Make the biggest number in python

Question

I'm looking for a Python object which is guaranteed to compare greater than any given int. It should be portable, platform-independent and work on both Python 2.7+ and 3.x.

For example:

x = float('inf')
while True:
    n = next(my_gen)
    if my_calc(n):
        x = min(n, x)
        if my_cond(x):
            break

Here I've used float('inf') for this purpose because it seems to behave correctly. But this feels dirty, because I think it relies on some underlying float specification and I don't know whether that's going to be platform dependent or break in unexpected ways.

I'm aware that I could create my own class and define the comparison operators, but I thought there might be an existing built-in way.

Is it safe to use float('inf') like this? Is there a less ugly way of creating this "biggest integer"?

http://stackoverflow.com/questions/3477283/maximum-float-in-python might be of assistance to you — Leifingson, Dec 29 '13 at 03:38

Martijn Pieters · Accepted Answer · 2013-12-29T17:49:46.487

9

float('inf') is guaranteed to test as larger than any number, including integers. This is not platform specific.

From the floatobject.c source code:

else if (!Py_IS_FINITE(i)) {
    if (PyInt_Check(w) || PyLong_Check(w))
        /* If i is an infinity, its magnitude exceeds any
         * finite integer, so it doesn't matter which int we
         * compare i with.  If i is a NaN, similarly.
         */
        j = 0.0;

Python integers themselves are only bounded by memory, so using 10 ** 3000 is not going to be big enough, probably.

float('inf') is always available; Python will handle underlying platform specifics for you to make this so.

edited Dec 29 '13 at 17:49

answered Dec 29 '13 at 03:37

Martijn Pieters

889,049
245
3,507
2,997

Is that true in Python 3 as well? – Mark Ransom Dec 29 '13 at 03:39
@MartijnPieters Now that we are on the topic, is there something which is an exact opposite of this? I mean, a number which is lesser than any number in Python? – thefourtheye Dec 29 '13 at 03:41
1

@thefourtheye: `float('-inf')`, negative infinity. – Martijn Pieters Dec 29 '13 at 03:42
@MartijnPieters Thanks :) – thefourtheye Dec 29 '13 at 03:42

score 1 · Answer 2 · edited May 23 '17 at 12:03

1

Why don't just use:

x = float('inf')

instead of:

x = 1e3000

Read this post for more information.

edited May 23 '17 at 12:03

Community

1
1

answered Dec 29 '13 at 03:38

Christian

31,246
6
45
67

Well, my question still stands for `float('inf')` – wim Dec 29 '13 at 03:41
I have removed the distracting part about 1e3000 because this was not crucial to my question – wim Dec 29 '13 at 03:44

score 0 · Answer 3 · answered Dec 29 '13 at 11:03

In the following I remove the need for that first sentinel x value by using an outer while loop to capture the first valid x then using it on the preserved inner while loop:

while True:
    n = next(my_gen)
    if my_calc(n):
        x = n
        if my_cond(x):
            break
        else:
            while True:
                n = next(my_gen)
                if my_calc(n):
                    x = min(n, x)
                    if my_cond(x):
                        break  
            break

It is more code. Usually the removal of sentinel values is good but the above would have to be assessed for maintainability.

Further factoring of the code gives the following, but the code above preserves more of the original conditionals.

while True:
    n = next(my_gen)
    if my_calc(n):
        x = n
        if not my_cond(x):
            while True:
                n = next(my_gen)
                if my_calc(n):
                    x = min(n, x)
                    if my_cond(x):
                        break  
        break

Could be @wim, but sentinels values have their own detractors too. Think of how unsure you were, leading to you having to ask this question, and people having to check the Python source to ensure that a comparison between any int and float('inf') will work the way you want it to. I guess you could use either method and comment with a link to this SO question :-) — Paddy3118, Dec 29 '13 at 11:29
p.s. Not needing a sentinel makes the code Duck-typing friendly. If you defined your own type that was not comparable to floats but was returned by my_cond and my_calc then all would remain OK. — Paddy3118, Dec 29 '13 at 11:34

Make the biggest number in python

3 Answers3