Optimizing an n-body simulation using Numpy and Glumpy












0














I am making the process of moving from MatLab to Python, and am attempting a real time N-body simulation as a way of getting to grips with the language. I have completed a script that uses glumpy for the visualisation, and vectorised numpy code for the numerical integration, but it slows waaaaaay down for more than about 150 points. Essentially, I am looking for tips on how to optimize the code so that it can run in real time with at least an order of magnitude more points. Particularly, is there any way of implementing the integration within the shader code using the GPU??



I am aware of the limitations with the Euler method that I have used (and the scaling of the acceleration), and am looking for input on optimization of the software implementation, as opposed to the numerical method itself.



import numpy as np
from glumpy import app, gloo, gl

#VERTEX SHADER CODE
vertex = """
attribute vec2 position;
void main() {
gl_PointSize = 5.0;
gl_Position = vec4(position, 0.0, 1.0);

} """

#FRAGMENT SHADER CODE
fragment = """
void main() {
gl_FragColor = vec4(vec3(0.5), 1.0);
} """

#CREATE ARRAY OF POSITIONS
n_points = int(300) #number of points
p = np.random.uniform(-1,1,(n_points,2)) #array of positions (x,y)
v = np.zeros((n_points,2)) #array of velocities (xdot,ydot)
m = np.ones((n_points,1)) #array of masses
a = np.zeros((n_points,2)) #array of accelerations
index = np.linspace(0,n_points-1,n_points)

#OPEN WINDOW AND CREATE SHADER PROGRAM
window = app.Window(512,512, color = (1,1,1,1))
points = gloo.Program(vertex,fragment,count = n_points)
points["position"] = p

@window.event
def on_draw(dt):
global v, p, a, m, theta

for i in range(n_points): #for each point
delta_x = p[index != i, 0] - p[i, 0] #distance in x to each other point
delta_y = p[index != i, 1] - p[i, 1] #distance in y to each other point
mass = m[index != i,:] #mass of other points
theta = np.arctan2(delta_y, delta_x) #angle to other points
r = (delta_x**2 + delta_y**2)*0.5 + 0.01 #distance to other points
a[i,0] = np.sum(mass * np.cos(theta) / r**2) #acceleration in x
a[i,1] = np.sum(mass * np.sin(theta) / r**2) #acceleration in y

print(dt)
a = a*0.00000001
v = v + a*dt
p = p + v*dt

window.clear()
points["position"] = p
points.draw(gl.GL_POINTS)

app.run()









share|improve this question



























    0














    I am making the process of moving from MatLab to Python, and am attempting a real time N-body simulation as a way of getting to grips with the language. I have completed a script that uses glumpy for the visualisation, and vectorised numpy code for the numerical integration, but it slows waaaaaay down for more than about 150 points. Essentially, I am looking for tips on how to optimize the code so that it can run in real time with at least an order of magnitude more points. Particularly, is there any way of implementing the integration within the shader code using the GPU??



    I am aware of the limitations with the Euler method that I have used (and the scaling of the acceleration), and am looking for input on optimization of the software implementation, as opposed to the numerical method itself.



    import numpy as np
    from glumpy import app, gloo, gl

    #VERTEX SHADER CODE
    vertex = """
    attribute vec2 position;
    void main() {
    gl_PointSize = 5.0;
    gl_Position = vec4(position, 0.0, 1.0);

    } """

    #FRAGMENT SHADER CODE
    fragment = """
    void main() {
    gl_FragColor = vec4(vec3(0.5), 1.0);
    } """

    #CREATE ARRAY OF POSITIONS
    n_points = int(300) #number of points
    p = np.random.uniform(-1,1,(n_points,2)) #array of positions (x,y)
    v = np.zeros((n_points,2)) #array of velocities (xdot,ydot)
    m = np.ones((n_points,1)) #array of masses
    a = np.zeros((n_points,2)) #array of accelerations
    index = np.linspace(0,n_points-1,n_points)

    #OPEN WINDOW AND CREATE SHADER PROGRAM
    window = app.Window(512,512, color = (1,1,1,1))
    points = gloo.Program(vertex,fragment,count = n_points)
    points["position"] = p

    @window.event
    def on_draw(dt):
    global v, p, a, m, theta

    for i in range(n_points): #for each point
    delta_x = p[index != i, 0] - p[i, 0] #distance in x to each other point
    delta_y = p[index != i, 1] - p[i, 1] #distance in y to each other point
    mass = m[index != i,:] #mass of other points
    theta = np.arctan2(delta_y, delta_x) #angle to other points
    r = (delta_x**2 + delta_y**2)*0.5 + 0.01 #distance to other points
    a[i,0] = np.sum(mass * np.cos(theta) / r**2) #acceleration in x
    a[i,1] = np.sum(mass * np.sin(theta) / r**2) #acceleration in y

    print(dt)
    a = a*0.00000001
    v = v + a*dt
    p = p + v*dt

    window.clear()
    points["position"] = p
    points.draw(gl.GL_POINTS)

    app.run()









    share|improve this question

























      0












      0








      0







      I am making the process of moving from MatLab to Python, and am attempting a real time N-body simulation as a way of getting to grips with the language. I have completed a script that uses glumpy for the visualisation, and vectorised numpy code for the numerical integration, but it slows waaaaaay down for more than about 150 points. Essentially, I am looking for tips on how to optimize the code so that it can run in real time with at least an order of magnitude more points. Particularly, is there any way of implementing the integration within the shader code using the GPU??



      I am aware of the limitations with the Euler method that I have used (and the scaling of the acceleration), and am looking for input on optimization of the software implementation, as opposed to the numerical method itself.



      import numpy as np
      from glumpy import app, gloo, gl

      #VERTEX SHADER CODE
      vertex = """
      attribute vec2 position;
      void main() {
      gl_PointSize = 5.0;
      gl_Position = vec4(position, 0.0, 1.0);

      } """

      #FRAGMENT SHADER CODE
      fragment = """
      void main() {
      gl_FragColor = vec4(vec3(0.5), 1.0);
      } """

      #CREATE ARRAY OF POSITIONS
      n_points = int(300) #number of points
      p = np.random.uniform(-1,1,(n_points,2)) #array of positions (x,y)
      v = np.zeros((n_points,2)) #array of velocities (xdot,ydot)
      m = np.ones((n_points,1)) #array of masses
      a = np.zeros((n_points,2)) #array of accelerations
      index = np.linspace(0,n_points-1,n_points)

      #OPEN WINDOW AND CREATE SHADER PROGRAM
      window = app.Window(512,512, color = (1,1,1,1))
      points = gloo.Program(vertex,fragment,count = n_points)
      points["position"] = p

      @window.event
      def on_draw(dt):
      global v, p, a, m, theta

      for i in range(n_points): #for each point
      delta_x = p[index != i, 0] - p[i, 0] #distance in x to each other point
      delta_y = p[index != i, 1] - p[i, 1] #distance in y to each other point
      mass = m[index != i,:] #mass of other points
      theta = np.arctan2(delta_y, delta_x) #angle to other points
      r = (delta_x**2 + delta_y**2)*0.5 + 0.01 #distance to other points
      a[i,0] = np.sum(mass * np.cos(theta) / r**2) #acceleration in x
      a[i,1] = np.sum(mass * np.sin(theta) / r**2) #acceleration in y

      print(dt)
      a = a*0.00000001
      v = v + a*dt
      p = p + v*dt

      window.clear()
      points["position"] = p
      points.draw(gl.GL_POINTS)

      app.run()









      share|improve this question













      I am making the process of moving from MatLab to Python, and am attempting a real time N-body simulation as a way of getting to grips with the language. I have completed a script that uses glumpy for the visualisation, and vectorised numpy code for the numerical integration, but it slows waaaaaay down for more than about 150 points. Essentially, I am looking for tips on how to optimize the code so that it can run in real time with at least an order of magnitude more points. Particularly, is there any way of implementing the integration within the shader code using the GPU??



      I am aware of the limitations with the Euler method that I have used (and the scaling of the acceleration), and am looking for input on optimization of the software implementation, as opposed to the numerical method itself.



      import numpy as np
      from glumpy import app, gloo, gl

      #VERTEX SHADER CODE
      vertex = """
      attribute vec2 position;
      void main() {
      gl_PointSize = 5.0;
      gl_Position = vec4(position, 0.0, 1.0);

      } """

      #FRAGMENT SHADER CODE
      fragment = """
      void main() {
      gl_FragColor = vec4(vec3(0.5), 1.0);
      } """

      #CREATE ARRAY OF POSITIONS
      n_points = int(300) #number of points
      p = np.random.uniform(-1,1,(n_points,2)) #array of positions (x,y)
      v = np.zeros((n_points,2)) #array of velocities (xdot,ydot)
      m = np.ones((n_points,1)) #array of masses
      a = np.zeros((n_points,2)) #array of accelerations
      index = np.linspace(0,n_points-1,n_points)

      #OPEN WINDOW AND CREATE SHADER PROGRAM
      window = app.Window(512,512, color = (1,1,1,1))
      points = gloo.Program(vertex,fragment,count = n_points)
      points["position"] = p

      @window.event
      def on_draw(dt):
      global v, p, a, m, theta

      for i in range(n_points): #for each point
      delta_x = p[index != i, 0] - p[i, 0] #distance in x to each other point
      delta_y = p[index != i, 1] - p[i, 1] #distance in y to each other point
      mass = m[index != i,:] #mass of other points
      theta = np.arctan2(delta_y, delta_x) #angle to other points
      r = (delta_x**2 + delta_y**2)*0.5 + 0.01 #distance to other points
      a[i,0] = np.sum(mass * np.cos(theta) / r**2) #acceleration in x
      a[i,1] = np.sum(mass * np.sin(theta) / r**2) #acceleration in y

      print(dt)
      a = a*0.00000001
      v = v + a*dt
      p = p + v*dt

      window.clear()
      points["position"] = p
      points.draw(gl.GL_POINTS)

      app.run()






      python numpy glumpy






      share|improve this question













      share|improve this question











      share|improve this question




      share|improve this question










      asked Nov 20 at 23:47









      SeBeast

      104




      104





























          active

          oldest

          votes











          Your Answer






          StackExchange.ifUsing("editor", function () {
          StackExchange.using("externalEditor", function () {
          StackExchange.using("snippets", function () {
          StackExchange.snippets.init();
          });
          });
          }, "code-snippets");

          StackExchange.ready(function() {
          var channelOptions = {
          tags: "".split(" "),
          id: "1"
          };
          initTagRenderer("".split(" "), "".split(" "), channelOptions);

          StackExchange.using("externalEditor", function() {
          // Have to fire editor after snippets, if snippets enabled
          if (StackExchange.settings.snippets.snippetsEnabled) {
          StackExchange.using("snippets", function() {
          createEditor();
          });
          }
          else {
          createEditor();
          }
          });

          function createEditor() {
          StackExchange.prepareEditor({
          heartbeatType: 'answer',
          autoActivateHeartbeat: false,
          convertImagesToLinks: true,
          noModals: true,
          showLowRepImageUploadWarning: true,
          reputationToPostImages: 10,
          bindNavPrevention: true,
          postfix: "",
          imageUploader: {
          brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
          contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
          allowUrls: true
          },
          onDemand: true,
          discardSelector: ".discard-answer"
          ,immediatelyShowMarkdownHelp:true
          });


          }
          });














          draft saved

          draft discarded


















          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53403315%2foptimizing-an-n-body-simulation-using-numpy-and-glumpy%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown






























          active

          oldest

          votes













          active

          oldest

          votes









          active

          oldest

          votes






          active

          oldest

          votes
















          draft saved

          draft discarded




















































          Thanks for contributing an answer to Stack Overflow!


          • Please be sure to answer the question. Provide details and share your research!

          But avoid



          • Asking for help, clarification, or responding to other answers.

          • Making statements based on opinion; back them up with references or personal experience.


          To learn more, see our tips on writing great answers.





          Some of your past answers have not been well-received, and you're in danger of being blocked from answering.


          Please pay close attention to the following guidance:


          • Please be sure to answer the question. Provide details and share your research!

          But avoid



          • Asking for help, clarification, or responding to other answers.

          • Making statements based on opinion; back them up with references or personal experience.


          To learn more, see our tips on writing great answers.




          draft saved


          draft discarded














          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53403315%2foptimizing-an-n-body-simulation-using-numpy-and-glumpy%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown





















































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown

































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown







          Popular posts from this blog

          Wiesbaden

          Marschland

          Dieringhausen