Mailman 3 February 2008 - SciPy-Dev

SLSQP Constrained Optimizer Status
by Rob Falck 24 Nov '13

24 Nov '13

I'm currently implementing the Sequential Least Squares Quadratic Programming (SLSQP) optimizer, by Dieter Kraft, for use in Scipy. The Fortran code being wrapped with F2PY is here: http://www.netlib.org/toms/733 (its use within Scipy has been cleared) SLSQP provides for bounds on the independent variables, as well as equality and inequality constraint functions, which is a capability that doesn't exist in scipy.optimize. Currently the code works, although the constraint normals are being generated by approximation. I'm working on a way to pass these in. I think the most elegant way will be a single function that returns the matrix of constraint normals. For a demonstration of what the code can do, here is an optimization of f(x,y) = 2xy + 2x - x**2 - 2y**2 Example 14.2 in Chapra & Canale gives the maximum as x=2.0, y=1.0. The unbounded optimization tests find this solution. As expected, its faster when derivatives are provided rather than approximated. Unbounded optimization. Derivatives approximated. Elapsed time: 1.45792961121 ms Results [[1.9999999515712266, 0.99999996181577444], -1.9999999999999984, 4, 0, 'Optimization terminated successfully.'] Unbounded optimization. Derivatives provided. Elapsed time: 1.03211402893 ms Results [[1.9999999515712266, 0.99999996181577444], -1.9999999999999984, 4, 0, 'Optimization terminated successfully.'] The following example uses an equality constraint to find the optimal when x=y. Bound optimization. Derivatives approximated. Elapsed time: 1.384973526 ms Results [[0.99999996845920858, 0.99999996845920858], -0.99999999999999889, 4, 0, 'Optimization terminated successfully.'] I've tried to conform to the syntax used by the other optimizers in scipy.optimize. The function definition and doc string are below. If anyone is interested in testing it out, let me know. def fmin_slsqp( func, x0 , eqcons=[], f_eqcons=None, ieqcons=[], f_ieqcons=None, bounds = [], fprime = None, fprime_cons=None,args = (), iter = 100, acc = 1.0E-6, iprint = 1, full_output = 0, epsilon = _epsilon ): """ Minimize a function using Sequential Least SQuares Programming Description: Python interface function for the SLSQP Optimization subroutine originally implemented by Dieter Kraft. Inputs: func - Objective function (in the form func(x, *args)) x0 - Initial guess for the independent variable(s). eqcons - A list of functions of length n such that eqcons[j](x0,*args) == 0.0 in a successfully optimized problem f_eqcons - A function of the form f_eqcons(x, *args) that returns an array in which each element must equal 0.0 in a successfully optimized problem. If f_eqcons is specified, eqcons is ignored. ieqcons - A list of functions of length n such that ieqcons[j](x0,*args) >= 0.0 in a successfully optimized problem f_ieqcons - A function of the form f_ieqcons(x0, *args) that returns an array in which each element must be greater or equal to 0.0 in a successfully optimized problem. If f_ieqcons is specified, ieqcons is ignored. bounds - A list of tuples specifying the lower and upper bound for each independent variable [ (xl0, xu0), (xl1, xu1), ...] fprime - A function that evaluates the partial derivatives of func fprime_cons - A function of the form f(x, *args) that returns the m by n array of constraint normals. If not provided, the normals will be approximated. Equality constraint normals precede inequality constraint normals. args - A sequence of additional arguments passed to func and fprime iter - The maximum number of iterations (int) acc - Requested accuracy (float) iprint - The verbosity of fmin_slsqp. iprint <= 0 : Silent operation iprint == 1 : Print summary upon completion (default) iprint >= 2 : Print status of each iterate and summary full_output - If 0, return only the minimizer of func (default). Otherwise, output final objective function and summary information. epsilon - The step size for finite-difference derivative estimates. Outputs: ( x, { fx, gnorm, its, imode, smode }) x - The final minimizer of func. fx - The final value of the objective function. its - The number of iterations. imode - The exit mode from the optimizer, as an integer. smode - A string describing the exit mode from the optimizer. Exit modes are defined as follows: -1 : Gradient evaluation required (g & a) 0 : Optimization terminated successfully. 1 : Function evaluation required (f & c) 2 : Number of equality constraints larger than number of independent variables 3 : More than 3*n iterations in LSQ subproblem 4 : Inequality constraints incompatible 5 : Singular matrix E in LSQ subproblem 6 : Singular matrix C in LSQ subproblem 7 : Rank-deficient equality constraint subproblem HFTI 8 : Positive directional derivative for linesearch 9 : Iteration limit exceeded -- - Rob Falck

4 16

Question about Kaiser Implementation in firwin
by Buehler, Eric (AGRE) 04 Oct '08

04 Oct '08

Hello, I have been using the signal.firwin FIR design tool and I have a question about the implementation. Scipy-0.4.8 Numpy-0.9.6 I am generating a kaiser window with the following parameters: N = 128 Cutoff = ~4e7 width = .3 All of the special kaiser generation parameters in the function follow Oppenheim and Schafer well. Even the sinc function for linear phase is follows exactly. However, the last line of the firwin function is causing me some heartburn. filter_design.py 1538 win = get_window(window,N,fftbins=1) 1539 alpha = N//2 1540 m = numpy.arange(0,N) 1541 h = win*special.sinc(cutoff*(m-alpha)) 1542 return h / sum(h) Line 1542 of filter_design.py, "return h / sum(h)", normalizes the function where it doesn't seem necessary, at least in the kaiser window case. Without the normalization, the kaiser window already returns a value of 1 at the zero frequency point. This normalization scales all of the data, making the window difficult to use in the frequency domain. Can someone point me to the rationale for this line? Looking at the code, this seems to be a pretty recent change (within the last year/year and a half). Thanks, Eric Buehler eric <dot> buehler <at> smiths-aerospace <dot> com ****************************************** The information contained in, or attached to, this e-mail, may contain confidential information and is intended solely for the use of the individual or entity to whom they are addressed and may be subject to legal privilege. If you have received this e-mail in error you should notify the sender immediately by reply e-mail, delete the message from your system and notify your system manager. Please do not copy it for any purpose, or disclose its contents to any other person. The views or opinions presented in this e-mail are solely those of the author and do not necessarily represent those of the company. The recipient should check this e-mail and any attachments for the presence of viruses. The company accepts no liability for any damage caused, directly or indirectly, by any virus transmitted in this email. ******************************************

3 2

Moving numexpr to its own project
by David M. Cooke 15 Apr '08

15 Apr '08

Hi, With the pending closing of the scipy.sandbox (anybody want to cough up more details on that, btw? When? How?), I want to move numexpr to its own project, likely hosted on Google Code. It's not something I think is "sciencey" enough for a scikit (plus, I don't like sharing a bug database with other, separate, projects), and it's more like weave (but simpler). Tim H., Francesc, Ivan? You're the main contributors; does this work for you? -- |>|\/|< /------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm(a)physics.mcmaster.ca

6 12

Status of sandbox.spline and my other code.
by John Travers 07 Apr '08

07 Apr '08

Hi all, This email is intended to clarify the status of various bits of code I've committed/not committed so I don't leave too much of a mess after me. I'm about to start writing my PhD thesis, so I'll be too busy to work on scipy until September. 1. sandbox.spline stuff The module here was supposed to be a tidy up of scipy.interpolate. All of the dierkx functionality has been moved to f2py wrappers, which I think makes maintenance easier. However, it does not add any new functionality, and in retrospect appears to have been a waste of my time. So I'll leave to you to decide if you want to integrate it into scipy.interpolate or not; though it was originaly planned to be a seperate module to clear up the ambiguity between interpolation and smoothing spline functionality. However I have added 20 unit tests to the code (most of which simply check the wrapper against the pure fortran output, but still useful I think) which could easily be moved over to the current scipy.interpolate module. 2. Radial basis function module in sandbox (rbf) This code has recently had attention from Robert Hetland and is quite improved. I'm not sure if I will ever go into scipy though?? If not, I will put it into a scikit at some point. Related to this, the wiki page has been updated, however, it is at http://www.scipy.org/RadialBasisFunctions but should probably be in the Cookbook. I don't know how to move it over, so some pointers would be helpful. In addition there are four unused attachments I uploaded that could be deleted. 3. Boundary value ODE solver (BVP_SOLVER) I have fairly functional code interfacing to some very good BVP solver code (http://cs.smu.ca/~muir/BVP_SOLVER_Webpage.shtml). The license is good for scipy (as confirmed by the authors), but the code if fortran 95 - therefore cannot apparently go into scipy. I'll make a scikit when I get time (probably September). 4. An interface to the pikaia genetic algorithm routine I have a python interface to pikaia fortran code for genetic algorithm optimisation of real valued functions (http://www.hao.ucar.edu/Public/models/pikaia/pikaia.html). The fortran code itself is robust and highly optimised for numerical work, though it is nowhere near as general as the sandbox.ga module. I think it is useful to go in scipy.optimization alongside the annealing module. If people disagree I'll make it into a scikit. I hope this clears a few things up. Best regards, John Travers

4 4

could anyone propose University of California workers to change bvls license?
by dmitrey 05 Mar '08

05 Mar '08

hi all, here's a fortran routine bvls, free for non-commercial purposes: http://lib.stat.cmu.edu/general/bvls Could anyone contact to the authors of the routine and propose them make it available in terms of GPL/LGPL or, moreover, BSD license? I have already working Python interface (via f2py, thanks to Pearu Peterson for assist), so, if BSD - it could be connected to scipy. In anyway, this routine would be of great interest in scikits.openopt users, because solvers ralg and lincher could be enhanced (in case of >1 constraints), and, of course, mere bvls solver could be included to LLSP class. Authors of the routine, as it is mentioned in the file header, are Robert L. Parker Philip B. Stark Scripps Institution of Oceanography Department of Statistics University of California, San Diego University of California La Jolla CA 92093 Berkeley CA 94720-3860 rlparker(a)ucsd.edu stark(a)stat.berkeley.edu I would contact the authors of the routine by myself, but IIRC some scipy developers work in this university and have better English language knowledge. Regards, D.

4 4

f2py + f90 file
by dmitrey 03 Mar '08

03 Mar '08

hi all, I have an .f90 file with a routine that I intend to connect to Python via f2py (the file is http://users.bigpond.net.au/amiller/bvls.f90 ) I have gfortran and f95 installed (KUBUNTU 7.10), *no f90* (f90 is absent in KUBUNTU software update channel as well). Both they work with the file correctly, so I can get both a.out and bvlc.o (however I don't know how to get *.so files) So f2py -c -m bvls bvls.f90 yields: ... Fortran f77 compiler: /usr/bin/g77 -g -Wall -fno-second-underscore -fPIC -O3 -funroll-loops -march=athlon64 -mmmx -m3dnow -msse2 -msse compile options: '-I/tmp/tmpK2eM1Z/src.linux-x86_64-2.5 -I/usr/lib/python2.5/site-packages/numpy/core/include -I/usr/include/python2.5 -c' error: f90 not supported by GnuFCompiler needed for bvls.f90 if I rename the file extension to f95 I have the same problem: error: f90 not supported by GnuFCompiler needed for bvls.f95 Any suggestions? Thank you in advance, D.

2 9

ANN: Enthought Python Distribution - Beta
by Travis Vaught 01 Mar '08

01 Mar '08

Greetings, Enthought is very excited about our pending wide-release of the Enthought Python Distribution (EPD). After much effort, we finally think we're close to the first non-beta release. As one more quality check, we'd love to impose on you guys one more time to try out a just- minted beta release for Windows (EPD 2.5.2001_beta1) and give us some feedback. Any major problems will, of course, be fixed for the next release, but we're open to any suggestions for improvement for future releases as well. http://www.enthought.com/epd For those of you unfamiliar with EPD, it's a "kitchen-sink-included" distribution of Python with over 60 additional tools and libraries. It's bundled into a nice MSI installer on Windows and includes NumPy, SciPy, IPython, 2D and 3D visualization, database adapters and a lot of other tools right out of the box. We'll have support for RedHat and Mac OS X in a general release very soon. For academic, non-profit or hobbyist use, EPD is, and will remain, free. We are charging an annual subscription for commercial and governmental access to downloads and updates of EPD. Downloaded files may be used indefinitely past the subscription term. You are welcome to try out the beta indefinitely, regardless of your commercial/non- commercial persuasion. When the final (non-beta) version is released, commercial folks can try it for 30 days. You can check out the license terms (http://www.enthought.com/products/epdlicense.php) if you're interested in the details. EPD is compelling because it solves a lingering packaging and distribution problem, but also because of the libraries which it includes. We owe many folks on this list a debt of gratitude for their work on some really great tools. So, thanks ... and enjoy! Best Regards, Travis N. Vaught

8 20

Experience with Goto's BLAS
by Nils Wagner 29 Feb '08

29 Feb '08

Hi all, Is there anyone who can report about the usage of Goto's BLAS with respect to numpy/scipy ? What can be said about advantages of ACML, MKL, ATLAS and Goto's BLAS ? Who can shed light on the differences between the different libraries ? Cheers, Nils References: http://www.tacc.utexas.edu/resources/software/software.php http://developer.amd.com/tools/acml/Pages/default.aspx http://www.intel.com/cd/software/products/asmo-na/eng/307757.htm http://math-atlas.sourceforge.net/

2 1

[arpack] eigen_symmetric and the sigma value
by Matthieu Brucher 28 Feb '08

28 Feb '08

Hi, Is there any plan of adding the support of the sigma keyword, like it is done in Matlab (http://www.mathworks.com/access/helpdesk/help/techdoc/ref/** eigs.html) ? It is needed for some algorithms, like the Locally Linear Embedding I'd like to put it the learn scikit. Matthieu -- French PhD student Website : http://matthieu-brucher.developpez.com/ Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92 LinkedIn : http://www.linkedin.com/in/matthieubrucher

2 6

Re: [SciPy-dev] Scipy, numpy and Abaqus
by Gilbert, Linda M CTR USAF AFRL/RXOC 27 Feb '08

27 Feb '08

> _____________________________________________ > From: Gilbert, Linda M CTR USAF AFRL/MLOC > Sent: Thursday, January 03, 2008 11:01 AM > To: 'scipy-dev(a)scipy.org'; 'pearu(a)cens.ioc.ee' > Subject: Scipy, numpy and Abaqus > > Hello, > > Do you know how to get numpy and scipy to work from within abaqus? > > Thank you in advance, > > Linda Gilbert > AFRL/RXOC > RCF Information Systems > 937-904-7632 > >

3 3