Add is_series() method and others to symbolic expressions

[vbraun]

In order to get more information out of symbolic expressions, I added is_series() and is_terminating_series() methods. This is really necessary if you want to write functions that accept symbolic series as input.

I also noticed that there are various methods _is_something(), which would likewise be useful but are not "public" because of the leading underscore. The second patch removes the leading underscore from them to make them part of the public API.

Apply

• attachment: trac_10859_remove_underscores_from_some_symbolic_expression_methods.patch
• attachment: trac_10859_add_is_series_to_symbolic_expression.patch

Depends on #11320

Component: symbolics

Keywords: sd31

Author: Volker Braun

Reviewer: Karl-Dieter Crisman, Burcin Erocal

Merged: sage-4.7.1.alpha4

Issue created by migration from https://trac.sagemath.org/ticket/10859

[vbraun]

comment:1

Apply trac_10859_add_is_series_to_symbolic_expression.patch, trac_10859_remove_underscores_from_some_symbolic_expression_methods.patch

[kcrisman]

comment:2

A few comments/questions:

• Are there any other symbolic objects that would look like series but not be caught by this? (For instance, ones gotten by doing SR(some Maxima thing that returns a series).) I guess I'm wondering if this is only supposed to find things created by .series() or if it's supposed to catch anything that's a series.
• Does this include Laurent series?
• There are a couple weird typos, including via the :meth:`series` method.a
• The Boolean. Whether ``self`` was constructed by :meth:`series`. should look more like it does for other outputs, or at least have a semicolon or colon or something for both of these functions.
• I think that the second patch should be part of a separate ticket. Presumably someone thought it was good for the methods to be hidden, so it would be helpful to know why that is.
• Should we maybe even discuss whether these new methods should be underscored?

I apologize for not having time to give this a proper review, though on a cursory examination it looks fine in terms of the actual code and tests.

[kcrisman]

Reviewer: Karl-Dieter Crisman

[vbraun]

comment:3

1. GiNac has a special operand for series, and is_series() returns true if and only if the expression is a series in this sense. The output will include an Order() term. Maxima does not have an order term as far as I know:

sage: maxima(sin(x)).taylor(maxima(x), 0, 10)
x-x^3/6+x^5/120-x^7/5040+x^9/362880

so converting maxima series to the symbolic ring necessarily creates ordinary polynomials.

2. Laurent series are included:

sage: (cos(x)/x).series(x,10)
1*x^(-1) + (-1/2)*x + 1/24*x^3 + (-1/720)*x^5 + 1/40320*x^7 + (-1/3628800)*x^9 + Order(x^10)
sage: _.is_series()
True

3. I think removing the underscores from other is_something() is in the spirit of this ticket of exposing more introspection capabilities to the user. I don't think there is any argument to be made that the user must not be able to access these functions. Rather, it adds obvious functionality to the public API.

I'll fix some english language typos and upload a revised version.

[burcin]

comment:4

Replying to @vbraun:

3. I think removing the underscores from other is_something() is in the spirit of this ticket of exposing more introspection capabilities to the user. I don't think there is any argument to be made that the user must not be able to access these functions. Rather, it adds obvious functionality to the public API.

I agree that these functions should be exposed to the user. I was the one to add them with an underscore in the first place, with the thought that they are only useful if you want to get into the internals of symbolic expressions. But anybody who wants to do any serious programming with symbolics needs them. They should be more visible.

I also appreciate any effort to clean up the public API of symbolic expressions. This has been a mess for a long time. AFAIK, it was written to expose some basic maxima functionality initially, then somehow it was decided to keep this old interface and not use the (much better designed) ginac one.

[vbraun]

Updated patch

[vbraun]

comment:5

Attachment: trac_10859_add_is_series_to_symbolic_expression.patch.gz

[burcin]

Changed reviewer from Karl-Dieter Crisman to Karl-Dieter Crisman, Burcin Erocal

[burcin]

comment:6

• Patchbot reports errors in sage/plot/plot.py and sage/symbolic/expression_conversions.py.
  • This also shows that there might be other users of these underscore methods and we should deprecate them instead of just renaming them.
  • In attachment: trac_10859_remove_underscores_from_some_symbolic_expression_methods.patch, a variable name self_is_zero is changed to selfis_zero. This hunk should be removed from the patch.

Thank you for taking the time to improve symbolics. :)

[vbraun]

comment:7

Thanks for having a look. The offending calls to underscored methods were not there in Sage-4.6.2 when I wrote the patch. So I'd argue they demonstrate once more the need for exposing this functionality. I don't think we should formally deprecate the methods since they are private, and nobody outside of the Sage library is supposed to use them.

I'll fix the other issues and will test it against the pynac-0.2.2 update.

[burcin]

comment:8

I am not a fan of the deprecation policy either, but I have been advertising those functions for a while ![1] and in this case it is not so much trouble to deprecate them.

![1] http://wiki.sagemath.org/symbolics/rewrite?action=AttachFile (which should be attached to a trac ticket)

You just need to do _is_something = deprecated_function_alias(is_something, 'Sage 4.7.1') for each function. deprecated_function_alias is in sage.misc.misc.

[vbraun]

Dependencies: #11320

[vbraun]

comment:9

All tests pass!

[burcin]

comment:10

The patchbot doesn't read the dependencies field. Depends on #11320.

We can switch this to a positive review once the bot agrees. The failing maxima tests reported are not relevant.

[vbraun]

comment:11

I think the patch buildbot also won't rebuild it just because you uttered "Depends on ...". Though you can take my word that it builds and doctests fine on Sage-4.7.rc1 ;-)

[burcin]

comment:12

2 weeks! I totally forgot about this. Sorry for the delay.

[jdemeyer]

comment:14

There are lots of doctest failures because of the DeprecationWarning.

[vbraun]

comment:15

The maxima interface uses the private _is_... methods at one point, this leads to the large number of doctest failures. The updated patch takes care of that. All doctests pass on Sage-4.7.1-alpha2

[vbraun]

Changed keywords from none to sd31

[vbraun]

Attachment: trac_10859_remove_underscores_from_some_symbolic_expression_methods.patch.gz

Updated patch

[jdemeyer]

Merged: sage-4.7.1.alpha4