Search Results

Search found 1909 results on 77 pages for 'adjacency matrix'.

Page 70/77 | < Previous Page | 66 67 68 69 70 71 72 73 74 75 76 77  | Next Page >

  • Bilinear interpolation - DirectX vs. GDI+

    - by holtavolt
    I have a C# app for which I've written GDI+ code that uses Bitmap/TextureBrush rendering to present 2D images, which can have various image processing functions applied. This code is a new path in an application that mimics existing DX9 code, and they share a common library to perform all vector and matrix (e.g. ViewToWorld/WorldToView) operations. My test bed consists of DX9 output images that I compare against the output of the new GDI+ code. A simple test case that renders to a viewport that matches the Bitmap dimensions (i.e. no zoom or pan) does match pixel-perfect (no binary diff) - but as soon as the image is zoomed up (magnified), I get very minor differences in 5-10% of the pixels. The magnitude of the difference is 1 (occasionally 2)/256. I suspect this is due to interpolation differences. Question: For a DX9 ortho projection (and identity world space), with a camera perpendicular and centered on a textured quad, is it reasonable to expect DirectX.Direct3D.TextureFilter.Linear to generate identical output to a GDI+ TextureBrush filled rectangle/polygon when using the System.Drawing.Drawing2D.InterpolationMode.Bilinear setting? For this (magnification) case, the DX9 code is using this (MinFilter,MipFilter set similarly): Device.SetSamplerState(0, SamplerStageStates.MagFilter, (int)TextureFilter.Linear); and the GDI+ path is using: g.InterpolationMode = InterpolationMode.Bilinear; I thought that "Bilinear Interpolation" was a fairly specific filter definition, but then I noticed that there is another option in GDI+ for "HighQualityBilinear" (which I've tried, with no difference - which makes sense given the description of "added prefiltering for shrinking") Followup Question: Is it reasonable to expect pixel-perfect output matching between DirectX and GDI+ (assuming all external coordinates passed in are equal)? If not, why not? Finally, there are a number of other APIs I could be using (Direct2D, WPF, GDI, etc.) - and this question generally applies to comparing the output of "equivalent" bilinear interpolated output images across any two of these. Thanks!

    Read the article

  • Tips on designing a .NET API for future use with F#

    - by Drew Noakes
    I'm in the process of designing a .NET API to allow developers to create RoboCup agents for the 3D simulated soccer league. I'm pretty happy with how the API work with C# code, however I would like to use this project to improve my F# skill (which is currently based on reading rather than practice). So I would like to ask what kinds of things I should consider when designing an API that is to be consumed by both C# and F# code. Some points. I make fairly heavy use of matrix and vector math. These are currently immutable classes/structs. The API currently defines a few interfaces with the consumer implements (eg: IAgent), using instances of their implementations (eg: MyAgent) to construct other API classes (eg: new Client(myAgent)). The API fires events. The API exposes a few delegate types. The API includes several enums. I'd like to release a version of the API as soon as possible and don't want to make major changes to it later if I realise it's too difficult to work with from F#. Any advice is appreciated.

    Read the article

  • Performance: float to int cast and clipping result to range

    - by durandai
    I'm doing some audio processing with float. The result needs to be converted back to PCM samples, and I noticed that the cast from float to int is surprisingly expensive. Whats furthermore frustrating that I need to clip the result to the range of a short (-32768 to 32767). While I would normally instictively assume that this could be assured by simply casting float to short, this fails miserably in Java, since on the bytecode level it results in F2I followed by I2S. So instead of a simple: int sample = (short) flotVal; I needed to resort to this ugly sequence: int sample = (int) floatVal; if (sample > 32767) { sample = 32767; } else if (sample < -32768) { sample = -32768; } Is there a faster way to do this? (about ~6% of the total runtime seems to be spent on casting, while 6% seem to be not that much at first glance, its astounding when I consider that the processing part involves a good chunk of matrix multiplications and IDCT) EDIT The cast/clipping code above is (not surprisingly) in the body of a loop that reads float values from a float[] and puts them into a byte[]. I have a test suite that measures total runtime on several test cases (processing about 200MB of raw audio data). The 6% were concluded from the runtime difference when the cast assignment "int sample = (int) floatVal" was replaced by assigning the loop index to sample. EDIT @leopoldkot: I'm aware of the truncation in Java, as stated in the original question (F2I, I2S bytecode sequence). I only tried the cast to short because I assumed that Java had an F2S bytecode, which it unfortunately does not (comming originally from an 68K assembly background, where a simple "fmove.w FP0, D0" would have done exactly what I wanted).

    Read the article

  • Introduction to vectorizing in MATLAB - any good tutorials?

    - by Gacek
    I'm looking for any good tutorials on vectorizing (loops) in MATLAB. I have quite simple algorithm, but it uses two for loops. I know that it should be simple to vectorize it and I would like to learn how to do it instead of asking you for the solution. But to let you know what problem I have, so you would be able to suggest best tutorials that are showing how to solve similar problems, here's the outline of my problem: B = zeros(size(A)); % //A is a given matrix. for i=1:size(A,1) for j=1:size(A,2) H = ... %// take some surrounding elements of the element at position (i,j) (i.e. using mask 3x3 elements) B(i,j) = computeSth(H); %// compute something on selected elements and place it in B end end So, I'm NOT asking for the solution. I'm asking for a good tutorials, examples of vectorizing loops in MATLAB. I would like to learn how to do it and do it on my own.

    Read the article

  • Who to access a UIButton using tag and button change image?

    - by ste
    I need to change the images on a matrix of UIButtons, and the only thing I know of to address the buttons, is the tag. But I can not find a way to actually us this identifier. The buttons are created programmatically during viewDidLoad. Here is the code for creating the buttons: #define N_ROWS 4 #define N_COLS 3 int N_IMG = 0; for (int a = 0; a < N_COLS; a++) { for (int j = 0; j < N_ROWS; j++) { UIButton *aButton = [UIButton buttonWithType:UIButtonTypeCustom]; aButton.frame = CGRectMake(a * 65.0 + 25, j * 65.0 + 15, 10.0, 10.0); aButton.tag = j + a * N_ROWS + 1; [aButton setBackgroundColor:[UIColor redColor]]; N_IMG = N_IMG++; [self.view addSubview:aButton]; number_sorted = 1; } } Here is the code for setting the image: - (IBAction)set_image:(id)sender { #define N_ROWS 4 #define N_COLS 3 int N_IMG = 0; for (int a = 0; a < N_COLS; a++) { for (int j = 0; j < N_ROWS; j++) { uibutton aButton.tag == (j + a * N_ROWS + 1) setImage:[UIImage imageNamed:[puzzles objectAtIndex:N_IMG]] forState:UIControlStateNormal]; N_IMG = N_IMG++; } } } This is the code where the truble starts: uibutton aButton.tag == (j + a * N_ROWS + 1) Who can I set this up to work?

    Read the article

  • Syntactical analysis with Flex/Bison part 2

    - by Imran
    Hallo, I need help in Lex/Yacc Programming. I wrote a compiler for a syntactical analysis for inputs of many statements. Now i have a special problem. In case of an Input the compiler gives the right output, which statement is uses, constant operator or a jmp instructor to which label, now i have to write so, if now a if statement comes, first the first command (before the else) must be give out when the assignment of the if is yes then it must jump to the end because the command after the else isnt needed, so after this jmp then the second command must be give out. I show it in an example maybe you understand what i mean. Input adr. Output if(x==0) 10 if(x==0) Wait 5 20 WAIT 5 else 30 JMP 50 Wait 1 40 WAIT 1 end 50 END like so. I have an idea, maybe i can do it whith a special if statement like IF exp jmp_stmt_end stmt_seq END when the if statement is given in the input the compiler has to recognize the end ofthe statement and like my jmp_stmt in my compiler ( you have to download the files from http://bitbucket.org/matrix/changed-tiny) only to jump to the end. I hope you understand my problem.thanks.

    Read the article

  • Problem using the find function in MATLAB

    - by Peter Etchells
    I have two arrays of data that I'm trying to amalgamate. One contains actual latencies from an experiment in the first column (e.g. 0.345, 0.455... never more than 3 decimal places), along with other data from that experiment. The other contains what is effectively a 'look up' list of latencies ranging from 0.001 to 0.500 in 0.001 increments, along with other pieces of data. Both data sets are X-by-Y doubles. What I'm trying to do is something like... for i = 1:length(actual_latency) row = find(predicted_data(:,1) == actual_latency(i)) full_set(i,1:4) = [actual_latency(i) other_info(i) predicted_info(row,2) ... predicted_info(row,3)]; end ...in order to find the relevant row in predicted_data where the look up latency corresponds to the actual latency. I then use this to created an amalgamated data set, full_set. I figured this would be really simple, but the find function keeps failing by throwing up an empty matrix when looking for an actual latency that I know is in predicted_data(:,1) (as I've double-checked during debugging). Moreover, if I replace find with a for loop to do the same job, I get a similar error. It doesn't appear to be systematic - using different participant data sets throws it up in different places. Furthermore, during debugging mode, if I use find to try and find a hard-coded value of actual_latency, it doesn't always work. Sometimes yes, sometimes no. I'm really scratching my head over this, so if anyone has any ideas about what might be going on, I'd be really grateful.

    Read the article

  • Python: Serial Transmission

    - by Silent Elektron
    I have an image stack of 500 images (jpeg) of 640x480. I intend to make 500 pixels (1st pixels of all images) as a list and then send that via COM1 to FPGA where I do my further processing. I have a couple of questions here: How do I import all the 500 images at a time into python and how do i store it? How do I send the 500 pixel list via COM1 to FPGA? I tried the following: Converted the jpeg image to intensity values (each pixel is denoted by a number between 0 and 255) in MATLAB, saved the intensity values in a text file, read that file using readlines(). But it became too cumbersome to make the intensity value files for all the 500 images! Used NumPy to put the read files in a matrix and then pick the first pixel of all images. But when I send it, its coming like: [56, 61, 78, ... ,71, 91]. Is there a way to eliminate the [ ] and , while sending the data serially? Thanks in Advance! :)

    Read the article

  • Patterns / Solutions to complicated Feature Management

    - by yclian
    Hi all, My company develops CDN / Web-Hosting solution. We have a middleware that's served as a business logic layer and exposes web service for the front-end. I would like to seek for a clean solution to feature management - there're uncertainties and ugly workarounds/solutions in the software that the dev would say "when it happens or is broken, we will fix it". For example, here're the following features that a web publisher can have: Sites limit Bandwidth limit SSL feature + SSL configuration per site If we downgrade a web publisher, when he's having 10 sites, down to 5 sites, we can choose not to suspend the rest of the 5 sites, or we shall prompt for suspension before the downgrade. For the case of bandwidth limit, the downgrade is easy, when the bandwidth check happens, if the publisher has it exceeded, then we will suspend his account. For the case of SSL feature. Every SSL configuration is tied to a site, what shall happen to these configuration object when the SSL feature is downgraded from enabled to disabled? So as you can see, there're many different situations and there are different ways of handling it. I can make a system that examines the impacts and prompts the user to make changes before the downgrade/upgrade. Or a system that ignores the impacts and just upgrade/downgrade. Bad. Or a system designed in a way that the client code need to be aware of the complex feature matrix (or I can expose a helper to the client code to check if a feature is not DEFUNCT) There can be many ways that I am still thinking but puzzled. I am wondering, how would you tackle this issue and is there any recommended patterns or books or software that you think I can refer to? Appreciate your help.

    Read the article

  • OpenGL circle rotation

    - by user350632
    I'm using following code to draw my circles: double theta = 2 * 3.1415926 / num_segments; double c = Math.Cos(theta);//precalculate the sine and cosine double s = Math.Sin(theta); double t; double x = r;//we start at angle = 0 double y = 0; GL.glBegin(GL.GL_LINE_LOOP); for(int ii = 0; ii < num_segments; ii++) { float first = (float)(x * scaleX + cx) / xyFactor; float second = (float)(y * scaleY + cy) / xyFactor; GL.glVertex2f(first, second); // output Vertex //apply the rotation matrix t = x; x = c * x - s * y; y = s * t + c * y; } GL.glEnd(); The problem is that when scaleX is different from scaleY then circles are transformed in the right way except for the rotation. In my code sequence looks like this: circle.Scale(tmp_p.scaleX, tmp_p.scaleY); circle.Rotate(tmp_p.rotateAngle); My question is what other calculations should i perform for circle to rotate properly when scaleX and scaleY are not equal?

    Read the article

  • C++, name collision across different namespace

    - by aaa
    hello. I am baffled by the following name collision: namespace mp2 { boost::numeric::ublas::matrix_range<M> slice(M& m, const R1& r1, const R2& r2) { namespace ublas = boost::numeric::ublas; ublas::range r1_(r1.begin(), r1.end()), r2_(r2.begin(), r2.end()); return ublas::matrix_range<M>(m, r1_, r2_); } double energy(const Wavefunction &wf) { const Wavefunction::matrix& C = wf.coefficients(); int No = wf.occupied().size(); foreach (const Basis::MappedShell& P, basis.shells()) { slice(C, range(No), range(P)); the error from g++4.4 is 7 In file included from mp2.cpp:1: 8 /usr/include/boost/numeric/ublas/fwd.hpp: In function âdouble mp2::energy(const Wavefunction&)â: 9 /usr/include/boost/numeric/ublas/fwd.hpp:32: error: âboost::numeric::ublas::sliceâ is not a function, 10 ../../src/mp2/energy.hpp:98: error: conflict with âtemplate<class M, class R1, class R2> boost::numeric::ublas::matrix_range<M> mp2::slice(M&, const R1&, const R2&)â 11 ../../src/mp2/energy.hpp:123: error: in call to âsliceâ 12 /usr/include/boost/numeric/ublas/fwd.hpp:32: error: âboost::numeric::ublas::sliceâ is not a function, 13 ../../src/mp2/energy.hpp:98: error: conflict with âtemplate<class M, class R1, class R2> boost::numeric::ublas::matrix_range<M> mp2::slice(M&, const R1&, const R2&)â 14 ../../src/mp2/energy.hpp:129: error: in call to âsliceâ 15 make: *** [mp2.lo] Error 1 ublas segment is namespace boost { namespace numeric { namespace ublas { typedef basic_slice<> slice; why is slice in ublas collides with slice in mp2? I and fairly certain there is no using namespace ublas in the code and in includes. thank you

    Read the article

  • SQL query for an access database needed

    - by masfenix
    Hey guys, first off all sorry, i can't login using my yahoo provider. anyways I have this problem. Let me explain it to you, and then I'll show you a picture. I have a access db table. It has 'report id', 'recpient id', and 'recipient name' and 'report req'. What the table "means" is that do the user using that report still require it or can we decommission it. Here is how the data looks like (blocked out company userids and usernames): *check the link below, I cant post pictures cuz yahoo open id provider isnt working. So basically I need to have 3 select queries: 1) Select all the reports where for each report, ALL the users have said no to 'reportreq'. In plain English, i want a listing of all the reports that we have to decommission because no user wants it. 2) Select all the reports where the report is required, and the batchprintcopy is more then 0. This way we can see which report needs to be printed and save paper instead of printing all the reports. 3)A listing of all the reports where the reportreq field is empty. I think i can figure this one out myself. This is using Access/VBA and the data will be exported to an excel spreadsheet. I just a simple query if it exists, OR an alogorithm to do it quickly. I just tried making a "matrix" and it took about 2 hours to populate. https://docs.google.com/uc?id=0B2EMqbpeBpQkMTIyMzA5ZjMtMGQ3Zi00NzRmLWEyMDAtODcxYWM0ZTFmMDFk&hl=en_US

    Read the article

  • Increasing figure size in Matplotlib

    - by Anirudh
    I am trying to plot a graph from a distance matrix. The code words fine and gives me a image in 800 * 600 pixels. The image being too small, All the nodes are packed together. I want increase the size of the image. so I added the following line to my code - figure(num=None, figsize=(10, 10), dpi=80, facecolor='w', edgecolor='k') After this all I get is a blank 1000 * 1000 image file. My overall code - import networkx as nx import pickle import matplotlib.pyplot as plt print "Reading from pickle." p_file = open('pickles/names') Names = pickle.load(p_file) p_file.close() p_file = open('pickles/distance') Dist = pickle.load(p_file) p_file.close() G = nx.Graph() print "Inserting Nodes." for n in Names: G.add_node(n) print "Inserting Edges." for i in range(601): for j in range(601): G.add_edge(Names[i],Names[j],weight=Dist[i][j]) print "Drawing Graph." nx.draw(G) print "Saving Figure." #plt.figure(num=None, figsize=(10, 10)) plt.savefig('new.png') print "Success!"

    Read the article

  • Combining two data sets and plotting in matlab

    - by bautrey
    I am doing experiments with different operational amplifier circuits and I need to plot my measured results onto a graph. I have two data sets: freq1 = [.1 .2 .5 .7 1 3 4 6 10 20 35 45 60 75 90 100]; %kHz Vo1 = [1.2 1.6 1.2 2 2 2.4 14.8 20.4 26.4 30.4 53.6 68.8 90 114 140 152]; %mV V1 = 19.6; Acm = Vo1/(1000*V1); And: freq2 = [.1 .5 1 30 60 70 85 100]; %kHz Vo1 = [3.96 3.96 3.96 3.84 3.86 3.88 3.88 3.88]; %V V1 = .96; Ad = Vo1/(2*V1); (I would show my plots but apparently I need more reps for that) I need to plot the equation, CMRR vs freq: CMRR = 20*log10(abs(Ad/Acm)); The size of Ad and Acm are different and the frequency points do not match up, but the boundaries of both of these is the same, 100Hz to 100kHz (x-axis). On the line of CMRR, Matlab says that Ad and Acm matrix dimensions do not agree. How I think I would solve this is using freq1 as the x-axis for CMRR and then taking approximated points from Ad according to the value on freq1. Or I could do function approximations of Ad and Acm and then do the divide operator on those. I do not know how I would code up this two ideas. Any other ideas would helpful, especially simpler ones. Thanks

    Read the article

  • Import & modify date data in MATLAB

    - by niko
    I have a .csv file with records written in the following form: 2010-04-20 15:15:00,"8.9915176259e+00","8.8562623697e+00" 2010-04-20 15:30:00,"8.5718021723e+00","8.6633827160e+00" 2010-04-20 15:45:00,"8.4484844117e+00","8.4336586330e+00" 2010-04-20 16:00:00,"1.1106980342e+01","8.4333062208e+00" 2010-04-20 16:15:00,"9.0643470589e+00","8.6885660103e+00" 2010-04-20 16:30:00,"8.2133517943e+00","8.2677822671e+00" 2010-04-20 16:45:00,"8.2499419380e+00","8.1523501983e+00" 2010-04-20 17:00:00,"8.2948492278e+00","8.2884797924e+00" From these data I would like to make clusters - I would like to add a column with number indicating the hour - so in case of the first row a value 15 has to be added in a new row. The first problem is that calling a function [numData, textData, rawData] = xlsread('testData.csv') creates an empty matrix numData and one-column textData and rawData structures. Is it possible to create any template which recognizes a yyyy, MM, dd, hh, mm, ss values from the data above? What I would basically like to do with these data is to categorize the values by hours so from the example row of input: 2010-04-20 15:15:00,"8.9915176259e+00","8.8562623697e+00" update 1: in Matlab the line above is recognized as a string: '2010-04-26 13:00:00,"1.0428104753e+00","2.3456394130e+00"' I would want this to be the output: 15, 8.9915176259e+00, 8.8562623697e+00 update 1: a string has to be parsed Does anyone know how to parse a string and retrieve a timestamp, value1 (1.0428104753e+00) and value2 (2.3456394130e+00) from it as separate values?

    Read the article

  • View results of affine transform

    - by stckjp
    I am trying to find out the reason why when I apply affine transformations on an image in OpenCV, the result of it is not visible in the preview window, but the entire window is black.How can I find workaround for this problem so that I can always view my transformed image (the result of the affine transform) in the window no matter the applied transformation? Update: I think that this happens because all the transformations are calculated with respect to the origin of the coordinate system (top left corner of the image). While for rotation I can specify the center of the rotation, and I am able to view the result, when I perform scaling I am not able to control where the transformed image goes. Is it possible to somehow move the coordinate system to make the image fit in the window? Update2: I have an image which contains only ROI at some position in it (the rest of the image is black), and I need to apply a set of affine transforms on it. To make things simpler and to see the effect of each individual transform, I applied each transform one by one. What I noticed is that, whenever I move (translate) the image such that the center of the ROI is in the center of the coordinate system (top left corner of the view window), all the affine transforms perform correctly without moving. However, by translating the center of ROI at the center of the coordinate system, the upper and the left part of the ROI remain cut out of the current view window. If I move ROI's central point to another point in the view window (for example the window center), an affine transform of type: A=[a 0 0; 0 b 0] (A is 2x3 matrix, parameter of the warpAffine function) moves the image (ROI), outside of the view window (which doesn't happen if the ROI's center is in the top-left corner). How can I modify the affine transform so the image doesn't move out of its place (behaves the same way as when the ROI center is in the center of the coordinate system)?

    Read the article

  • Why do R objects not print in a function or a "for" loop?

    - by Sal Leggio
    I have an R matrix named ddd. When I enter this, everything works fine: i <- 1 shapiro.test(ddd[,y]) ad.test(ddd[,y]) stem(ddd[,y]) print(y) The calls to Shapiro Wilk, Anderson Darling, and stem all work, and extract the same column. If I put this code in a "for" loop, the calls to Shapiro Wilk, and Anderson Darling stop working, while the the stem & leaf call and the print call continue to work. for (y in 7:10) { shapiro.test(ddd[,y]) ad.test(ddd[,y]) stem(ddd[,y]) print(y) } The decimal point is 1 digit(s) to the right of the | 0 | 0 0 | 899999 1 | 0 [1] 7 The same thing happens if I try and write a function. SW & AD do not work. The other calls do. > D <- function (y) { + shapiro.test(ddd[,y]) + ad.test(ddd[,y]) + stem(ddd[,y]) + print(y) } > D(9) The decimal point is at the | 9 | 000 9 | 10 | 00000 [1] 9 Why don't all the calls behave the same way?

    Read the article

  • Finding What You Need in R: function arguments/parameters from outside the function's package

    - by doug
    Often in R, there are a dozen functions scattered across as many packages--all of which have the same purpose but of course differ in accuracy, performance, theoretical rigor, and so on. How do you gather all of these in one place before you start your task? So for instance: the generic plot function. Setting secondary ticks is much easier (IMHO) using a function outside of the base package, minor.tick(nx=n, ny=n, tick.ratio=n), found in Hmisc. Of course, that doesn't show up in plot's docstring. Likewise, the data-input arguments to 'plot' can be supplied by an object returned from the function 'hexbin', again, from a library outside of the base installation (where 'plot' resides). What would be great obviously is a programmatic way to gather these function arguments from the various libraries and put them in a single namespace. edit: (trying to re-state my example just above more clearly:) the arguments to plot supplied in the base package for, e.g., setting the axis tick frequency are xaxp/yaxp; however, one can also set a/t/f via a function outside of the base package, again, as in the minor.tick function from the Hmisc package--but you wouldn't know that just from looking at the plot method signature. Is there a meta function in R for this? So far, as i come across them, i've been manually gathering them in a TextMate 'snippet' (along with the attendant library imports). This isn't that difficult or time consuming, but i can only update my snippet as i find out about these additional arguments/parameters. Is there a canonical R way to do this, or at least an easier way? Just in case that wasn't clear, i am not talking about the case where multiple packages provide functions directed to the same statistic or view (e.g., 'boxplot' in the base package; 'boxplot.matrix' in gplots; and 'bplots' in Rlab). What i am talking is the case in which the function name is the same across two or more packages.

    Read the article

  • identifier ... is undefined when trying to run pure C code in Cuda using nvcc

    - by Lostsoul
    I'm new and learning Cuda. A approach that I'm trying to use to learn is to write code in C and once I know its working start converting it to Cuda since I read that nvcc compiles Cuda code but complies everything else using plain old c. My code works in c(using gcc) but when I try to compile it using nvcc(after changing the file name from main.c to main.cu) I get main.cu(155): error: identifier "num_of_rows" is undefined main.cu(155): error: identifier "num_items_in_row" is undefined 2 errors detected in the compilation of "/tmp/tmpxft_00002898_00000000-4_main.cpp1.ii". Basically in my main method I send data to a function like this: process_list(count, countListItem, list); the first two items are ints and the last item(list) is a matrix. Then I create my function like this: void process_list(int num_of_rows, int num_items_in_row, int current_list[num_of_rows][num_items_in_row]) { This line is where I get my errors when using nvcc(line 155). I need to convert this code to cuda anyway so no need to troubleshoot this specific issue(plus code is quite large) but I'm wondering if I'm wrong about nvcc treating the C part of your code like plain C. In the book cuda by example I just used nvcc to compile but do I need any extra flags when just using pure c?

    Read the article

  • Out-of-memory algorithms for addressing large arrays

    - by reve_etrange
    I am trying to deal with a very large dataset. I have k = ~4200 matrices (varying sizes) which must be compared combinatorially, skipping non-unique and self comparisons. Each of k(k-1)/2 comparisons produces a matrix, which must be indexed against its parents (i.e. can find out where it came from). The convenient way to do this is to (triangularly) fill a k-by-k cell array with the result of each comparison. These are ~100 X ~100 matrices, on average. Using single precision floats, it works out to 400 GB overall. I need to 1) generate the cell array or pieces of it without trying to place the whole thing in memory and 2) access its elements (and their elements) in like fashion. My attempts have been inefficient due to reliance on MATLAB's eval() as well as save and clear occurring in loops. for i=1:k [~,m] = size(data{i}); cur_var = ['H' int2str(i)]; %# if i == 1; save('FileName'); end; %# If using a single MAT file and need to create it. eval([cur_var ' = cell(1,k-i);']); for j=i+1:k [~,n] = size(data{j}); eval([cur_var '{i,j} = zeros(m,n,''single'');']); eval([cur_var '{i,j} = compare(data{i},data{j});']); end save(cur_var,cur_var); %# Add '-append' when using a single MAT file. clear(cur_var); end The other thing I have done is to perform the split when mod((i+j-1)/2,max(factor(k(k-1)/2))) == 0. This divides the result into the largest number of same-size pieces, which seems logical. The indexing is a little more complicated, but not too bad because a linear index could be used. Does anyone know/see a better way?

    Read the article

  • My kernel only works in block (0,0)

    - by ZeroDivide
    I am trying to write a simple matrixMultiplication application that multiplies two square matrices using CUDA. I am having a problem where my kernel is only computing correctly in block (0,0) of the grid. This is my invocation code: dim3 dimBlock(4,4,1); dim3 dimGrid(4,4,1); //Launch the kernel; MatrixMulKernel<<<dimGrid,dimBlock>>>(Md,Nd,Pd,Width); This is my Kernel function __global__ void MatrixMulKernel(int* Md, int* Nd, int* Pd, int Width) { const int tx = threadIdx.x; const int ty = threadIdx.y; const int bx = blockIdx.x; const int by = blockIdx.y; const int row = (by * blockDim.y + ty); const int col = (bx * blockDim.x + tx); //Pvalue stores the Pd element that is computed by the thread int Pvalue = 0; for (int k = 0; k < Width; k++) { Pvalue += Md[row * Width + k] * Nd[k * Width + col]; } __syncthreads(); //Write the matrix to device memory each thread writes one element Pd[row * Width + col] = Pvalue; } I think the problem may have something to do with memory but I'm a bit lost. What should I do to make this code work across several blocks?

    Read the article

  • Solving quadratic programming using R

    - by user702846
    I would like to solve the following quadratic programming equation using ipop function from kernlab : min 0.5*x'*H*x + f'*x subject to: A*x <= b Aeq*x = beq LB <= x <= UB where in our example H 3x3 matrix, f is 3x1, A is 2x3, b is 2x1, LB and UB are both 3x1. edit 1 My R code is : library(kernlab) H <- rbind(c(1,0,0),c(0,1,0),c(0,0,1)) f = rbind(0,0,0) A = rbind(c(1,1,1), c(-1,-1,-1)) b = rbind(4.26, -1.73) LB = rbind(0,0,0) UB = rbind(100,100,100) > ipop(f,H,A,b,LB,UB,0) Error in crossprod(r, q) : non-conformable arguments I know from matlab that is something like this : H = eye(3); f = [0,0,0]; nsamples=3; eps = (sqrt(nsamples)-1)/sqrt(nsamples); A=ones(1,nsamples); A(2,:)=-ones(1,nsamples); b=[nsamples*(eps+1); nsamples*(eps-1)]; Aeq = []; beq = []; LB = zeros(nsamples,1); UB = ones(nsamples,1).*1000; [beta,FVAL,EXITFLAG] = quadprog(H,f,A,b,Aeq,beq,LB,UB); and the answer is a vector of 3x1 equals to [0.57,0.57,0.57]; However when I try it on R, using ipop function from kernlab library ipop(f,H,A,b,LB,UB,0)) and I am facing Error in crossprod(r, q) : non-conformable arguments I appreciate any comment

    Read the article

  • Ruby CSV in Array

    - by mattrock
    I read a CSV file. This file looks like this 1.00 cm; 2.00cm ; 3.00 cm; ... ; 100 cm 2.00 cm; 4.00 cm; 6.00 cm; ... ; 100 cm 4.00 cm; 8.00 cm; 12.00 cm; ... ; 100cm 8.00 cm; 16.00 cm; 24.00 cm; ... ; 100cm I have already written the following code CSV.foreach("/Users/testUser/Entwicklung/coverrechner/CoverPaperDE.csv", col_sep: ';') do |row| puts row[0] end This produces the following output: 1.00 cm 2.00 cm 4.00 cm 8.00 cm Example: My matrix is constructed 1.1 1.2 1.3 1.4 2.1 2.2 2.3 2.4 3.1 3.2 3.3 3.4 4.1 4.2 4.3 4.4 I want the following output 1.1 2.1 3.1 4.1 1.2 2.2 3.2 4.2 ... 4.4 How does it work?

    Read the article

  • MATLAB: svds() not converging

    - by Paul
    So using MATLAB's svds() function on some input data as such: [U, S, V, flag] = svds(data, nSVDs, 'L') I noticed that from run to run with the same data, I'd get drastically different output SVD sizes from run to run. When I checked whether 'flag' was set, I found that it was, indicating that the SVDs had not converged. My normal system here would be that if it really needs to converge, I'd do something like this: flag = 1 svdOpts = struct('tol', 1e-10, 'maxit', 600, 'disp', 0); while flag: if svdOpts.maxit > 1e6 error('There''s a real problem here.') end [U, S, V, flag] = svds(data, nSVDs, 'L', svdOpts) svdOpts.maxit = svdOpts.maxit*2 end But from what I can tell, when you use 'L' as the third argument, the fourth argument is ignored, meaning I just have to deal with the fact that it's not converging? I'm not even really sure how to use the 'sigma' argument in place of the 'L' argument. I've also tried reducing the number of SVDs calculated to no avail. Any help on this matter would be much appreciated. EDIT While following up on the comments below, I found that the problem had to do with the way I was building my data matrices. Turned out I had accidentally inverted a matrix and had an input of size (4000x1) rather than (20x200), which was what was refusing to converge. I also did some more timing tets and found that the fourth argument was not, in fact, being ignored, so that's on me. Thanks for the help guys.

    Read the article

  • rbind.zoo doesn't seem create consistent zoo object

    - by a-or-b
    I want to rbind.zoo two zoo object together. When I was testing I came across the following issue(?)... Note: The below is an example, there is clearly no point to it apart from being illustrative. I have an zoo object, call it, 'X'. I want to break it into two parts and then rbind.zoo them together. When I compare it to the original object then all.equal gives differences. It appears that the '$class' attribute differs, but I can't see how or why. Is I make these xts objects then the all.equal works as expected. i.e. ..... X.date <- as.POSIXct(paste("2003-", rep(1:4, 4:1), "-", sample(1:28, 10, replace = TRUE), sep = "")) X <- zoo(matrix(rnorm(24), ncol = 2), X.date) a <- X[c(1:3), ] # first 3 elements b <- X[c(4:6), ] # second 3 elements c <- rbind.zoo(a, b) # rbind into an object of 6 elements d <- X[c(1:6), ] # all 6 elements all.equal(c, d) # are they equal? ~~~~ all.equal gives me the following difference: "Attributes: < Component 3: Attributes: < Length mismatch: comparison on first 1 components "

    Read the article

< Previous Page | 66 67 68 69 70 71 72 73 74 75 76 77  | Next Page >