Hello,
I appreciate your patience in explaining my issue. Checking dimensions in
Sympy is useful for my work, so I am seeking to implement the check
correctly.
I understand what you're saying about Python variables versus Sympy
Symbols. There are two examples in my response below. The first ("Example
1") is an attempt to use the substitution-based approach. In Example 2 I
realized I could reduce my issue to three lines (which exclude irrelevant
aspects of Latex and Python variables).
Example 1 using latex to Sympy to dimensions
Here's what I get when I use substitution of dimensions:
>>> import sys
>>> sys.version
'3.6.9 (default, Apr 18 2020, 01:56:04) \n[GCC 8.4.0]'
>>> import sympy
>>> sympy.__version__
'1.7.dev'
>>> from sympy.physics.units import mass, length, time
>>> from sympy.physics.units.systems.si import dimsys_SI
>>> from sympy.parsing.latex import parse_latex
Define the expression in Latex, then convert to Sympy expression
>>> eq = parse_latex("F = m a")
>>> F, m, a = sympy.symbols('F m a')
These three on the left side are Python variables
For each Sympy symbol in eq, replace with the Sympy dimensions
>>> eq.subs({F: mass*length/time**2, m: mass, a: length/time**2})
False
That boolean return wasn't what I was aiming for, so below I break the
process down
Separate substitution into two steps and review the output
>>> eq.lhs.subs({F: mass*length/time**2, m: mass, a: length/time**2})
Dimension(length*mass/time**2)
>>> eq.rhs.subs({F: mass*length/time**2, m: mass, a: length/time**2})
Dimension(length/time**2)*Dimension(mass, M)
Now that I have confidence the Sympy substitution is behaving as expected,
save the results to two Python variables
>>> RHS = eq.rhs.subs({F: mass*length/time**2, m: mass, a: length/time**2})
>>> RHS
Dimension(length/time**2)*Dimension(mass, M)
>>> type(RHS)
<class 'sympy.core.mul.Mul'>
That Mul is potentially the cause of my issue with dimsys_SI.equivalent_dims
below. If there is a way to simplify the Mul of Dimension class objects,
that would probably eliminate the issue.
The left-hand side of eq returns a Dimension class object
>>> LHS = eq.lhs.subs({F: mass*length/time**2, m: mass, a: length/time**2})
>>> LHS
Dimension(length*mass/time**2)
>>> type(LHS)
<class 'sympy.physics.units.dimensions.Dimension'>
The issue of failing to compare the left-side dimensions and the right-side
dimensions could be due to the Mul, but I do not know how to simplify RHS.
Or I may be incorrectly supplying Python variables LHS and RHS
>>> dimsys_SI.equivalent_dims( RHS, LHS )
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File
"/usr/local/lib/python3.6/dist-packages/sympy/physics/units/dimensions.py",
line 454, in equivalent_dims
deps1 = self.get_dimensional_dependencies(dim1)
File
"/usr/local/lib/python3.6/dist-packages/sympy/physics/units/dimensions.py",
line 448, in get_dimensional_dependencies
dimdep = self._get_dimensional_dependencies_for_name(name)
File
"/usr/local/lib/python3.6/dist-packages/sympy/physics/units/dimensions.py",
line 422, in _get_dimensional_dependencies_for_name
for k, v in d.items():
AttributeError: 'NoneType' object has no attribute 'items'
After writing my example 1, I realized there is a simpler example of the
issue.
Example 2: the short version
>>> import sys
>>> sys.version
'3.6.9 (default, Apr 18 2020, 01:56:04) \n[GCC 8.4.0]'
>>> import sympy
>>> sympy.__version__
'1.7.dev'
>>> from sympy.physics.units import mass, length
>>> from sympy.physics.units.systems.si import dimsys_SI
>>> sympy.Mul(mass, length)
Dimension(length, L)*Dimension(mass, M)
>>> mass*length
Dimension(length*mass)
The above two lines appear equivalent but not equal
Same issue as in Example 1,
>>> dimsys_SI.equivalent_dims( sympy.Mul(mass, length), mass*length )
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/opt/sympy/physics/units/dimensions.py", line 461, in
equivalent_dims
deps1 = self.get_dimensional_dependencies(dim1)
File "/opt/sympy/physics/units/dimensions.py", line 455, in
get_dimensional_dependencies
dimdep = self._get_dimensional_dependencies_for_name(name)
File "/opt/sympy/physics/units/dimensions.py", line 429, in
_get_dimensional_dependencies_for_name
for k, v in d.items():
AttributeError: 'NoneType' object has no attribute 'items'
I think the reason is the Mul. However, I do not know how to simplify to
Dimension
>>> type(mass*length)
<class 'sympy.physics.units.dimensions.Dimension'>
>>> type(sympy.Mul(mass, length))
<class 'sympy.core.mul.Mul'>
>>> sympy.simplify(sympy.Mul(mass, length))
Dimension(length, L)*Dimension(mass, M)
Kindly,
Ben
On Tuesday, June 2, 2020 at 11:46:41 AM UTC-4, Aaron Meurer wrote:
>
> The issue here is you're confusing Python variables and SymPy symbols.
> The use of eval() isn't helping, because eval() evaluates based on the
> Python variables (I would try to avoid eval() if possible). See
> https://docs.sympy.org/latest/tutorial/gotchas.html. Once you define
> F, m, and a, they have no bearing on the symbols that are present in
> eq. SymPy has no way to know that there are Python variables named F,
> m, and a when looking at eq.
>
> What you can do is
>
> F, m, a = symbols('F m a')
> eq.subs({F: mass*length/time**2, m: mass, a: length/time**2}).
>
> to replace the symbols with units in the equation.
>
> Aaron Meurer
>
>
--
You received this message because you are subscribed to the Google Groups
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/sympy/993a7df4-b126-4ba1-8859-cd7000c2d3bc%40googlegroups.com.
