IEEE Floats and Python

Everybody knows that "floating point numbers cause problems." Many people have misconceptions about what those problems are, and how to deal with them. There are some great references for people who want to understand things, the (justly) most famous being What Every Computer Scientist Should Know About Floating-Point Arithmetic by David Goldberg.

But many Python programmers like to learn by experimenting. That's why you've chosen a language with an interactive REPL, powerful introspection, and, when it comes down to it, a readable open-source implementation, right?

So, let's look at how floating point numbers are represented in Python; understanding what all the bits mean may make it easier to understand how rounding errors and so forth work.

On nearly any platform you care about, floating point numbers—both binary (float) and decimal (Decimal)—are going to be stored in IEEE 754-2008 formats. So, let's look at how those work.

IEEE binary floats

Scientific notation

IEEE floats are based on scientific notation, but in binary.

Scientific notation looks like:

-1.042e3

The "-" sign is obvious. The "1.042" is almost obvious, but note that in "normalized form" there always must be exactly one nonzero digit left of the "." (that's what makes it scientific notation), and you're allowed to have trailing zeros (if, e.g., you had 11 to 4 significant digits, you'd write 1.100e1, not 1.1e1). The "e3" means that you multiple by 10**3.

So:

- (1 + 0/10 + 4/100 + 2/1000) * 10**3 == -104.2

For binary scientific notation, everything works the same way, except with 2 in place of 10. So:

-1.01e2

- (1 + 0/2 + 1/4) * 2**2 = -101 (binary) or -5 (decimal)

Representation in bits

Obviously, any float based on scientific notation is going to have three things to store: the sign, the mantissa (the decimal 1.042 or binary 1.01), and the exponent (the 3 in 10**3 or 2 in 2**2).

The standard doesn't really specify how the bits have to be stored inside the computer, but they have to be represented for interchange as if they were stored as:

1 sign bit
w exponent bits
p-1 significand bits

Let's pick a specific, very small, size to make our lives easy: just 3 exponent bits and 2 significand bits. Of course in real life you'd never deal with 6-bit floats, but having only 64 possible values means we can look at an exhaustive table of them (which I've included below), which can be very helpful.

We'll call this 6-bit format "binary6", even though technically it isn't a legal interchange format. (For one thing, interchange formats must have either 16 bits, or a multiple of 32 other than 96. For another, a k-bit interchange format for any k but 16 and 32 must have w = round(4 * math.log2(k)) − 13, which in our case would give -3 bits.)

For the binary64 format, everything is exactly the same as our binary6 format, except that w is 11 instead of 3, and p-1 is 52 instead of 2.

Sign

The sign is easy. Unlike integers, there's no 2s complement or anything fancy going on here. You compute the number from the exponent and significand, and then if the sign bit is 1, you negate it. So, -5.0 looks exactly like 5.0 except for the first bit. This does have some strange consequences—it means -0.0 and +0.0 are different numbers, and it means the bits for the smallest negative number and the smallest positive number are very far apart—but it also makes a lot of things easier. So, in our example of -1.01e2, the sign bit is 1.

Mantissa and significand

Next, we've got the mantissa, the 1.01 part. But notice that we don't have 3 mantissa bits, we have 2 significand bits. How does this work? Well, in scientific notation, there's always exactly one non-zero digit to the left of the decimal point. In binary, the only non-zero digit is 1, so the format saves a bit by not storing that 1.

So, for 2 bits, the possible values are 1.00 (stored as 00), 1.01 (as 01), 1.10 (as 10), and 1.11 (as 11).

Exponent

Finally, we've got the exponent of 2. To handle the possibility of negative exponents, instead of storing the exponent itself, we store the biased exponent—that is, we add 2**(w-1)-1. Also, the smallest and largest values are reserved for special purposes, which we'll get to in a moment. For 3 bits, the possible values are special (stored as 000), -2 (as 001), -1 (as 010), 0 (as 011), 1 (as 100), 2 (as 101), 3 (as 110), and special (as 111).

So, our number, -1.01e2, is stored as 1 (sign) 101 (exponent) 01 (significand), or 110101.

Infinity, NaN, and subnormals

The smallest exponent value is used to store 0. After all, if there were no special rule, 0 000 00 would mean +1.00 * 2**-3, which is obviously not 0. But just saying exponent 0 means the value is 0 is a waste of bits—that would mean we have 8 different zero values (0 000 00, 0 000 01, 0 000 10, 0 000 11, 1 000 00, 1 000 01, 1 000 10, and 1 000 11). Having positive and negative zero can sometimes be useful, but having 4 different positive zeros is pointless.

So, instead, a 000 exponent actually means the same thing as a 001 exponent, except that a 0 instead of a 1 is prepended to the significand to get the mantissa. A number like 0.01e-2 isn't properly normalized scientific notation—that would be 1.00e-4. So, these numbers are called "subnormal" or "denormal".

Meanwhile, the largest exponent value is used to store infinity. Again, though, that would give us 8 infinities when we only need 2. So, if the significand is nonzero, instead of infinity, it's NaN. The first bit of the significand is the quiet bit, and the remaining bits (in this case, just 1) are the payload, which is reserved for anything a user-level library wants.

Table

So, in our 6-bit format, here's what all the possible numbers mean:

    sign exponent significand    binary scientific  decimal
       0      000          00    0.00e-2 = 0        0.0
       0      000          01    0.01e-2 = 1e-4     0.0625
       0      000          10    0.10e-2 = 1e-3     0.125
       0      000          11    0.11e-2 = 1.1e-3   0.1875
       0      001          00    1.00e-2            0.25
       0      001          01    1.01e-2            0.3125
       0      001          10    1.10e-2            0.375
       0      001          11    1.11e-2            0.4375
       0      010          00    1.00e-1            0.5
       0      010          01    1.01e-1            0.625
       0      010          10    1.10e-1            0.75
       0      010          11    1.11e-1            0.875
       0      011          00    1.00e0             1.
       0      011          01    1.01e0             1.25
       0      011          10    1.10e0             1.5
       0      011          11    1.11e0             1.75
       0      100          00    1.00e1             2.
       0      100          01    1.01e1             2.5
       0      100          10    1.10e1             3.
       0      100          11    1.11e1             3.5
       0      101          00    1.00e2             4.
       0      101          01    1.01e2             5.
       0      101          10    1.10e2             6.
       0      101          11    1.11e2             7
       0      110          00    1.00e3             8.
       0      110          01    1.01e3             10.
       0      110          10    1.10e3             12.
       0      110          11    1.11e3             14.
       0      111          00    inf                inf
       0      111          01    snan1              snan1
       0      111          10    qnan0              qnan0
       0      111          11    qnan1              qnan1
       1      000          00    -0.00e-2 = -0      -0.0
       1      000          01    -0.01e-2 = -1e-4   -0.0625
       1      000          10    -0.10e-2 = -1e-3   -0.125
       1      000          11    -0.11e-2 = -1.1e-3 -0.1875
       1      001          00    -1.00e-2           -0.25
       1      001          01    -1.01e-2           -0.3125
       1      001          10    -1.10e-2           -0.375
       1      001          11    -1.11e-2           -0.4375
       1      010          00    -1.00e-1           -0.5
     ...
       1      110          00    1.00e3             -8.
       1      110          01    1.01e3             -10.
       1      110          10    1.10e3             -12.
       1      110          11    1.11e3             -14.
       1      111          00    -inf               -inf
       1      111          01    -snan1             -snan1
       1      111          10    -qnan0             -qnan0
       1      111          11    -qnan1             -qnan1

Rounding

Given the table above, rounding errors should be pretty easy to understand.

If you want to store 1.1 as a binary6 float, the closest value is 1.0 (0 100 00), which is off by 0.1; the next best would be 1.25 (0 100 01), which is off by even more 0.15.

Now, let's say you want to test 0.6 / 4 == 0.15. The closest binary6 values to start with are 0.625 and 4. Divide those, and the closest result is 0.1875. But the closest value to 0.15 is 0.125, and 0.1875 != 0.125.

You can work out the same thing with more familiar examples with binary64, like 0.3 / 3 == 0.1.

Reading floats as ints

Notice that, as long as you stick to finite nonnegative numbers, you can treat the exact same bits as if they were integers, and two adjacent floats will always be adjacent integers.

For example, Python doesn't provide a function like C99/POSIX-2001 nextafter, but it's easy to implement yourself. For a quick&dirty case, if you know you're dealing with finite positive numbers, just add or subtract 1 as an integer, and you've got the next or previous float. Dealing with negative numbers and nonfinite numbers gets a bit trickier, but it should be obvious how to do it (once you choose exactly what you want to do for, e.g., nextafter(-0) and nextbefore(inf)).

Similarly, to get the "float difference" in ulps (Units of Least Precision, or Units in the Last Place), just subtract the two floats as integers (again, with the obvious extensions for negative and nonfinite numbers). Note that this is often very different from the delta, the difference in real numbers. For example, 1.25 is 2 ulp away from both 0.875 and 1.75, but the deltas are not the same (0.375 and 0.5, or relative 30% and 40%).

You can even begin to understand tricks like the fast inverse sqrt hack.

Manipulating float bits in Python

OK, now we know how a float is stored. But how can we play with this in Python?

While it's possible (at least in CPython) to get at the values under the covers via ctypes, it's generally simpler (and a lot safer) to just use the struct module to pack a float in big-endian C double format. Then we've got the bits as a sequence of bytes—or, if you prefer, you can turn it into a 64-bit unsigned integer, or use a third-party library like bitarray or bitstring to make your life a whole lot easier.

So, let's look at how -123.456 is stored as a float:

    >>> f = -123.456
    >>> b = bitstring.pack('>d', f)
    >>> sbit, wbits, pbits = b[:1], b[1:12], b[12:]
    >>> sbit.bin
    '1'

That's easy: sign bit 1 means negative.

    >>> wbits.bin
    '10000000101'
    >>> wbits.uint - (1<<10) + 1
    6

So, the exponent is (decimal) 6.

    >>> pbits.bin
    '1110110111010010111100011010100111111011111001110111'
    >>> 1 + pbits.uint / (1<<52)
    1.929

So, the mantissa is (decimal) 1.929.

So, the number is -1.929 * 2**6. Which checks out:

    >>> -1.929 * 2**6
    -123.456

Or, if you prefer:

    >>> pbits.hex
    'edd2f1a9fbe77'

So, in Python's float hex format, the number is -1.edd2f1a9fbe77p+6, which we can check:

    >>> f.hex()
    '-0x1.edd2f1a9fbe77p+6'

You can also use this to create and parse NaN payloads, which Python doesn't do natively:

    >>> quiet, payload = 1, 3
    >>> bits = (0b11111111111 << 52) | (quiet << 51) | payload
    >>> qnan3 = struct.unpack('>d', struct.pack('>Q', bits))[0]

Of course Python is going to print qnan3 as just plain "nan", but if your platform has some way of printing out non-default NaN values, you may be able to see it like this:

    >>> ctypes.printf(b'%f', ctypes.c_double(qnan3))
    QNaN3

And if you want to do the nextafter or float_difference hacks described in the last section:

    >>> f = 1.1
    >>> d = struct.unpack('>Q', struct.pack('>d', f))[0]
    >>> struct.unpack('>d', struct.pack('>Q', d-1))[0]
    1.0999999999999999
    >>> struct.unpack('>d', struct.pack('>Q', d+1))[0]
    1.1000000000000003

The floatextras package

Now that you know how to do this all manually, the floatextras package on PyPI wraps it all up for you:

    >>> f = 1.1
    >>> floatextras.next_plus(f)
    1.1000000000000003
    >>> floatextras.float_difference(1.0999999999999999, 1.1000000000000003)
    -2
    >>> floatextras.as_tuple(-123.456)
    FloatTuple(sign=1, digits=(1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1), exponent=6)
    >>> floatextras.as_tuple(qnan3)
    FloatTuple(sign=0, digits=(1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1), exponent=1024)

Caveat for some platforms

What I described above is all true for the current IEEE 754-2008 standard, but most platforms are actually based on the older IEEE 754-1987. There are a number of differences and expansions in areas I didn't get to above, and some of the terminology is different (e.g., "double" instead of "binary64").

Also, there are also a few things that the older standard left open to the implementation, which I've assumed will always be interpreted the same way. If you're on almost any platform on x86, x86_64, or ARM hardware, everything will be as above; on different hardware, there may be minor differences (e.g., Irix for MIPS used the NaN quiet/signaling bit backward, so 0 means quiet; another platform which I've mercifully forgotten used the last bit of the significand rather than the first; POWER stored floats big-endian even in little-endian mode so you had to flip the bits around to do the "treat them as ints" trick).

Decimal floats

There's no reason floats have to be stored in binary. In fact, IEEE 754-2008 specifies how you can use the same basic ideas with decimal fractions instead of binary. And Python includes an implementation of this (well, of its predecessor, IEEE 854-1987, but again, it's close enough) in the decimal module.

Most of the differences between binary and decimal formats are obvious, except for one. Decimals can't use the "free mantissa bit" digit, because the digit left of the decimal point can be anything from 1 through 9, so you have to store it. This also means decimal floats can have non-canonical representations (after all, 0.1e2 means the same thing as 1.0e1). Which also means you can store subnormal values, and 0, with different levels of precision, and some operations do so.

The decimal module docs, and the PEP linked from there, cover all of the details very nicely.

The obvious advantage of decimal floats is that there's no rounding error converting to and from decimal strings. For example, the closest binary64 float to 1.1 is 1.10000000000000009e+0; the closest decimal float (as long as you have at least 2 digits of precision) is 1.1e+0. Magic, right? Well, no; plenty of things that are exact with binary floats are inexact with decimal floats. Still, decimals are often a good solution to many floating-point problems, as long as you understand how they fall short of magic. Which, hopefully, you now do.

Also, because the decimal module is implemented from scratch, rather than relying on a native C type, it provides a lot more flexibility. Instead of being stuck with a decimal64 format, you can define and use any format. You can specify rounding modes and signaling modes. All of the optional APIs from the standard are available, instead of just the ones C/POSIX/your platform chose to implement, plus some extras to do things like picking a float apart into sign, exponent, and mantissa, or building NaN values with custom payloads.

It's been more than a decade since Typical Programmer Greg Jorgensen taught the word about Abject-Oriented Programming.

Much of what he said still applies, but other things have changed.

I haven't posted anything new in a couple years (partly because I attempted to move to a different blogging platform where I could write everything in markdown instead of HTML but got frustrated—which I may attempt again), but I've had a few private comments and emails on some of the old posts, so I

Looking before you leap

Python is a duck-typed language, and one where you usually trust EAFP ("Easier to Ask Forgiveness than Permission") over LBYL ("Look Before You Leap").

Background

Currently, CPython’s internal bytecode format stores instructions with no args as 1 byte, instructions with small args as 3 bytes, and instructions with large args as 6 bytes (actually, a 3-byte EXTENDED_ARG followed by a 3-byte real instruction).

If you want to skip all the tl;dr and cut to the chase, jump to Concrete Proposal.

Many people, when they first discover the heapq module, have two questions:

Why does it define a bunch of functions instead of a container type? Why don't those functions take a key or reverse parameter, like all the other sorting-related stuff in Python? Why not a type?

At the abstract level, it'

Currently, in CPython, if you want to process bytecode, either in C or in Python, it’s pretty complicated.

The built-in peephole optimizer has to do extra work fixing up jump targets and the line-number table, and just punts on many cases because they’re too hard to deal with.

One common "advanced question" on places like StackOverflow and python-list is "how do I dynamically create a function/method/class/whatever"? The standard answer is: first, some caveats about why you probably don't want to do that, and then an explanation of the various ways to do it when you reall

A few years ago, Cesare di Mauro created a project called WPython, a fork of CPython 2.6.4 that “brings many optimizations and refactorings”. The starting point of the project was replacing the bytecode with “wordcode”. However, there were a number of other changes on top of it.

Many languages have a for-each loop.

When the first betas for Swift came out, I was impressed by their collection design. In particular, the way it allows them to write map-style functions that are lazy (like Python 3), but still as full-featured as possible.

In a previous post, I explained in detail how lookup works in Python.

The documentation does a great job explaining how things normally get looked up, and how you can hook them.

But to understand how the hooking works, you need to go under the covers to see how that normal lookup actually happens.

When I say "Python" below, I'm mostly talking about CPython 3.5.

In Python (I'm mostly talking about CPython here, but other implementations do similar things), when you write the following:

def spam(x): return x+1 spam(3) What happens?

Really, it's not that complicated, but there's no documentation anywhere that puts it all together.

I've seen a number of people ask why, if you can have arbitrary-sized integers that do everything exactly, you can't do the same thing with floats, avoiding all the rounding problems that they keep running into.

In a recent thread on python-ideas, Stephan Sahm suggested, in effect, changing the method resolution order (MRO) from C3-linearization to a simple depth-first search a la old-school Python or C++.

Note: This post doesn't talk about Python that much, except as a point of comparison for JavaScript.

Most object-oriented languages out there, including Python, are class-based. But JavaScript is instead prototype-based.

About a year and a half ago, I wrote a blog post on the idea of adding pattern matching to Python.

I finally got around to playing with Scala semi-seriously, and I realized that they pretty much solved the same problem, in a pretty similar way to my straw man proposal, and it works great.

About a year ago, Jules Jacobs wrote a series (part 1 and part 2, with part 3 still forthcoming) on the best collections library design.

In three separate discussions on the Python mailing lists this month, people have objected to some design because it leaks something into the enclosing scope. But "leaks into the enclosing scope" isn't a real problem.

There's a lot of confusion about what the various kinds of things you can iterate over in Python. I'll attempt to collect definitions for all of the relevant terms, and provide examples, here, so I don't have to go over the same discussions in the same circles every time.

Python has a whole hierarchy of collection-related abstract types, described in the collections.abc module in the standard library. But there are two key, prototypical kinds. Iterators are one-shot, used for a single forward traversal, and usually lazy, generating each value on the fly as requested.

There are a lot of novice questions on optimizing NumPy code on StackOverflow, that make a lot of the same mistakes. I'll try to cover them all here.

What does NumPy speed up?

Let's look at some Python code that does some computation element-wise on two lists of lists.

When asyncio was first proposed, many people (not so much on python-ideas, where Guido first suggested it, but on external blogs) had the same reaction: Doing the core reactor loop in Python is going to be way too slow. Something based on libev, like gevent, is inherently going to be much faster.

Let's say you have a good idea for a change to Python.

There are hundreds of questions on StackOverflow that all ask variations of the same thing. Paraphrasing:

lst is a list of strings and numbers. I want to convert the numbers to int but leave the strings alone.

In Haskell, you can section infix operators. This is a simple form of partial evaluation. Using Python syntax, the following are equivalent:

(2*) lambda x: 2*x (*2) lambda x: x*2 (*) lambda x, y: x*y So, can we do the same in Python?

Grammar

The first form, (2*), is unambiguous.

Many people—especially people coming from Java—think that using try/except is "inelegant", or "inefficient". Or, slightly less meaninglessly, they think that "exceptions should only be for errors, not for normal flow control".

These people are not going to be happy with Python.

If you look at Python tutorials and sample code, proposals for new language features, blogs like this one, talks at PyCon, etc., you'll see spam, eggs, gouda, etc. all over the place.

Most control structures in most most programming languages, including Python, are subordinating conjunctions, like "if", "while", and "except", although "with" is a preposition, and "for" is a preposition used strangely (although not as strangely as in C…).

There are two ways that some Python programmers overuse lambda. Doing this almost always mkes your code less readable, and for no corresponding benefit.

Some languages have a very strong idiomatic style—in Python, Haskell, or Swift, the same code by two different programmers is likely to look a lot more similar than in Perl, Lisp, or C++.

There's an advantage to this—and, in particular, an advantage to you sticking to those idioms.

Python doesn't have a way to clone generators.

At least for a lot of simple cases, however, it's pretty obvious what cloning them should do, and being able to do so would be handy. But for a lot of other cases, it's not at all obvious.

Every time someone has a good idea, they believe it should be in the stdlib. After all, it's useful to many people, and what's the harm? But of course there is a harm.

This confuses every Python developer the first time they see it—even if they're pretty experienced by the time they see it:

>>> t = ([], []) >>> t[0] += [1] --------------------------------------------------------------------------- TypeError Traceback (most recent call last) <stdin> in <module>()