Uniformly random generation of unlabelled rooted trees.
More...
#include <rand_ulab_rooted_trees.hpp>
|
virtual void | init (uint32_t seed=0) noexcept |
| Sets the size of the unlabelled trees to generate.
|
|
std::pair< uint32_t, uint32_t > | ranrut (uint32_t n, uint32_t lr, uint32_t nt) noexcept |
| Generates uniformly at random a rooted unlabelled tree of n nodes.
|
|
void | init_rn () noexcept |
| Initialiases m_rn with 31 values from the OEIS (see [33]).
|
|
const numeric::integer & | get_rn (uint32_t n) noexcept |
| Computes all the values \(t_i\) for \(i \in [1,n]\).
|
|
std::pair< uint32_t, uint32_t > | choose_jd_from_T (uint32_t n) noexcept |
| Chooses uniformly at random a pair \((j,d)\), according to some probability.
|
|
|
const uint32_t | m_n |
| Number of nodes of the tree.
|
|
std::mt19937 | m_gen |
| Random number generator.
|
|
std::uniform_real_distribution< double > | m_unif |
| Distribution of the numbers.
|
|
std::vector< numeric::integer > | m_rn |
| The number of unlabelled rooted trees.
|
|
internal::data_array< uint32_t > | m_head_vector |
| The head vector of the tree under construction.
|
|
Uniformly random generation of unlabelled rooted trees.
Users should refrain from using this class. The generation of random labelled trees should be done using the wrapper class rand_ulab_rooted_trees. This class, however, contains the actual code to generate labelled rooted trees uniformly at random.
Every call to get_tree generates rooted unlabelled trees uniformly at random using the ranrut procedure (see [27], chapter 29).
◆ _rand_ulab_rooted_trees() [1/3]
lal::generate::_rand_ulab_rooted_trees::_rand_ulab_rooted_trees |
( |
uint32_t | n, |
|
|
uint32_t | seed = 0 ) |
|
noexcept |
Constructor with size of tree and seed for the random number generator.
In case the seed given is '0', a random seed will be generated.
- Parameters
-
n | Number of nodes. |
seed | The seed used for the random generator. If the seed is 0 then a random seed is generated and used. |
◆ _rand_ulab_rooted_trees() [2/3]
Copy constructor.
- Parameters
-
Gen | Random unlabelled rooted tree generator. |
◆ _rand_ulab_rooted_trees() [3/3]
Move constructor.
- Parameters
-
Gen | Random unlabelled rooted tree generator. |
◆ choose_jd_from_T()
std::pair< uint32_t, uint32_t > lal::generate::_rand_ulab_rooted_trees::choose_jd_from_T |
( |
uint32_t | n | ) |
|
|
protectednoexcept |
Chooses uniformly at random a pair \((j,d)\), according to some probability.
Probability of choosing \((j,d)\) is: \(\frac{d \cdot t_{k - jd} \cdot t_d}{(k - 1)t_k}\).
- Parameters
-
- Returns
- A pair of integers \((j,d)\) such that \(j \ge 1\), \(jd \le n\) and \(j \ge 1\), \(jd \le n\).
◆ clear()
void lal::generate::_rand_ulab_rooted_trees::clear |
( |
| ) |
|
|
noexcept |
Clears the memory occupied.
In order to save computation time, this class has been designed to reuse memory when generating trees. For example, since it needs the values of well-known integer sequences (see attribute m_rn) that are costly to compute every time they are needed, they are stored in memory and reused over time.
So, if the user wants to generate trees of 1000 nodes there will be too much memory occupied (and unused) if then this class is used to generate trees of 10 nodes. In cases like this it is recommended to clear the memory occupied.
- Postcondition
- After calling this method, the contents of the attributes m_rn are cleared. Attribute m_rn is then assigned the same 31 values that it is assigned when creating an object of this class.
◆ get_rn()
const numeric::integer & lal::generate::_rand_ulab_rooted_trees::get_rn |
( |
uint32_t | n | ) |
|
|
protectednoexcept |
Computes all the values \(t_i\) for \(i \in [1,n]\).
Here \(n\) is m_n. In case these values have already been calculated, this method does nothing.
◆ get_tree()
Generates uniformly at random a free unlabelled tree.
- Returns
- An unlabelled rooted tree. The tree is rooted at vertex 0.
◆ init()
virtual void lal::generate::_rand_ulab_rooted_trees::init |
( |
uint32_t | seed = 0 | ) |
|
|
protectedvirtualnoexcept |
Sets the size of the unlabelled trees to generate.
Initialises m_rn with 31 values are extracted from [33].
Initialises the random number generator with seed. When seed is 0, a random seed is used.
- Parameters
-
seed | Integer value used to seed the random number generator. |
Reimplemented in lal::generate::_rand_ulab_free_trees.
◆ ranrut()
std::pair< uint32_t, uint32_t > lal::generate::_rand_ulab_rooted_trees::ranrut |
( |
uint32_t | n, |
|
|
uint32_t | lr, |
|
|
uint32_t | nt ) |
|
protectednoexcept |
Generates uniformly at random a rooted unlabelled tree of n nodes.
The first call to this method should have lr = m_n + 1.
- Parameters
-
n | Number of nodes of the rooted tree to generate. |
lr | Pointer to the root of the last tree added. m_head_vector[lr] is the node that the root points to. |
nt | Index to m_head_vector where we have to place the new tree. |
- Returns
- Two indices: the index of the root of the last tree generated and where to store the next tree in m_head_vector.
◆ m_head_vector
internal::data_array<uint32_t> lal::generate::_rand_ulab_rooted_trees::m_head_vector |
|
protected |
The head vector of the tree under construction.
This list has n values for m_n nodes. The first position contains the root vertex.
Do not use its actual type (lal::head_vector) in an attempt to make memory usage a bit more efficient.
◆ m_rn
The number of unlabelled rooted trees.
Contains \(r_n\) for \(n\ge 0\).
The documentation for this class was generated from the following file: