class QRandomGenerator#

The QRandomGenerator class allows one to obtain random values from a high-quality Random Number Generator. More

Inherited by: QRandomGenerator64

Synopsis#

Methods#

Static functions#

Note

This documentation may contain snippets that were automatically translated from C++ to Python. We always welcome contributions to the snippet translation. If you see an issue with the translation, you can also let us know by creating a ticket on https:/bugreports.qt.io/projects/PYSIDE

Detailed Description#

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

QRandomGenerator may be used to generate random values from a high-quality random number generator. Like the C++ random engines, QRandomGenerator can be seeded with user-provided values through the constructor. When seeded, the sequence of numbers generated by this class is deterministic. That is to say, given the same seed data, QRandomGenerator will generate the same sequence of numbers. But given different seeds, the results should be considerably different.

securelySeeded() can be used to create a QRandomGenerator that is securely seeded with system() , meaning that the sequence of numbers it generates cannot be easily predicted. Additionally, global() returns a global instance of QRandomGenerator that Qt will ensure to be securely seeded. This object is thread-safe, may be shared for most uses, and is always seeded from system()

system() may be used to access the system’s cryptographically-safe random generator. On Unix systems, it’s equivalent to reading from /dev/urandom or the getrandom() or getentropy() system calls.

The class can generate 32-bit or 64-bit quantities, or fill an array of those. The most common way of generating new values is to call the generate() , generate64() or fillRange() functions. One would use it as:

value = QRandomGenerator.global().generate()

Additionally, it provides a floating-point function generateDouble() that returns a number in the range [0, 1) (that is, inclusive of zero and exclusive of 1). There’s also a set of convenience functions that facilitate obtaining a random number in a bounded, integral range.

Seeding and determinism#

QRandomGenerator may be seeded with specific seed data. When that is done, the numbers generated by the object will always be the same, as in the following example:

prng1 = QRandomGenerator(1234), prng2(1234)
Q_ASSERT(prng1.generate() == prng2.generate())
Q_ASSERT(prng1.generate64() == prng2.generate64())

The seed data takes the form of one or more 32-bit words. The ideal seed size is approximately equal to the size of the QRandomGenerator class itself. Due to mixing of the seed data, QRandomGenerator cannot guarantee that distinct seeds will produce different sequences.

global() , like all generators created by securelySeeded() , is always seeded from system() , so it’s not possible to make it produce identical sequences.

Bulk data#

When operating in deterministic mode, QRandomGenerator may be used for bulk data generation. In fact, applications that do not need cryptographically-secure or true random data are advised to use a regular QRandomGenerator instead of system() for their random data needs.

For ease of use, QRandomGenerator provides a global object that can be easily used, as in the following example:

x = QRandomGenerator.global().generate()
y = QRandomGenerator.global().generate()
w = QRandomGenerator.global().bounded(16384)
h = QRandomGenerator.global().bounded(16384)

System-wide random number generator#

system() may be used to access the system-wide random number generator, which is cryptographically-safe on all systems that Qt runs on. This function will use hardware facilities to generate random numbers where available. On such systems, those facilities are true Random Number Generators. However, if they are true RNGs, those facilities have finite entropy sources and thus may fail to produce any results if their entropy pool is exhausted.

If that happens, first the operating system then QRandomGenerator will fall back to Pseudo Random Number Generators of decreasing qualities (Qt’s fallback generator being the simplest). Whether those generators are still of cryptographic quality is implementation-defined. Therefore, system() should not be used for high-frequency random number generation, lest the entropy pool become empty. As a rule of thumb, this class should not be called upon to generate more than a kilobyte per second of random data (note: this may vary from system to system).

If an application needs true RNG data in bulk, it should use the operating system facilities (such as /dev/random on Linux) directly and wait for entropy to become available. If the application requires PRNG engines of cryptographic quality but not of true randomness, system() may still be used (see section below).

If neither a true RNG nor a cryptographically secure PRNG are required, applications should instead use PRNG engines like QRandomGenerator ‘s deterministic mode and those from the C++ Standard Library. system() can be used to seed those.

Fallback quality#

system() uses the operating system facilities to obtain random numbers, which attempt to collect real entropy from the surrounding environment to produce true random numbers. However, it’s possible that the entropy pool becomes exhausted, in which case the operating system will fall back to a pseudo-random engine for a time. Under no circumstances will system() block, waiting for more entropy to be collected.

The following operating systems guarantee that the results from their random-generation API will be of at least cryptographically-safe quality, even if the entropy pool is exhausted: Apple OSes (Darwin), BSDs, Linux, Windows. Barring a system installation problem (such as /dev/urandom not being readable by the current process), system() will therefore have the same guarantees.

On other operating systems, QRandomGenerator will fall back to a PRNG of good numeric distribution, but it cannot guarantee proper seeding in all cases. Please consult the OS documentation for more information.

Applications that require QRandomGenerator not to fall back to non-cryptographic quality generators are advised to check their operating system documentation or restrict their deployment to one of the above.

Reentrancy and thread-safety#

QRandomGenerator is reentrant, meaning that multiple threads can operate on this class at the same time, so long as they operate on different objects. If multiple threads need to share one PRNG sequence, external locking by a mutex is required.

The exceptions are the objects returned by global() and system() : those objects are thread-safe and may be used by any thread without external locking. Note that thread-safety does not extend to copying those objects: they should always be used by reference.

Standard C++ Library compatibility#

QRandomGenerator is modeled after the requirements for random number engines in the C++ Standard Library and may be used in almost all contexts that the Standard Library engines can. Exceptions to the requirements are the following:

  • QRandomGenerator does not support seeding from another seed sequence-like class besides std::seed_seq itself;

  • QRandomGenerator is not comparable (but is copyable) or streamable to std::ostream or from std::istream.

QRandomGenerator is also compatible with the uniform distribution classes std::uniform_int_distribution and std:uniform_real_distribution, as well as the free function std::generate_canonical. For example, the following code may be used to generate a floating-point number in the range [1, 2.5):

std.uniform_real_distribution dist(1, 2.5)
return dist(QRandomGenerator.global())
__init__(begin, end)#
Parameters:
  • beginquint32

  • endquint32

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

This is an overloaded function.

Initializes this QRandomGenerator object with the values found in the range from begin to end as the seed. Two objects constructed or reseeded with the same seed value will produce the same number sequence.

This constructor is equivalent to:

std.seed_seq sseq(begin, end)
generator = QRandomGenerator(sseq)
__init__(seedBuffer, len)
Parameters:
  • seedBufferquint32

  • len – int

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

This is an overloaded function.

Initializes this QRandomGenerator object with len values found in the array seedBuffer as the seed. Two objects constructed or reseeded with the same seed value will produce the same number sequence.

This constructor is equivalent to:

std.seed_seq sseq(seedBuffer, seedBuffer + len)
generator = QRandomGenerator(sseq)
__init__([seedValue=1])
Parameters:

seedValue – int

Initializes this QRandomGenerator object with the value seedValue as the seed. Two objects constructed or reseeded with the same seed value will produce the same number sequence.

__init__(other)
Parameters:

otherQRandomGenerator

Creates a copy of the generator state in the other object. If other is system() or a copy of that, this object will also read from the operating system random-generating facilities. In that case, the sequences generated by the two objects will be different.

In all other cases, the new QRandomGenerator object will start at the same position in the deterministic sequence as the other object was. Both objects will generate the same sequence from this point on.

For that reason, it is not advisable to create a copy of global() . If one needs an exclusive deterministic generator, consider instead using securelySeeded() to obtain a new object that shares no relationship with the global() .

bounded(lowest, highest)#
Parameters:
  • lowest – int

  • highest – int

Return type:

int

This is an overloaded function.

Generates one random 64-bit quantity in the range between lowest (inclusive) and highest (exclusive). The highest parameter must be greater than lowest.

The same result may also be obtained by using std::uniform_int_distribution<quint64> ` <http://en.cppreference.com/w/cpp/numeric/random/uniform_int_distribution>`_ with parameters lowest and \a highest - 1.

Note that this function cannot be used to obtain values in the full 64-bit range of quint64. Instead, use generate64() .

Note

This function is implemented as a loop, which depends on the random value obtained. On the long run, on average it should loop just under 2 times, but if the random generator is defective, this function may take considerably longer to execute.

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

bounded(highest)
Parameters:

highest – int

Return type:

int

This is an overloaded function.

Generates one random 64-bit quantity in the range between 0 (inclusive) and highest (exclusive). The same result may also be obtained by using std::uniform_int_distribution<quint64> ` <http://en.cppreference.com/w/cpp/numeric/random/uniform_int_distribution>`_ with parameters 0 and highest - 1.

Note that this function cannot be used to obtain values in the full 64-bit range of quint64. Instead, use generate64() .

Note

This function is implemented as a loop, which depends on the random value obtained. On the long run, on average it should loop just under 2 times, but if the random generator is defective, this function may take considerably longer to execute.

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

This is an overloaded function.

Generates one random 32-bit quantity in the range between lowest (inclusive) and highest (exclusive). The highest parameter must be greater than lowest.

The same result may also be obtained by using std::uniform_int_distribution ` <http://en.cppreference.com/w/cpp/numeric/random/uniform_int_distribution>`_ with parameters lowest and \a highest - 1. That class can also be used to obtain quantities larger than 32 bits.

For example, to obtain a value between 1000 (incl.) and 2000 (excl.), one would write:

v = QRandomGenerator.global().bounded(1000, 2000)

Note that this function cannot be used to obtain values in the full 32-bit range of quint32. Instead, use generate() .

bounded(highest)
Parameters:

highest – int

Return type:

int

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

This is an overloaded function.

Generates one random 32-bit quantity in the range between 0 (inclusive) and highest (exclusive). The same result may also be obtained by using std::uniform_int_distribution ` <http://en.cppreference.com/w/cpp/numeric/random/uniform_int_distribution>`_ with parameters 0 and highest - 1. That class can also be used to obtain quantities larger than 32 bits; for 64 bits, the 64-bit bounded() overload can be used too.

For example, to obtain a value between 0 and 255 (inclusive), one would write:

v = QRandomGenerator.global().bounded(256)

Naturally, the same could also be obtained by masking the result of generate() to only the lower 8 bits. Either solution is as efficient.

Note that this function cannot be used to obtain values in the full 32-bit range of quint32. Instead, use generate() .

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

bounded(highest)
Parameters:

highest – int

Return type:

int

This is an overloaded function.

Generates one random 64-bit quantity in the range between 0 (inclusive) and highest (exclusive). highest must be positive.

Note that this function cannot be used to obtain values in the full 64-bit range of qint64. Instead, use generate64() and cast to qint64 or instead use the unsigned version of this function.

Note

This function is implemented as a loop, which depends on the random value obtained. On the long run, on average it should loop just under 2 times, but if the random generator is defective, this function may take considerably longer to execute.

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

This is an overloaded function.

Generates one random 32-bit quantity in the range between lowest (inclusive) and highest (exclusive), both of which may be negative, but highest must be greater than lowest.

Note that this function cannot be used to obtain values in the full 32-bit range of int. Instead, use generate() and cast to int.

bounded(highest)
Parameters:

highest – int

Return type:

int

This is an overloaded function.

Generates one random 32-bit quantity in the range between 0 (inclusive) and highest (exclusive). highest must be positive.

Note that this function cannot be used to obtain values in the full 32-bit range of int. Instead, use generate() and cast to int.

bounded(highest)
Parameters:

highest – float

Return type:

float

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

Generates one random double in the range between 0 (inclusive) and highest (exclusive). This function is equivalent to and is implemented as:

return generateDouble() * highest

If the highest parameter is negative, the result will be negative too; if it is infinite or NaN, the result will be infinite or NaN too (that is, not random).

bounded(lowest, highest)
Parameters:
  • lowest – int

  • highest – int

Return type:

int

This is an overloaded function.

Generates one random 64-bit quantity in the range between lowest (inclusive) and highest (exclusive), both of which may be negative, but highest must be greater than lowest.

Note that this function cannot be used to obtain values in the full 64-bit range of qint64. Instead, use generate64() and cast to qint64.

Note

This function is implemented as a loop, which depends on the random value obtained. On the long run, on average it should loop just under 2 times, but if the random generator is defective, this function may take considerably longer to execute.

discard(z)#
Parameters:

z – int

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

Discards the next z entries from the sequence. This method is equivalent to calling generate() z times and discarding the result, as in:

while z--:
    generator.generate()
generate()#
Return type:

int

Generates a 32-bit random quantity and returns it.

See also

operator()() generate64()

generate64()#
Return type:

int

Generates a 64-bit random quantity and returns it.

See also

operator()() generate()

generateDouble()#
Return type:

float

Warning

This section contains snippets that were automatically translated from C++ to Python and may contain errors.

Generates one random qreal in the canonical range [0, 1) (that is, inclusive of zero and exclusive of 1).

This function is equivalent to:

rd = QRandomGenerator64()
return std.generate_canonical<qreal, std.numeric_limits<qreal>.digits>(rd)

The same may also be obtained by using std::uniform_real_distribution ` <http://en.cppreference.com/w/cpp/numeric/random/uniform_real_distribution>`_ with parameters 0 and 1.

static global_()#
Return type:

QRandomGenerator

static max()#
Return type:

int

Returns the maximum value that QRandomGenerator may ever generate. That is, std::numeric_limits<result_type>::max().

See also

min() max()

static min()#
Return type:

int

Returns the minimum value that QRandomGenerator may ever generate. That is, 0.

See also

max() min()

__ne__(rng2)#
Parameters:

rng2QRandomGenerator

Return type:

bool

Returns true if the two engines rng1 and rng2 are at different states or if one of them is reading from the operating system facilities and the other is not, false otherwise.

static securelySeeded()#
Return type:

QRandomGenerator

Returns a new QRandomGenerator object that was securely seeded with system() . This function will obtain the ideal seed size for the algorithm that QRandomGenerator uses and is therefore the recommended way for creating a new QRandomGenerator object that will be kept for some time.

Given the amount of data required to securely seed the deterministic engine, this function is somewhat expensive and should not be used for short-term uses of QRandomGenerator (using it to generate fewer than 2600 bytes of random data is effectively a waste of resources). If the use doesn’t require that much data, consider using global() and not storing a QRandomGenerator object instead.

See also

global() system()

seed([s=1])#
Parameters:

s – int

Reseeds this object using the value seed as the seed.

static system()#
Return type:

QRandomGenerator

Returns a pointer to a shared QRandomGenerator that always uses the facilities provided by the operating system to generate random numbers. The system facilities are considered to be cryptographically safe on at least the following operating systems: Apple OSes (Darwin), BSDs, Linux, Windows. That may also be the case on other operating systems.

They are also possibly backed by a true hardware random number generator. For that reason, the QRandomGenerator returned by this function should not be used for bulk data generation. Instead, use it to seed QRandomGenerator or a random engine from the <random> header.

The object returned by this function is thread-safe and may be used in any thread without locks. It may also be copied and the resulting QRandomGenerator will also access the operating system facilities, but they will not generate the same sequence.

See also

securelySeeded() global()