binary. But knowing binary will make you a stronger programmer, even if indirectly.
Once you venture into those areas you will need to know binary (and its ``sister'' base, hexadecimal). Without knowing it:
Embedded systems programming would be impossible.
Debugging would be hard because you wouldn't know what you were looking at in memory.
Numerical calculations with decimals would give you answers you don't understand.
I learned to twiddle bits back when c and asm were still used for "mainstream" programming. Although I no longer have much use for that knowledge, I recently used it to solve a real-world business problem.
We use a fax service that posts a message back to us when the fax has been sent or failed after x number of retries. The only way I had to identify the fax was a 15 character field. We wanted to consolidate this into one URL for all of our clients. Before we consolidated, all we had to fit in this field was the FaxID PK (32 bit int) column which we just sent as a string.
Now we had to identify the client (a 4 character code) and the database (32 bit int) underneath the client. I was able to do this using base 64 encoding. Without understanding the binary representation of numbers and characters, I probably would never have even thought of this solution.
Some useful information about the number system.
Binary | base 2
Hexadecimal | base 16
Decimal | base 10
Octal | base 8
These are the most common.
Converting them is faily easy.
112 base 8 = (1 x 8^2) + (2 x 8^1) + (4 x 8^0)
74 base 10 = (7 x 10^1) + (4 x 10^0)
The AND, OR, XOR, and etc. are used in logic gates. Search boolean algebra, something well worth the time knowing.
Say for instance, you have 11001111 base 2 and you want to extract the last four only.
Truth table for AND:
P | Q | R
T | T | T
T | F | F
F | F | F
F | T | F
You can use 11001111 base 2 AND 00111111 base 2 = 00001111 base 2
There are plenty of resources on the internet.