The more explanation I see from the OP, the more I oppose the idea. It
becomes more clear that this is a specialization, basically only to help
SymPy, while introducing semantic changes for everyone else (even if minor
ones).
Moreover, SymPy already has several readily available ways to do this,
albeit with slightly less appealing spellings.
Now, I WILL note that NumPy motivated several changes to Python. I believe
even strides in slices first benefitted NumPy. Likewise the ellipse object,
extended slices, and the `@` operator. However...
1. Unlike this proposal, none of those were backwards-incompatible changes
(beyond the trivial sense that "code that used to raise a SyntaxError
doesn't now).
2. The NumPy community (especially counting indirect users such as Panda,
scikit-learn, matplotlib, etc) is more than 10x the size of SimPy, and
probably more than 100x.
On Tue, May 18, 2021, 5:42 PM Martin Teichmann <martin.teichm...@gmail.com>
wrote:
> Hi Shreyan,
>
> we need this for symbolic math, e.g. in sympy. Probably you have never
> seen somebody doing symbolic math with Python, but believe me, there are
> many.
>
> Let me walk you through a toy problem to show where the issues are. Let's
> say we want to solve the equation x**2 == 3. Not a very tough equation, in
> reality things would be much more complicated. To start, we import sympy:
>
> >>> from sympy import symbols, solve
>
> Then we need to declare our variable:
>
> >>> x = symbols("x")
>
> Now we would just use the solve function to solve the equation.
> Unfortunately the == operator does not work (another, independent problem),
> so we have to reformulate our equation to x**2 - 3 == 0, and knowing that
> sympy implicitly adds "== 0" to the equation, we can just type
>
> >>> solve(x**2-3)
> [-sqrt(3), sqrt(3)]
>
> The result is not very astonishing. But wait, we realize we did a mistake,
> it is not actually 3 that we have to equate to, but 2/3! So we type
>
> >>> solve(x**2-2/3)
> [-0.816496580927726, 0.816496580927726]
>
> That went wrong. We wanted to have a symbolic result, not a numeric
> solution! How did that come about? Well, that's the point of discussion:
> 2/3 gets immediately turned into 0.66666..., and sympy has no clue where
> this is coming from. Sure, we can reformulate our problem:
>
> >>> solve(3*x**2-2)
> [-sqrt(6)/3, sqrt(6)/3]
>
> in our toy problem that is simple, but imagine what happens if you have
> 20+ terms that need to be changed. This is when it comes in handy to just
> be able to write 2/3 to mean two-third. Sympy gives you other ways than
> reformulating the problem to solve it, but that's a different topic. My
> point is, with my proposal, you can just write
>
> >>> solve(x**2 - 2/3)
> [-sqrt(6)/3, sqrt(6)/3]
>
> no weird things necessary.
>
> But you might ask, what if the fraction is not a constant? Well, in this
> case we need to declare another variable, and also tell solve in a second
> parameter which variable we want to solve for:
>
> >>> a=symbols("a")
> >>> solve(x**2 - a/3, x)
> [-sqrt(3)*sqrt(a)/3, sqrt(3)*sqrt(a)/3]
>
> ok, but if we now put some concrete value for a, would that not cause
> problems? Well the way you do that is to substitute:
>
> >>> equation = x**2 - a/3
> >>> solve(equation.subs(a, 2))
> [-sqrt(6)/3, sqrt(6)/3]
> >>> solve(equation.subs(a, 3))
> [-1, 1]
>
> as you see, there is no problem here. This is why the discussion whether
> the ratiofloat should also appear when we are dividing two variables which
> contain an int was not very interesting to me, as this is not a relevant
> problem for symbolic math.
>
> I hope I was able to show you where fractional constants are useful.
> Symbolic math in Python has many more problems, the fact that we cannot use
> == for example (I actually don't know why that is not possible, we should
> ask people from SymPy). Also the declaration of variables is a bit weird.
> But those problems are for another day.
>
> Cheers
>
> Martin
> _______________________________________________
> Python-ideas mailing list -- python-ideas@python.org
> To unsubscribe send an email to python-ideas-le...@python.org
> https://mail.python.org/mailman3/lists/python-ideas.python.org/
> Message archived at
> https://mail.python.org/archives/list/python-ideas@python.org/message/WQA2NVP7ECYEQFV67767547DWD2Q4PG6/
> Code of Conduct: http://python.org/psf/codeofconduct/
>
_______________________________________________
Python-ideas mailing list -- python-ideas@python.org
To unsubscribe send an email to python-ideas-le...@python.org
https://mail.python.org/mailman3/lists/python-ideas.python.org/
Message archived at
https://mail.python.org/archives/list/python-ideas@python.org/message/N7DGN5TN6H4HH7VIDVL2GPV5772QY6VX/
Code of Conduct: http://python.org/psf/codeofconduct/
