Updates:
Status: NeedsDecision
Cc: Vinzent.Steinberg [email protected] [email protected]
[email protected]
Comment #1 on issue 2622 by asmeurer: Map functions over iterables
http://code.google.com/p/sympy/issues/detail?id=2622
There's also @vectorize, though see issue 2182.
But I'm not a fan of this. First off, using a decorator destroys
information about the function. For example, do apart?? in isympy (or
source(apart) if you don't have IPython).
But to the point, I don't like implicit vectorization anyway. Remember
that explicit is better than implicit. You mention one problem with this,
which is that you might want a function to apply differently over a
container object, like exp(Matrix) representing the exponential of the
matrix, rather than the exponential of each item of the matrix.
Another example is functions that apply over lists of expressions. The
best example is solve(), which doesn't vectorize over a list, but rather
treats the list like a system of equations.
The solution in my opinion is to not have any automatic vectorization at
all. You can explicitly vectorize very easily in Python. If you have a
list, you can use a list comprehension ([sin(i) for i in yourlist]). For
matrix, we have Matrix.applyfunc. For sets and dictionaries, in Python 2.7
and Python 3 we have set and dictionary comprehensions.
So I'm tempted to mark this as WontFix. But I'll take the more cautious
path and mark it as NeedsDecision. For one thing, we already do have
functions that do this, and it's very inconsistant. Perhaps this should be
a separate issue, but there is also an inconsistency with automatic
sympification of strings.
Also, this behavior already exists, albiet inconsistently, for many SymPy
functions:
In [26]: simplify(a)
Out[26]:
⎡ 1 x ⎤
⎢──────, ──────⎥
⎢ 2 2 ⎥
⎣x - 1 x - 1⎦
In [27]: simplify(Tuple(*a))
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
/Users/aaronmeurer/Documents/python/sympy/sympy/<ipython-input-27-cc6abdb5bbd7>
in
<module>()
----> 1 simplify(Tuple(*a))
/Users/aaronmeurer/Documents/python/sympy/sympy/sympy/simplify/simplify.pyc
in simplify(expr, ratio)
1625 return together(powsimp(expr))
1626
-> 1627 expr = together(cancel(powsimp(expr)).expand())
1628
1629 if not isinstance(expr, Basic): # XXX: temporary hack
AttributeError: 'tuple' object has no attribute 'expand'
As for Functions, it doesn't already work in most cases:
In [28]: sin(a)
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
/Users/aaronmeurer/Documents/python/sympy/sympy/<ipython-input-28-66bf5e82d1e2>
in
<module>()
----> 1 sin(a)
/Users/aaronmeurer/Documents/python/sympy/sympy/sympy/core/cache.pyc in
wrapper(*args, **kw_args)
98 k = k + tuple(map(lambda x: type(x), k))
99 try:
--> 100 return func_cache_it_cache[k]
101 except KeyError:
102 pass
TypeError: unhashable type: 'list'
In [29]: sin(Tuple(*a))
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
/Users/aaronmeurer/Documents/python/sympy/sympy/<ipython-input-29-e37745b0e5f5>
in
<module>()
----> 1 sin(Tuple(*a))
/Users/aaronmeurer/Documents/python/sympy/sympy/sympy/core/cache.pyc in
wrapper(*args, **kw_args)
101 except KeyError:
102 pass
--> 103 func_cache_it_cache[k] = r = func(*args, **kw_args)
104 return r
105 return wrapper
/Users/aaronmeurer/Documents/python/sympy/sympy/sympy/core/function.pyc in
__new__(cls, *args, **options)
188 evaluate = options.pop('evaluate', True)
189 if evaluate:
--> 190 evaluated = cls.eval(*args)
191 if evaluated is not None:
192 return evaluated
/Users/aaronmeurer/Documents/python/sympy/sympy/sympy/functions/elementary/trigonometric.pyc
in
eval(cls, arg)
159 return
160
--> 161 if arg.could_extract_minus_sign():
162 return -cls(-arg)
163
AttributeError: 'Tuple' object has no attribute 'could_extract_minus_sign'
So we need to decide what should be done universally, and apply it
universally.
--
You received this message because you are subscribed to the Google Groups
"sympy-issues" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sympy-issues?hl=en.
