Executes step given an action. Args ---- action : tuple Action to be executed. An action is a tuple int

Determines all parents and actions that lead to state. Args ---- state : list Representation of a state,

:returns: numpy array of atom positions of shape [num_atoms, 3]

Defines the structure of the output of the policy model, from which an action is to be determined or sampled, by returning a vector w

Returns a vector of length the action space + 1: True if forward action is invalid given the current state, False otherwise.

Resets the environment.

Prepares a list of states in "GFlowNet format" for the oracles: a list of length n_dim with values in the range [cell_min, cell_max]

assumes original dataset contains no duplicates :param samples: must be np array padded to equal length. If a combination of new and original

Prepares a state in "GFlowNet format" for the oracles: a list of length n_dim with values in the range [cell_min, cell_max] for each

Transforms a batch of states into a one-hot encoding. The output is a numpy array of shape [n_states, length * n_dim]. See s

Returns a vector of length the action space + 1: True if forward action is invalid given the current state, False otherwise.

Prepares the output of an oracle for GFlowNet: the inputs proxy_vals is expected to be a negative value (energy), unless self.denorm_

Builds a batch of data if train == True: Each item in the batch is a list of 7 elements (all tensors): -

Constructs list with all possible actions. An action is described by a tuple (element, n), indicating that the count of element will

Converts numerical values to ATGC-format :param sequences: numerical DNA sequences to be converted :return: DNA sequences in

Recursively converts the numpy types into native Python types in order to enable proper dumping into YAML files: Parameters --------

Creates a one-hot encoding vector. :param value: The value for which the encoding should be one :param choices: A list of pos

Converts a state into human-readable representation.

Converts a state (a list of positions) into a human-readable string representing a state.

Executes the actions of a list on the environments of a list. Args ---- envs : list of GFlowNetEnv or derived

compute simple sum of distances between sample vectors: distance = disagreement of allele elements. :param samples: :return:

:returns: a dict of all tostion angles in the molecule with their values

Uniformly randomize torsion angles defined by self.freely_rotatable_tas

Converts a human-readable string representing a state into a state as a list of positions.

Computes the rewards of a batch of states in "GFlownet format"

Samples one action on each environment of a list. Args ---- envs : list of GFlowNetEnv or derived A list

Prepares a batch of states in "GFlowNet format" for the proxy. Args ---- state : list A state

Helper that checks whether there is a configuration of oxidation states that can produce a neutral charge for the given state.

convert from (((...))) notation to 111222333

Download a file from google drive if path doestn't exist. url should be in the format: https://drive.google.com/uc?id=FILE_ID

Combines YAML configuration file, command line arguments and default arguments into a single configuration dictionary. - Values in YAML

Compute energy of a molecule defined by atom_positions and atomic_numbers

:returns: a list of values of self.freely_rotatable_tas

Kept as a function variable of GflowNetAgent because logq calculation requires tfloat and tbool member functions of GflowNet

Computes log probabilities of actions given policy outputs and actions.

Returns a vector with the length of the discrete part of the action space + 1: True if action is invalid going backward given the cur

Determines all parents and actions that lead to state. Args ---- state : list Representation of a state

Defines the structure of the output of the policy model, from which an action is to be determined or sampled, by returning a vector w

Get an index in the dgl graph of the first edge corresponding to the input torsion_angle :param torsion_angle: tuple of 4 integers de

:param torsion_angle: tuple of 4 integers defining the torsion angle :param increment: a float value of the increment of the angle (i

generate an initial toy dataset with a given number of samples need an extra factor to speed it up (duplicate filtering is very slow)

Constructs a data set as a DataFrame according to the configuration.

Converts rdkit.Chem.rdchem.Mol to dgl.heterograph.DGLHeteroGraph. Takes into account chemical properties of atoms and bonds, without

Recursively converts a dictionary and its internal dictionaries into an argparse.Namespace Parameters ---------- data_dict : dic

print a string to command line output and a text file :param statement: :return:

Sets the reward to min_reward if any value of the state is larger than max_val.

assign correct scores to selected sequences :param queries: sequences to be scored :return: computed scores

Set atom positions of the self.dgl_graph to the input atom_positions values :param atom_positions: 2d numpy array of shape [num atoms, 3] with

Sets the state and done of an environment.

Transforms the state given as argument (or self.state if None) into a one-hot encoding. The output is a list of len n_angles * n_dim,

Transforms the state given as argument (or self.state if None) into a one-hot encoding. The output is a list of len length * n_dim,