View Full Version : plotting in Python

05-01-2009, 07:57 AM
Why is VNI promoting the open-sourced matplotlib library for plotting in Python instead of PV-WAVE (as Steve is doing in this post: [Dead link removed]?

Can't PV-WAVE be called from Python via CWAVEC? Is this a tacit admission that matplotlib is better than PV-WAVE? Or is VNI going to develop a PV-WAVE plugin for Python?

Secondly, if VNI recommends matplotlib over PV-WAVE, why don't they recommend pylab over PyIMSL?

05-01-2009, 01:41 PM
I read that post you linked to. I think Steve is just recommending matplotlib as the graphics library to use for Python programmers. Seems pretty straightforward.

I am not sure about PV-WAVE from Python via CWAVEC, but that seems doable. Please report back with results if you try it.

Finally, your last question does not make sense. First, I don?t think someone saying ?if you are a Python programmer consider matplotlib for plotting? as a recommendation for using Python over PV-WAVE. Second, and more importantly, there is little to no overlap between pylab and PyIMSL: pylab is a MATLAB-like interface to the matplotlib graphics libraries and PyIMSL is a set of numerical analysis libraries.

05-05-2009, 10:12 AM
Yes, VNI is recommending Python programmers use PyIMSL and matplotlib for Python development.

Is VNI one company or two? They ported IMSL to Python; can't they make a PV-WAVE plugin for Python? You would think that a company whose financial interest is being threatened by matplotlib would bring their product up to par and present an commercial alternative to matplotlib. Instead they are recommending their competition.

But apparently they want it both ways. They would have you buy PyIMSL when you can use the open-source SciPy. If matplotlib is good enough for visualization, why wouldn't SciPy be good enough for numeric analysis?

Margaret Journey
05-06-2009, 08:59 AM
I think you have missed the whole concept of PyIMSL Studio. PyIMSL Studio is recommended for prototype to production of advanced analytical models into production applications and predominantly in C. IMSL is the defining technology of PyIMSL Studio, not the Python tools. Please read the following to help clarify your knowledge of the product.

White paper: Introducing PyIMSL Studio - [Dead link removed]

White paper: Analytic Modeling in Python - [Dead link removed]

We also did not port IMSL to Python, please keep the facts straight, read the previous posts and our Website, PyIMSL are Python wrappers. The PyIMSL Studio product is not about matplotlib or Scipy for production use, it is about embedding the IMSL C Numerical Library into production use at a significantly lower cost than traditional prototype to production methods.

05-06-2009, 03:03 PM
Many developers such as myself are not required to deploy their analytical models in C but are still interested in doing scientific computing in Python -- including visualization. In fact, Python can be a acceptable production environment for some cases, as your references point out:

For some applications, performance of Python might be adequate. If this is the case, the prototype developed using PyIMSL Studio can simply be deployed into production. ... Python can be used for the production deployment of applications because there are Python components available for creating high quality user interfaces for standalone applications and for web-based application development.

So my question still stands: Does VNI stand by the quality of PV-WAVE's visualization like it does the IMSL analytics? If so, why would they promote an open-sourced visualization module like matplotlib? Why would they not put wrappers around the PV-WAVE plotting capability and package it with PyIMSL?

When I say VNI ported IMSL to Python, I mean ported the functionality for use in Python. PV-WAVE is described as incorporating "sophisticated analysis routines based on the industry-standard IMSL Numerical Libraries for reliable and precise numerical analysis." I.e., you have exposed the analytic side of PV-WAVE to Python. Can't you expose the visualization side? (After all, you are a software company, and PV-WAVE is your flagship visualization product.)

And the reason I mention SciPy is for counter-argument sake. If you say, "Well, matplotlib is better suited for Python visualization", then I can respond, "If matplotlib is the de facto standard for Python visualization, is not SciPy the new de facto industry standard for Python analysis?"