Piecewise float/int <= bounds

[majidaldo]

This difference doesn't make sense.
When 1 is a float

>>> x = Symbol('x')
>>>Piecewise((123, x < 1), (456, x >= 1.0))
Piecewise((123, x < 1), (456, x >= 1.0))

but if 1 is an int

>>> Piecewise( (123, x<1), (456 , x>=1) )
Piecewise((123, x < 1), (456, True))

I think there is the 'otherwise' behavior here. If I put more (exprs, conds) after the second Piecewise, they don't show up.

>>>Piecewise( (123, x<1), (456 , x>=1) , (789, x>4) )
Piecewise((123, x < 1), (456, True))

It shows up using the float(1).

>>>Piecewise( (123, x<1), (456 , x>=1.) , (789, x>4) )
Piecewise( (123, x<1), (456 , x>=1.) , (789, x>4) )

[majidaldo]

..wait was i supposed to nest Piecewise?

[bjodah]

Looks like there is deduction going on (assuming x is real I suppose). Is this what you're looking for?:

>>> pw = sympy.Piecewise((123, x<1), (456, x<=4), (789, 4<x))
>>> pw
Piecewise((123, x < 1), (456, x <= 4), (789, x > 4))

then using e.g. lambdify you get:

>>> l = sympy.lambdify(x, pw)
>>> l(0.5)
array(123.)
>>> l(1.5)
array(456.)
>>> l(4.5)
array(789.)

[majidaldo]

why do i have to do it like that? i don't see the logical difference.

[asmeurer]

I wouldn't be surprised if this boils down to #11707.