Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Dayanidhi Krishna
    @dkrish24
    @cristobalmedinalopez I am so sorry about the delay. I am almost at the terminal stage of my bachelor’s capstone project that I wasn’t able to devote time. I’ll get this done shortly. Sorry again
    Akashdeep Ramnaney
    @flaredragon
    @cristobalmedinalopez Sorry Sir to keep you waiting.I had my examination going on so wasn't able to work on the idea before.I'm working on the implementation of the project and proposal and get back with it soon.Any tips for the same will be appreciated☺
    Davi Laerte
    @davilaerte
    Hi, I'm Davi Laerte, currently studying Computer Science, I also found this project very interesting. I am familiar with static and data analysis, I have experience with pandas for data analysis in python and some data visualization techniques. I wonder if it would be interesting for this project to use a bit of data analysis for suggestions or predictions about the data collected which could be shown in the application ?
    Cristóbal Medina López
    @cristobalmedinalopez
    Hi @davilaerte Welcome to the P2PSP project!
    Regarding your question, of course, it would be very interesting. It could help us to predict a team behaviour in N more rounds. Are you meaning that?
    Davi Laerte
    @davilaerte
    Yes @cristobalmedinalopez , I think it would be interesting for the application
    siremey
    @siremey
    Hello, I am Emeka Orji, an MSc GIS student at the University of Ibadan.
    I have been working with Python packages for data analysis and visualization almost a year now. I also have some basic knowledge of JavaScript. I find this project very interesting and would like to contribute towards the GSoc 2018.
    Plotly or Bokeh server seems quite good for streaming and real time data visualization
    I like to be guided on where I can get started contributing to the project with the community
    Thank you
    Vicente González Ruiz
    @vicente-gonzalez-ruiz
    Hello @siremey, and welcome to the P2PSP.org project! At this moment, we prefer to improve the simulator (https://github.com/P2PSP/simulator) as much as we can, and of course, in this case, you could try to implement the statistic-module on it. The alternative is to work on the C++ implementation (https://github.com/P2PSP/console, https://github.com/P2PSP/core). However, as I has mentioned, we prefer to develop the simulator. Anyway, working with both implementations would be OK. So, if I were you, my first step would be to install the simulator and try to run it. An thank you for your interest!
    siremey
    @siremey
    Hi @vicente-gonzalez-ruiz, Implementing the statistical-module for (https://github.com/P2PSP/simulator) is great, as I've limited C++ knowledge. Will jump right in. Thank you
    Vicente González Ruiz
    @vicente-gonzalez-ruiz
    :+1:
    Sachin
    @sachinsngh165
    Hey guys! I have been working on the graph plotting method since past week. Since matplolib is quite slow and sluggish I was trying to move the plotting task to pyqtgraph library. As of now I have implement team status and network overlay plotting methods in pyqtgraph completely. And results are astonishing with great fps and low delay.
    This https://github.com/sachinsngh165/PyQtGraph-Experiments is my github repositry where I have implemented both of the methods.
    Please run this on your machine and review the results.
    • I've yet to implement the Buffer Status plotting.
    Cristóbal Medina López
    @cristobalmedinalopez
    Great! Let us take a look...
    Cristóbal Medina López
    @cristobalmedinalopez
    It looks really great! The networks overlay plotting methods is the one that more resources consumes in the matplotlib version. In this pyqtgraph version the improvement of the performance is enourmous. If after adding the buffer plotting the performance is similar, I think that it would be a perfect solution allowing us to check the behaviour of the simulation in a record time :) In fact, we could add playback controls as a new feature (pause, stop, rewind, fast forward). Thanks a lot!!
    BTW, I assume that it is running in the same thread. Is it not?
    Vicente González Ruiz
    @vicente-gonzalez-ruiz
    Amazing work, @sachinsngh165! Congratulations.
    Sachin
    @sachinsngh165
    thank you guys!
    @cristobalmedinalopez , I've yet to integrate that plotting method with main simulator program for simulation in real-time. I hope, I will do that real soon.
    Cristóbal Medina López
    @cristobalmedinalopez
    @sachinsngh165 Since the drawing log is stored in a file, I think that it could be as easy as run the draw() method at the same time that we are running the run() method. We tried it in the past but matplotlib was too slow. What do you think?
    Sachin
    @sachinsngh165
    Yes, It is working great if we run draw() and run() simultaneously as a different program. I have tested it.
    Sachin
    @sachinsngh165
    Hey, As I said, I have add buffer plotting method to Our Play Class. Code is on the same github repo. Please have a look.
    Vicente González Ruiz
    @vicente-gonzalez-ruiz
    Hi @sachinsngh165. I think that it is ideal. Good job. A question, how do you control the speed of the plot?
    Sachin
    @sachinsngh165
    Thanks, @vicente-gonzalez-ruiz to review my work.
    Currently, I don't know how to control plotting speed in pyqtgraph. But I do that by putting some delay function in our plotting methods.
    Cristóbal Medina López
    @cristobalmedinalopez
    What do you think about painting each thing from a different process?
    Now I can feel the plotting process slower. Probably it is due to the buffer.
    Sachin
    @sachinsngh165
    Yup
    There is too much buffer processing
    Cristóbal Medina López
    @cristobalmedinalopez
    Yes, any idea to optimize it?
    Sachin
    @sachinsngh165
    We can use pyqtgraph multiprocessing library
    Cristóbal Medina López
    @cristobalmedinalopez
    it sounds good :)
    Sachin
    @sachinsngh165
    In matplotlib implementation we were facing issues because, matplotlib were plotting new data points without clearing the previous one. So number of plot objects were increasing exponentially and matplotlib have to manage each and every plot objects created so far.
    You must have seen that initially, when plotting processes starts it works great but after few seconds it's performance start degrading and finally it get stuck ( at least on my mac ).
    Cristóbal Medina López
    @cristobalmedinalopez
    Yes, that is.
    Sachin
    @sachinsngh165
    And it is happening due to buffer plot because, it have so many data points to plot in each epoch . If you see the text file created.
    Cristóbal Medina López
    @cristobalmedinalopez
    Yes, you are right. We have to plot the status of the buffer for all peer in the team, and for each round.
    Maybe we could rethink about that.
    Sachin
    @sachinsngh165
    I've tried to overcome these issues with pyqtgraph by clearing the previous plotted points. And minimise points overlapping. Have you seen some difference ?
    Cristóbal Medina López
    @cristobalmedinalopez
    If you mean about a difference between the current version using matplotlib and your proposal using pyqtgraph. Yes, I saw a difference. I think that the performance with pyqtgraph is better.
    But I'm thinking about improving it more (if possible).
    Sachin
    @sachinsngh165
    There is still room for improvements in current implementation. Like using pyqtgraph multiprocessing and RemoteGraphicsView etc.
    Cristóbal Medina López
    @cristobalmedinalopez
    Ok, it could be a good option.
    In any case, I think that you have been a great jobs with pyqtgraph!!
    Sachin
    @sachinsngh165
    Thanks :smile:
    Cristóbal Medina López
    @cristobalmedinalopez
    Matplotlib has a problem with real time plotting...
    Sachin
    @sachinsngh165
    Yes, It have .
    I'll try to make further improvements whenever I'll get time.
    Cristóbal Medina López
    @cristobalmedinalopez
    It would be great. Thanks!
    I will try to take a look too
    Vicente González Ruiz
    @vicente-gonzalez-ruiz
    ?
    golokeshpatra45
    @golokeshpatra45
    @vicente-gonzalez-ruiz , if you go to the above stackoverflow, I have used Standard Deviation to solve my problem. But i dont know if there is a better/simpler way to perform the judgment of whether or not to terminate/restart an analytics task/job.
    So, I have explained every single detail in the link , please go through once.
    Thanks in advance