[SciPy-dev] Scipy-dev Digest, Vol 72, Issue 6

Arkapravo Bhaumik arkapravobhaumik@gmail....
Mon Oct 5 22:32:29 CDT 2009


Hi

My immediate plan was to try out scipy for some of those 'exotic' integrals
which can only be sought through the route of contour integration (
http://programming-unlimited.blogspot.com/2009/10/exotic-integrals-in-scipy.html).
My immediate observation has been the comments made in scipy along with the
output is not very consistent. I will soon post my detailed observations.

Comparing with Matlab etc , I am sure we cannot 'copy' a module (and I doubt
that may help). My plan was to used Matlab as a motivation , I guess that in
about 3-5 years scipy will take over all the numerical/scientific software.
I was wondering that employing some Pythonic-GUI (wx, tk,turtle, frog etc)
is it possible to develop software as Simulink ?

Best regards

Arkapravo

2009/10/6 <scipy-dev-request@scipy.org>

> Send Scipy-dev mailing list submissions to
>        scipy-dev@scipy.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
>        http://mail.scipy.org/mailman/listinfo/scipy-dev
> or, via email, send a message with subject or body 'help' to
>        scipy-dev-request@scipy.org
>
> You can reach the person managing the list at
>        scipy-dev-owner@scipy.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Scipy-dev digest..."
>
>
> Today's Topics:
>
>   1. Re: Volunteer for Scipy Project (Anne Archibald)
>   2. Re: Volunteer for Scipy Project (David Goldsmith)
>   3. Re: Volunteer for Scipy Project (Anne Archibald)
>   4. Re: Volunteer for Scipy Project (Charles R Harris)
>   5. Re: Volunteer for Scipy Project (Anne Archibald)
>   6. Re: Volunteer for Scipy Project (Charles R Harris)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 5 Oct 2009 16:40:46 -0400
> From: Anne Archibald <peridot.faceted@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <ce557a360910051340l2e5af73es5b7498abf677db7@mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> 2009/10/5 David Cournapeau <david@ar.media.kyoto-u.ac.jp>:
> > Hi Arkapravo,
> >
> > Arkapravo Bhaumik wrote:
> >>
> >> ? ? Dear Sir/Ma'am
> >>
> >> ? ? I was in e-mail contact with one of your colleagues , Travis ; ?I
> >> ? ? am very interested in contributing to scipy project as a
> >> ? ? volunteer. I believe that I can contribute in
> >>
> >> ? ? (1) Documentation
> >> ? ? (2) Suggesting and developing newer functionality
> >> ? ? (3) Study similar software as Matlab, Mathematica, Maple etc
> >> ? ? trying to look for inspiration for possible improvements in scipy
> >>
> >
> > Thanks for your help. As in most open source projects, the best way to
> > contribute is to work on something that you actually need for yourself,
> > be it code enhancement, bug fixes, documentation, etc...
> >
> > Depending on your interests and proficiency in python vs. C vs. Fortran,
> > there are several things that could be worked on: the ones which need
> > the most work ATM are scipy.special and scipy.interpolate.
> >
> > Please note also that you should be careful when 'studying' proprietary
> > software - reimplementing from scratch a similar functionality is OK,
> > re-using the implementation is almost never ok.
>
> I would add a mode of working that is definitely okay: find a
> reference to a journal article on how to compute the thing you're
> interested in (possibly by finding the reference in
> matlab/r/maple/mathematica/whatever source ordoumentation) and
> implementing what is described in the research paper. This is a great
> way to fill gaps in scipy with good, well-thought-out algorithms.
>
> Anne
>
>
> ------------------------------
>
> Message: 2
> Date: Mon, 5 Oct 2009 13:54:11 -0700
> From: David Goldsmith <d.l.goldsmith@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <45d1ab480910051354m3cc0f210hdb49b5a281948807@mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Just curious, Anne: have you anything in particular in mind (i.e., are
> there
> some small - or gaping - holes in scipy (IYO, of course) which you know
> could be filled by a careful implementation of something(s) extant in the
> literature)?
>
> DG
>
> On Mon, Oct 5, 2009 at 1:40 PM, Anne Archibald <peridot.faceted@gmail.com
> >wrote:
>
> > 2009/10/5 David Cournapeau <david@ar.media.kyoto-u.ac.jp>:
> > > Hi Arkapravo,
> > >
> > > Arkapravo Bhaumik wrote:
> > >>
> > >>     Dear Sir/Ma'am
> > >>
> > >>     I was in e-mail contact with one of your colleagues , Travis ;  I
> > >>     am very interested in contributing to scipy project as a
> > >>     volunteer. I believe that I can contribute in
> > >>
> > >>     (1) Documentation
> > >>     (2) Suggesting and developing newer functionality
> > >>     (3) Study similar software as Matlab, Mathematica, Maple etc
> > >>     trying to look for inspiration for possible improvements in scipy
> > >>
> > >
> > > Thanks for your help. As in most open source projects, the best way to
> > > contribute is to work on something that you actually need for yourself,
> > > be it code enhancement, bug fixes, documentation, etc...
> > >
> > > Depending on your interests and proficiency in python vs. C vs.
> Fortran,
> > > there are several things that could be worked on: the ones which need
> > > the most work ATM are scipy.special and scipy.interpolate.
> > >
> > > Please note also that you should be careful when 'studying' proprietary
> > > software - reimplementing from scratch a similar functionality is OK,
> > > re-using the implementation is almost never ok.
> >
> > I would add a mode of working that is definitely okay: find a
> > reference to a journal article on how to compute the thing you're
> > interested in (possibly by finding the reference in
> > matlab/r/maple/mathematica/whatever source ordoumentation) and
> > implementing what is described in the research paper. This is a great
> > way to fill gaps in scipy with good, well-thought-out algorithms.
> >
> > Anne
> > _______________________________________________
> > Scipy-dev mailing list
> > Scipy-dev@scipy.org
> > http://mail.scipy.org/mailman/listinfo/scipy-dev
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://mail.scipy.org/pipermail/scipy-dev/attachments/20091005/d95b297e/attachment-0001.html
>
> ------------------------------
>
> Message: 3
> Date: Mon, 5 Oct 2009 17:20:53 -0400
> From: Anne Archibald <peridot.faceted@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <ce557a360910051420i3d23703m81f8a032ef61c1e0@mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> 2009/10/5 David Goldsmith <d.l.goldsmith@gmail.com>:
> > Just curious, Anne: have you anything in particular in mind (i.e., are
> there
> > some small - or gaping - holes in scipy (IYO, of course) which you know
> > could be filled by a careful implementation of something(s) extant in the
> > literature)?
>
> Well, not exactly - the examples I had in mind were minor and/or in
> the past. I ran into problems with scipy's hyp2f1, for example, so I
> went and looked up the best algorithm I could find for it (and I think
> I contributed that code). I wanted the Kuiper test as an alternative
> to the Kolmogorov-Smirnov test (it's invariant under cyclic
> permutations, and is sensitive to different features of the
> distribution) so I looked up the test and the special function needed
> to interpret its results. (I haven't contributed this to scipy yet,
> mostly because I chose an interface that's not particularly compatible
> with that for scipy's K-S test.) And on a larger scale, that's what
> scipy.spatial's kdtree implementation is.
>
> For examples where I think a bit of lit review plus implementation
> work might help, I'd say that the orthogonal polynomials could use
> some work - the generic implementation in scipy.special falls apart
> rapidly as you go to higher orders. I always implement my own
> Chebyshev polynomials using the cos(n*arccos(x)) expression, for
> example, and special implementations for the others might be very
> useful.
>
> Anne
>
>
> ------------------------------
>
> Message: 4
> Date: Mon, 5 Oct 2009 18:16:50 -0600
> From: Charles R Harris <charlesr.harris@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <e06186140910051716q2a881b39q999e545f21f06b8d@mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> On Mon, Oct 5, 2009 at 3:20 PM, Anne Archibald <peridot.faceted@gmail.com
> >wrote:
>
> > 2009/10/5 David Goldsmith <d.l.goldsmith@gmail.com>:
> > > Just curious, Anne: have you anything in particular in mind (i.e., are
> > there
> > > some small - or gaping - holes in scipy (IYO, of course) which you know
> > > could be filled by a careful implementation of something(s) extant in
> the
> > > literature)?
> >
> > Well, not exactly - the examples I had in mind were minor and/or in
> > the past. I ran into problems with scipy's hyp2f1, for example, so I
> > went and looked up the best algorithm I could find for it (and I think
> > I contributed that code). I wanted the Kuiper test as an alternative
> > to the Kolmogorov-Smirnov test (it's invariant under cyclic
> > permutations, and is sensitive to different features of the
> > distribution) so I looked up the test and the special function needed
> > to interpret its results. (I haven't contributed this to scipy yet,
> > mostly because I chose an interface that's not particularly compatible
> > with that for scipy's K-S test.) And on a larger scale, that's what
> > scipy.spatial's kdtree implementation is.
> >
> > For examples where I think a bit of lit review plus implementation
> > work might help, I'd say that the orthogonal polynomials could use
> > some work - the generic implementation in scipy.special falls apart
> > rapidly as you go to higher orders. I always implement my own
> > Chebyshev polynomials using the cos(n*arccos(x)) expression, for
> > example, and special implementations for the others might be very
> > useful.
> >
> >
> At what order does the scipy implementation of the Chebyshev polynomials
> fall apart? I looked briefly at that package a long time ago, but never
> used
> it. I ask so I can check the chebyshev module that is going into numpy.
>
> Chuck
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://mail.scipy.org/pipermail/scipy-dev/attachments/20091005/41c05d5e/attachment-0001.html
>
> ------------------------------
>
> Message: 5
> Date: Mon, 5 Oct 2009 21:29:52 -0400
> From: Anne Archibald <peridot.faceted@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <ce557a360910051829n370b6a69s8d592d29b0511e9e@mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> 2009/10/5 Charles R Harris <charlesr.harris@gmail.com>:
> >
> >
> > On Mon, Oct 5, 2009 at 3:20 PM, Anne Archibald <
> peridot.faceted@gmail.com>
> > wrote:
> >>
> >> 2009/10/5 David Goldsmith <d.l.goldsmith@gmail.com>:
> >> > Just curious, Anne: have you anything in particular in mind (i.e., are
> >> > there
> >> > some small - or gaping - holes in scipy (IYO, of course) which you
> know
> >> > could be filled by a careful implementation of something(s) extant in
> >> > the
> >> > literature)?
> >>
> >> Well, not exactly - the examples I had in mind were minor and/or in
> >> the past. I ran into problems with scipy's hyp2f1, for example, so I
> >> went and looked up the best algorithm I could find for it (and I think
> >> I contributed that code). I wanted the Kuiper test as an alternative
> >> to the Kolmogorov-Smirnov test (it's invariant under cyclic
> >> permutations, and is sensitive to different features of the
> >> distribution) so I looked up the test and the special function needed
> >> to interpret its results. (I haven't contributed this to scipy yet,
> >> mostly because I chose an interface that's not particularly compatible
> >> with that for scipy's K-S test.) And on a larger scale, that's what
> >> scipy.spatial's kdtree implementation is.
> >>
> >> For examples where I think a bit of lit review plus implementation
> >> work might help, I'd say that the orthogonal polynomials could use
> >> some work - the generic implementation in scipy.special falls apart
> >> rapidly as you go to higher orders. I always implement my own
> >> Chebyshev polynomials using the cos(n*arccos(x)) expression, for
> >> example, and special implementations for the others might be very
> >> useful.
> >>
> >
> > At what order does the scipy implementation of the Chebyshev polynomials
> > fall apart? I looked briefly at that package a long time ago, but never
> used
> > it. I ask so I can check the chebyshev module that is going into numpy.
>
> By n=30 they are off by as much as 0.0018 on [-1,1]; n=31 they are off
> by 0.1, and by n=35 they are off by four - not great for values that
> should be in the interval [-1,1]. This may seem like an outrageously
> high degree for a polynomial, but there's no reason they need be this
> bad, and one could quite reasonably want to use an order this high,
> say for function approximation.
>
> I think this inaccuracy is probably inevitable in a scheme that
> computes values using a recurrence relation, and something like it
> probably occurs for all the orthogonal polynomials that don't have
> special-purpose evaluators.
>
> Anne
>
>
> ------------------------------
>
> Message: 6
> Date: Mon, 5 Oct 2009 21:12:49 -0600
> From: Charles R Harris <charlesr.harris@gmail.com>
> Subject: Re: [SciPy-dev] Volunteer for Scipy Project
> To: SciPy Developers List <scipy-dev@scipy.org>
> Message-ID:
>        <e06186140910052012i39245147g6b3769bc315d8f7c@mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> On Mon, Oct 5, 2009 at 7:29 PM, Anne Archibald <peridot.faceted@gmail.com
> >wrote:
>
> > 2009/10/5 Charles R Harris <charlesr.harris@gmail.com>:
> > >
> > >
> > > On Mon, Oct 5, 2009 at 3:20 PM, Anne Archibald <
> > peridot.faceted@gmail.com>
> > > wrote:
> > >>
> > >> 2009/10/5 David Goldsmith <d.l.goldsmith@gmail.com>:
> > >> > Just curious, Anne: have you anything in particular in mind (i.e.,
> are
> > >> > there
> > >> > some small - or gaping - holes in scipy (IYO, of course) which you
> > know
> > >> > could be filled by a careful implementation of something(s) extant
> in
> > >> > the
> > >> > literature)?
> > >>
> > >> Well, not exactly - the examples I had in mind were minor and/or in
> > >> the past. I ran into problems with scipy's hyp2f1, for example, so I
> > >> went and looked up the best algorithm I could find for it (and I think
> > >> I contributed that code). I wanted the Kuiper test as an alternative
> > >> to the Kolmogorov-Smirnov test (it's invariant under cyclic
> > >> permutations, and is sensitive to different features of the
> > >> distribution) so I looked up the test and the special function needed
> > >> to interpret its results. (I haven't contributed this to scipy yet,
> > >> mostly because I chose an interface that's not particularly compatible
> > >> with that for scipy's K-S test.) And on a larger scale, that's what
> > >> scipy.spatial's kdtree implementation is.
> > >>
> > >> For examples where I think a bit of lit review plus implementation
> > >> work might help, I'd say that the orthogonal polynomials could use
> > >> some work - the generic implementation in scipy.special falls apart
> > >> rapidly as you go to higher orders. I always implement my own
> > >> Chebyshev polynomials using the cos(n*arccos(x)) expression, for
> > >> example, and special implementations for the others might be very
> > >> useful.
> > >>
> > >
> > > At what order does the scipy implementation of the Chebyshev
> polynomials
> > > fall apart? I looked briefly at that package a long time ago, but never
> > used
> > > it. I ask so I can check the chebyshev module that is going into numpy.
> >
> > By n=30 they are off by as much as 0.0018 on [-1,1]; n=31 they are off
> > by 0.1, and by n=35 they are off by four - not great for values that
> > should be in the interval [-1,1]. This may seem like an outrageously
> > high degree for a polynomial, but there's no reason they need be this
> > bad, and one could quite reasonably want to use an order this high,
> > say for function approximation.
> >
> >
> Hmm, I get an maximum error of about 1e-13 for n=100 using my routine,
> which
> isn't great and can be improved a bit with a few tricks, but is probably
> good enough.  That's using simple Clenshaw recursion for the evaluation.
> There must be something seriously wrong with the scipy version. I routinely
> use fits with n > 50 because truncating such a series gives a good
> approximation to the minmax fit and it's also nice to see how the
> coefficients fall off to estimate the size of the truncation error.
>
> I think this inaccuracy is probably inevitable in a scheme that
> > computes values using a recurrence relation
>
>
> Not so, using the Cheybshev recurrence in either direction should be stable
> for |x| <= 1. It's like multiplying with a complex number of modulus 1,
> i.e., a unitary matrix.
>
>
> > and something like it probably occurs for all the orthogonal polynomials
> > that don't have
> > special-purpose evaluators.
> >
>
> Depends on the recursion, the direction of the recursion, and the domain.
>
> Chuck
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://mail.scipy.org/pipermail/scipy-dev/attachments/20091005/f991a2ef/attachment.html
>
> ------------------------------
>
> _______________________________________________
> Scipy-dev mailing list
> Scipy-dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-dev
>
>
> End of Scipy-dev Digest, Vol 72, Issue 6
> ****************************************
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/scipy-dev/attachments/20091006/520b6163/attachment-0001.html 


More information about the Scipy-dev mailing list