Re: [Matplotlib-users] Problem with plotting fast spiking data
Just to conclude the problem. I've just installed a fresh svn matplotlib (1.0-svn to be exact) and the problem with path optimisation disappeared, meaning with path.simplify False plots are correct as far as my data is concerned. BTW it very nicely builds out-of-the-box on 10.6 in 64-bit mode. Cheers, Jakub On 18 Feb 2010, at 22:37, Eric Firing wrote: Jakub Nowacki wrote: Hi, I work with neural models and I have problem with plotting fast spiking data. The spikes on the plot appear to have different hight which changes when I for example resize the plot window. The same problem is with saving data into files, especially in vector formats. I found the information about changing the join style, it helps a bit (rounded is the best) but doesn't solve the problem. For raster formats the workaround is to save the data in higher resolution, using DPI option. Below I included links to examples. This sounds like a problem with the path simplification algorithm. If so, the question is whether it is a bug that has been fixed or a new bug. What mpl version are you using? Can you build and install from svn? Eric Normal example (100 DPI): https://docs.google.com/leaf?id=0B3NZY3443E1VY2JjNTc3MjAtZDI5NC00OThjLTgwY2EtNTVhMDVkZWQ2YzIwhl=en Example 300 DPI: https://docs.google.com/leaf?id=0B3NZY3443E1VNDdkMzUzNzAtMjdmNC00NjFmLTliMzMtODE5MzExMmNjNjQzhl=en The problem is vector files (I'm especially interested in EPS) ignore DPI option. I've experimented with different backends and the problem is persistent on GUI and 'file-writting' back ends. I also tested it on Linux and Mac, and the outcome does't change. Creating larger figures sometimes helps a bit but not always solve the problem. Just to mention that plotting this kind of data is possible 'out of the box' in Matlab or XPPAut (which is not the most fancy plotting tool) I get a proper outcome. Maybe there should be a option to plot raw data in some sense, or join style function that deals with such a plots in a proper fashion. If you need some additional information do not hesitate to ask. Thanks for the help in advance. Cheers, Jakub -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Problem with plotting fast spiking data
Jakub Nowacki wrote: Hi, I work with neural models and I have problem with plotting fast spiking data. The spikes on the plot appear to have different hight which changes when I for example resize the plot window. The same problem is with saving data into files, especially in vector formats. I found the information about changing the join style, it helps a bit (rounded is the best) but doesn't solve the problem. For raster formats the workaround is to save the data in higher resolution, using DPI option. Below I included links to examples. This sounds like a problem with the path simplification algorithm. If so, the question is whether it is a bug that has been fixed or a new bug. What mpl version are you using? Can you build and install from svn? Eric Normal example (100 DPI): https://docs.google.com/leaf?id=0B3NZY3443E1VY2JjNTc3MjAtZDI5NC00OThjLTgwY2EtNTVhMDVkZWQ2YzIwhl=en Example 300 DPI: https://docs.google.com/leaf?id=0B3NZY3443E1VNDdkMzUzNzAtMjdmNC00NjFmLTliMzMtODE5MzExMmNjNjQzhl=en The problem is vector files (I'm especially interested in EPS) ignore DPI option. I've experimented with different backends and the problem is persistent on GUI and 'file-writting' back ends. I also tested it on Linux and Mac, and the outcome does't change. Creating larger figures sometimes helps a bit but not always solve the problem. Just to mention that plotting this kind of data is possible 'out of the box' in Matlab or XPPAut (which is not the most fancy plotting tool) I get a proper outcome. Maybe there should be a option to plot raw data in some sense, or join style function that deals with such a plots in a proper fashion. If you need some additional information do not hesitate to ask. Thanks for the help in advance. Cheers, Jakub -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Problem with plotting fast spiking data
On Thu, Feb 18, 2010 at 4:19 PM, Jakub Nowacki j.s.nowa...@googlemail.com wrote: Hi, I work with neural models and I have problem with plotting fast spiking data. The spikes on the plot appear to have different hight which changes when I for example resize the plot window. The same problem is with saving data into files, especially in vector formats. I found the information about changing the join style, it helps a bit (rounded is the best) but doesn't solve the problem. For raster formats the workaround is to save the data in higher resolution, using DPI option. Below I included links to examples. Normal example (100 DPI): https://docs.google.com/leaf?id=0B3NZY3443E1VY2JjNTc3MjAtZDI5NC00OThjLTgwY2EtNTVhMDVkZWQ2YzIwhl=en Example 300 DPI: https://docs.google.com/leaf?id=0B3NZY3443E1VNDdkMzUzNzAtMjdmNC00NjFmLTliMzMtODE5MzExMmNjNjQzhl=en The problem is vector files (I'm especially interested in EPS) ignore DPI option. I've experimented with different backends and the problem is persistent on GUI and 'file-writting' back ends. I also tested it on Linux and Mac, and the outcome does't change. Creating larger figures sometimes helps a bit but not always solve the problem. Just to mention that plotting this kind of data is possible 'out of the box' in Matlab or XPPAut (which is not the most fancy plotting tool) I get a proper outcome. Maybe there should be a option to plot raw data in some sense, or join style function that deals with such a plots in a proper fashion. If you need some additional information do not hesitate to ask. Thanks for the help in advance. This is a known bug in the latest release of mpl that is fixed in svn. We do need to get a new release out soon. It is easy to fix by uncomennting the path.simplify : False line in your matplotlibrc file (the default is True). Note that path.simplify does work in the svn release of matplotlib -- it is designed to reduce paths but be imperceptible to the eye at the resolution plotted, but due to a bug in the released version it can result in improper simplifications. See http://matplotlib.sourceforge.net/users/customizing.html for information on how to change your matplotlibrc settings. JDH -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Problem with plotting fast spiking data
Dear John and Eric, Thank you for the fast response. I'm using version 0.99.1.1 (to be exact), the MacOSX binary release for Python 2.5. The fix with path.simplify: False in matplotlibrc works perfectly! I googled quite a lot beforehand in order to find a fix for this issue, but obviously failed to find this bug. Once more thank you very much for the help. Cheers, Jakub On 18 Feb 2010, at 22:39, John Hunter wrote: On Thu, Feb 18, 2010 at 4:19 PM, Jakub Nowacki j.s.nowa...@googlemail.com wrote: Hi, I work with neural models and I have problem with plotting fast spiking data. The spikes on the plot appear to have different hight which changes when I for example resize the plot window. The same problem is with saving data into files, especially in vector formats. I found the information about changing the join style, it helps a bit (rounded is the best) but doesn't solve the problem. For raster formats the workaround is to save the data in higher resolution, using DPI option. Below I included links to examples. Normal example (100 DPI): https://docs.google.com/leaf?id=0B3NZY3443E1VY2JjNTc3MjAtZDI5NC00OThjLTgwY2EtNTVhMDVkZWQ2YzIwhl=en Example 300 DPI: https://docs.google.com/leaf?id=0B3NZY3443E1VNDdkMzUzNzAtMjdmNC00NjFmLTliMzMtODE5MzExMmNjNjQzhl=en The problem is vector files (I'm especially interested in EPS) ignore DPI option. I've experimented with different backends and the problem is persistent on GUI and 'file-writting' back ends. I also tested it on Linux and Mac, and the outcome does't change. Creating larger figures sometimes helps a bit but not always solve the problem. Just to mention that plotting this kind of data is possible 'out of the box' in Matlab or XPPAut (which is not the most fancy plotting tool) I get a proper outcome. Maybe there should be a option to plot raw data in some sense, or join style function that deals with such a plots in a proper fashion. If you need some additional information do not hesitate to ask. Thanks for the help in advance. This is a known bug in the latest release of mpl that is fixed in svn. We do need to get a new release out soon. It is easy to fix by uncomennting the path.simplify : False line in your matplotlibrc file (the default is True). Note that path.simplify does work in the svn release of matplotlib -- it is designed to reduce paths but be imperceptible to the eye at the resolution plotted, but due to a bug in the released version it can result in improper simplifications. See http://matplotlib.sourceforge.net/users/customizing.html for information on how to change your matplotlibrc settings. JDH -- Download Intel#174; Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users