r - Why a self-written Rcpp vectorized mathematical function is faster than its base counterpart? -

-

OK, I know the answer, but due to this being motivated, I want to know some good advice about the following : Why is RCPP exercise CA down? 15% faster (for long vectors) than the built-in exp () ? We all know that Rcpp is a wrapper for R / C API, so we should expect a little worse performance than the one. 

  Rcpp :: cppFunction ("NumericVector exp2" (NumericVector X) {NumericVector z = RCPP :: Clone (x); Int N = Z SEE (); for (intI = 0; I & lt; n; ++ i) Z [I] = XP (Z [I]); Return Z;} ") Library (" microbenchmark ") X & lt; - rcauchy (1000000) microbenchmark (equation (x), exp2 (x), unit = "relative") ## Unit: relative ## expr min LQ median uq max neval ## exp (x) 1,159893 1,154143 1 , 155856 1,154482 0,926272 100 100 ## exp2 (x) 1,00,000,000 1,00,000,000 1,00,000,000 1,00,000,000 1,00,000,000 100

base r NA is used to check further, we can not do this. Also keep in mind that loop is doing tricks like unrolling (as RCpp did in Chinese), we can do a bit better still.

Then I added 

  Rcpp :: cppFunction ("NumericVector expSugar (NumericVector X) {return exp (x);}")

And with that I get another advantage - towards the user with less code: 

  R & gt; Microbenchmark (consonant (x), exp2 (x), expsugar (x), unit = "relative") unit: relative expr min LQ means median uq max neval exp (x) 1,11190 1,11130 1,11718 1,10799 1,08938 1,02590 100 Exp2 (X) 1.08,184 1.08937 1.07289 1.07621 1.06382 1.00462 100 Express (X) 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000 100 R & gt;

Comments

Post a Comment

Popular posts from this blog

c# - SignalR: "Protocol error: Unknown transport." when navigating to hub -

-
Image
I am trying to use the ASP.NET signal for the first time - here's what I did: created a new web Has been added to the project and the following signal R packages: I have created the following category: [Hub-name ("chat")] Public class ChatHub: Hub {Public Zero Halo () {Clients.All.hello ( ); }} and the following startup configuration: public partial class startup {public zero configuration (IAppBuilder app) {configure (app); App.MapSignalR (); }} However, whenever I navigate localhost: (port) / api / signalr / chat I get the following error: Protocol error: Unknown transport. I am running a new browser and trying different people so remove it from the API / URL, and just use the localhost: (potter) / signaller / Chat
Read more

class - Kivy: how to instantiate a dynamic classes in python -

-
Using KV lang on my Python code, I have taken a hard time trying to instantiate a dynamic class, the following Consider the code: My test.kv file looks like this: & lt; MyPopup @ popup: Title: 'Hello' size_hunt: (1, .6) Grid layout: ID: RouteGrid column: 2 padding: ['8dp', '4dp', '8dp', '4dp'] Difference: '8dp' label: text: 'some text here' button: text: 'OK' On_press: do_something () & lt; MyGrid & gt;: Rows: 1 Button: Text: 'Show Popup' on_press: root.pop.show () Then import Kivy.app into my test.py: I get the following error: raise Factory Exception ('Uncount Class' '% name) kivy.factory.FactoryException: Unknown class Can anyone tell me how to do this properly, what do I remember? If you need more information please let me know Thank you. Takes a call to the factory before the KV file is loaded, so you want the class to Are not yet in existence. As long as no clas...
Read more

python - mayavi mapping a discrete colorbar on a surface -

-
Image
I want to change the color of the surface on the basis of another value (an ID) that is considered discrete (hence I In the simplified example given below, I have prepared a field with 3 different IDs: 0 / red on left 2 / blue 1 / green left But with the code below, I have some strange behavior on the border between red and blue (green Point) I think this is probably due to an interpolation! Code: import from import mlab np # np # my dataset -simplified-x, y, z = np.mgrid [-3: 3: 100] H, -3: 3: 100h, -3: 3: 100 h] value = NP. SQLT (x ** 2 + y ** 2 + z ** 2) # My color value: Volume is divided into 3 sub-versions taking colorvalues ​​with X = np.empty (values.shape) colorvalues [0:33,:,:] = 0 colorvalues ​​[33:66,:,:] = 2. Look [66:,:,:] = 1 src = mlab.pipeline.scalar_field (value) src.image_data.point_data.add_array (colorvalues.T.ravel ()) (1) .name = 'MyID' src.image_data. point_data.update src.image_data.point_data.get_array () # surface, contour = mlab.pipe...
Read more