Automorphism Fields: Notation of Inverse

[gh-DeRhamSource]

At this stage we have the following problem with the notation of inverses:

sage: M = Manifold(2, 'M')
sage: X.<x,y> = M.chart()
sage: aib = M.automorphism_field([[0, 2], [-1, 0]], name='a^(-1)*b')
sage: c = aib.inverse(); c
Field of tangent-space automorphisms a^(-1)*b^(-1) on the 2-dimensional differentiable manifold M

That is definitely misleading. In this ticket it gets fixed by adding brackets around the element before taking its inverse if the name could cause any confusion:

sage: c = aib.inverse(); c
Field of tangent-space automorphisms (a^(-1)*b)^(-1) on the 2-dimensional differentiable manifold M

To solve this, the ticket provides a modification of is_atomic in sage/tensor/modules/format_utilities.py.

[gh-DeRhamSource]

New commits:

​8b9ac9a

Trac #28973: Brackets around element before inverse

[egourgoulhon]

Thanks for taking care of this! However, for simple names like a, having (a)^(-1) for the inverse is not desirable. One should maintain a^(-1) in this case. To cover all cases, you should use something like is_atomic() (defined in src/sage/tensor/modules/format_utilities.py); see the handling of names in FreeModuleAltForm.wedge(). More precisely, you should define, in format_utilities.py, a new function similar to is_atomic and adapted to the inverse naming.

[gh-DeRhamSource]

You're right. That's definitely preferrable.

This was the fastest solution I had in mind so far. I see what I can do.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​15067c1

Trac #28973: is_atomic adapted and used

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​c7fbb72

Trac #28973: Small improvement of is_atomic and Python3 compatibility in doctest

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​69ee528

Trac #28973: No raw string needed

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​48d84fe

Trac #28973: Automorphism fields adapted to new code

[egourgoulhon]

Thanks for the useful enhancement of the function is_atomic().

The patchbot complains about the change introduced in the commit of comment:10

+inside file:  b/src/sage/tensor/modules/format_utilities.py
+@@ -52,16 +55,31 @@ def is_atomic(expression):
++        sage: is_atomic("a<>b", sep=['<>'])
++        sage: is_atomic("(a<>b)", sep=['<>'])
+Python3 incompatible code inserted on 2 non-empty lines

Could you revert this change?

[egourgoulhon]

There is also a doctest error reported by one of the patchbots:

File "src/sage/manifolds/differentiable/automorphismfield.py", line 951, in sage.manifolds.differentiable.automorphismfield.AutomorphismFieldParal
Failed example:
    latex(inv)
Expected:
    R
Got:
    R^{-1}
**********************************************************************
1 item had failures:
   1 of  12 in sage.manifolds.differentiable.automorphismfield.AutomorphismFieldParal

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​dce291f	Trac #28973: Python3 compatibility and doctest
​db09dbc	Trac #28973: Merge branch 'develop' into automorphism_inverse_notation

[gh-DeRhamSource]

Like this?

[egourgoulhon]

Replying to gh-DeRhamSource:

Like this?

As you can see from the latest patchbot report, <> still causes some trouble:

-Python3 incompatible code inserted on 0 non-empty lines
+inside file:  b/src/sage/tensor/modules/format_utilities.py
+@@ -52,16 +55,31 @@ def is_atomic(expression):
++        sage: is_atomic(r"a<>b", sep=[r'<>'])
++        sage: is_atomic(r"(a<>b)", sep=[r'<>'])
+Python3 incompatible code inserted on 2 non-empty lines

Apparently, the operator <> has been removed in Python 3: it does not appear in the list ​https://docs.python.org/3/library/operator.html. So please remove this doctest.

[gh-DeRhamSource]

I see. Well, this is on the symbolic level, so no actual Python code is involved here. Even though another operator as example would be certainly more convenient.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​8403049

Trac #28973: Doctest improved

[gh-DeRhamSource]

Perhaps this is a good example because the separator consists of more than one letter.

[egourgoulhon]

Thanks for the change. The patchbot is now full green ==> good to go.