cyrxnopt package

Submodules

cyrxnopt.NestedVenv module

class cyrxnopt.NestedVenv.NestedVenv(virtual_dir: str | Path)

Bases: EnvBuilder

activate() → None

Activates the current virtual environment as the primary virtual environment. If the venv is active but not primary, it will be reactivated as the primary venv.

Raises:

RuntimeError – The virtual environment does not exist.

property binary_directory: Path

The venv subdirectory containing binaries based on operating system.

Returns:

Full path to the venv binary directory

Return type:

Path

check_package(package: str, version: str = '') → bool

create(env_dir: Any = '') → None

Creates the virtual environment at the given location.

Parameters:

env_dir (AnyPath) – Desired venv directory

deactivate() → None

Deactivates the virtual environment regardless of if it is the primary virtual environment.

delete() → None

is_active() → bool

Checks if the virtual environment is active or not.

This checks if the virtual environment path is in the PATH environment variable or not.

Returns:

Whether the venv is active (True) or not (False).

Return type:

bool

is_primary() → bool

Checks if the virtual environment is the primary active virtual environment.

This checks if the virtual environment path is the first entry in the PATH environment variable. This menas its packages will be found first.

TODO: Recognize other virtual environments to ensure we are

the first virtual environment without needing to be the first element in the PATH environment variable.

Returns:

Whether the venv is primary (True) or not (False).

Return type:

bool

pip_freeze() → list[str]

Returns the list of modules in the virtual environment as they would be returned by ‘pip freeze’.

Raises:

CalledProcessError – An error occurred when running pip freeze

pip_install(package_name: str, package_path: Path | None = None, editable: bool = False) → None

Install a package to the active virtual environment using pip install for an editable install.

Parameters:

• package_name (str) – Name of the package

• package_path (Path) – Path to the package location

• editable (bool) – Whether to use an editable install

Raises:

CalledProcessError – An error occurred when running pip freeze

pip_install_e(package_path: Path, package_name: str = '') → None

Install a package to the active virtual environment using pip install for an editable install.

Parameters:

• package_path (Path) – Path to the package location

• package_name (str, optional) – Name of the package, defaults to “”. If not provided, the package name is assumed to the the last part of package_path.

Raises:

CalledProcessError – An error occurred when running pip install

pip_install_r(req_file: Path) → None

Installs package requirements from a “requirements.txt”-style file.

Parameters:

req_file (str) – Requirements file to use

Raises:

CalledProcessError – An error occurred when running pip install for a package

property prefix: Path

The prefix directory for this venv.

Returns:

Full path to the prefix directory for this venv

Return type:

Path

property python: Path

The python binary of the venv based on operatingsystem.

Returns:

Full path to the Python binary of the venv

Return type:

Path

property site_packages: Path

cyrxnopt.OptimizerABC module

class cyrxnopt.OptimizerABC.OptimizerABC(venv: NestedVenv)

Bases: ABC

This is the abstract class for general optimizer algorithms which include all the abstract functions need.

check_install() → bool

Check if an installation for this optimizer exists or not.

Returns:

Whether the optimizer is installed (True) or not (False).

Return type:

bool

property dependencies: list[str]

Dependencies required by this optimizer.

Returns:

Dependency package names

Return type:

list[str]

abstractmethod get_config() → list[dict[str, Any]]

Provides descriptions for valid configuration settings for an optimizer.

This method provides information about the possible configuration options and their default values for a given optimizer. The intent is for this method to provide the necessary information for a user-facing program to present the configuration options to the user. Once the user has chosen their desired configurations, they can be set for the optimizer using set_config().

The configuration descriptions returned by this function are dictionaries with three keys, “name”, “type”, and “value”:

• “name” will contain the name for the config option in snake_case, with “continuous” or “categorical” prepended to the name if two versions of the option are needed for continuous and categorical variables.

• “type” will contain the Python type annotation describing the type to use for this option (for translating to strictly typed languages)

• “value” will provide the default value of the option.

• “range” is optional and used to specify the allowed range for numbers or to constrain what options a string input can accept. For example, a description of an option for the optimization direction could be:

  {
      "name": "optimization_direction",
      "type": str,
      "value": "min",
      "range": ["min", "max"]
  }
  

• “description” is an optional description of the purpose of a config option, along with any caveats that may come with it.

install(local_paths: dict[str, str] = {}) → None

Install the optimizer and its dependencies.

The list of packages to be installed can be checked with the OptimizerABC.dependencies() property.

Parameters:

local_paths (dict[str, str], optional) – Mapping of package names to local paths to the packages to be installed. The package names in the mapping must match a name returned by OptimizerABC.dependencies(). Defaults to {}

abstractmethod predict(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable | None = None) → list[Any]

Abstract optimizer prediction function.

Parameters:

• prev_param (list[Any]) – Previous suggested reaction conditions

• yield_value (float) – Yield value from previous reaction conditions

• experiment_dir (str) – Output directory for the current experiment

• config (dict[str, Any]) – Optimizer configuration

• obj_func (Optional[Callable], optional) – Objective function to optimize, defaults to None

Returns:

The next suggested conditions to perform

Return type:

list[Any]

abstractmethod set_config(experiment_dir: str, config: dict[str, Any]) → None

Generates initial configurations and files for an optimizer.

Valid configuration options should be retrieved using get_config() before calling this function. The key for each option must match the corresponding “name” field and the value type must match the one assigned in the “type” field from the optimizer’s get_config() method.

Parameters:

• experiment_dir (str) – Experimental directory for saving data files.

• config (dict[str, Any]) – Configuration settings defined from get_config().

abstractmethod train(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Abstract optimizer training function.

Parameters:

• prev_param (list[Any]) – Parameters provided from the previous prediction or training step.

• yield_value (float) – Result from the previous prediction or training step.

• experiment_dir (str) – Output directory for the optimizer algorithm.

• config (dict[str, Any]) – Optimizer config

• obj_func (Optional[Callable[..., float]], optional) – Objective function to optimize, defaults to None

Returns:

The next suggested reaction to perform

Return type:

list[Any]

cyrxnopt.OptimizerAmlro module

class cyrxnopt.OptimizerAmlro.OptimizerAmlro(venv: NestedVenv)

Bases: OptimizerABC

get_config() → list[dict[str, Any]]

Gets the configuration options available for this optimizer.

See OptimizerABC.get_config() for more information about the config descriptions returned by this method and for general usage information.

Returns:

Configuration option descriptions

Return type:

list[dict[str, Any]]

predict(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Searches for the best parameters and records results from prior steps.

OptimizerAmlro.set_config() and OptimizerAmlro.train() must be called prior to this method to generate the necessary files and initial training data for the model.

The previous parameter+result is recorded during the next call to predict(). The prev_param and yield_value values provided here are always recorded. Importantly, this also means that the last suggested parameter+result pair from prediction will not be recorded unless either another predict() call is made afterward!

Parameters:

• prev_param (list[Any]) – Parameters provided from the previous prediction or from the final call to OptimizerAmlro.train()

• yield_value (float) – Result from the previous suggested conditions

• experiment_dir (str) – Output directory for saving data files

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

• obj_func (Optional[Callable[..., float]], optional) – Ignored for this optimizer, defaults to None

Returns:

The next suggested reaction to perform

Return type:

list[Any]

set_config(experiment_dir: str, config: dict[str, Any]) → None

Generates necessary data files based on the given config.

See OptimizerABC.set_config() for more information about how to form the config dictionary and for general usage information.

Parameters:

• experiment_dir (str) – Output directory for generated files

• config (dict[str, Any]) – Configuration options for this optimizer instance

train(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Suggests and records training data points needed for AMLRO.

OptimizerAmlro.set_config() must be called prior to this method to generate the necessary files.

The previous parameter+result is recorded during the next call to either train() or predict(). This means that on the first call here, the prev_param and yield_value values provided are ignored. Importantly, this also means that the last suggested parameter+result pair from training will not be recorded unless either another train() or a subsequent predict() call are made afterward!

Parameters:

• prev_param (list[Any]) – Experimental parameter combination from the previous experiment, provide an empty list for the first call

• yield_value (float) – Experimental yield

• experiment_dir (str) – Output directory for saving data files

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

• obj_func (Optional[Callable[..., float]], optional) – Ignored for this optimizer, defaults to None

Returns:

Next parameter combination to perform, or an empty list ([]) if all training points have been performed

Return type:

list[Any]

cyrxnopt.OptimizerController module

cyrxnopt.OptimizerController.check_install(optimizer_name: str, venv: NestedVenv) → bool

Checks if an optimizer is installed in the given environment.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

Returns:

Whether the optimizer is installed (True) or not (False)

Return type:

bool

cyrxnopt.OptimizerController.get_config(optimizer_name: str, venv: NestedVenv) → list[dict[str, Any]]

Gets the description of the options available for an optimizer.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

Returns:

Descriptions for valid configuration values of the optimizer.

Return type:

list[dict[str, Any]]

cyrxnopt.OptimizerController.get_optimizer(optimizer_name: str, venv: NestedVenv) → OptimizerABC

Gets an instance of the requested optimizer algorithm

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

Raises:

RuntimeError – Invalid optimizer name given

Returns:

Requested optimizer

Return type:

OptimizerABC

cyrxnopt.OptimizerController.install(optimizer_name: str, venv: NestedVenv, local_paths: dict[str, str] = {}) → None

Installs an optimizer into the given environment.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

• local_paths (dict[str, str], optional) – Mapping of package names to local paths to the packages to be installed, defaults to {}

cyrxnopt.OptimizerController.predict(optimizer_name: str, venv: NestedVenv, prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable | None = None) → list[Any]

Predicts new reaction conditions using the given optimizer.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

• prev_param (list[Any]) – experimental parameter combination for previous experiment

• yield_value (float) – experimental yield

• experiment_dir (str) – experimental directory for saving data files

• config (dict[str, Any]) – Initial reaction feature configurations

• obj_func (Optional[Callable], optional) – Objective function needed to optimize, defaults to None

Returns:

Next suggested reaction conditions

Return type:

list[Any]

cyrxnopt.OptimizerController.set_config(optimizer_name: str, venv: NestedVenv, config: dict[str, Any], experiment_dir: str) → None

Sets the provided options for the given optimizer.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

• config (dict[str, Any]) – Desired optimizer configuration

• experiment_dir (str) – Directory to be used for the current experiment. This is where the config files will be output.

cyrxnopt.OptimizerController.train(optimizer_name: str, venv: NestedVenv, prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable | None = None) → list[Any]

Predicts new reaction conditions using the given optimizer.

Parameters:

• optimizer_name (str) – Name of the optimizer algorithm

• venv (NestedVenv) – Environment containing the optimizer installation

• prev_param (list[Any]) – Previous suggested reaction conditions

• yield_value (float) – Yield value from previous reaction conditions

• experiment_dir (str) – Output directory for the current experiment

• config (dict[str, Any]) – Optimizer configuration

• obj_func (Optional[Callable], optional) – Objective function to optimize, defaults to None

Returns:

The next suggested conditions to perform

Return type:

list[Any]

cyrxnopt.OptimizerEDBOp module

class cyrxnopt.OptimizerEDBOp.OptimizerEDBOp(venv: NestedVenv)

Bases: OptimizerABC

get_config() → list[dict[str, Any]]

Get the configuration options available for this optimizer.

See OptimizerABC.get_config() for more information about the config descriptions returned by this method and for general usage information.

Returns:

List of configuration options with option name, data type, and information about which values are allowed/defaulted.

Return type:

list[dict[str, Any]]

predict(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Searches for the best parameters and records results from prior steps.

OptimizerEDBOp.set_config() must be called prior to this method to generate the necessary files.

Parameters:

• prev_param (list[Any]) – Parameters provided from the previous prediction, provide an empty list for the first call

• yield_value (float) – Experimental yield

• experiment_dir (str) – Output directory for any generated files

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

• obj_func (Optional[Callable[..., float]], optional) – Ignored for this optimizer, defaults to None

Returns:

The next suggested reaction to perform

Return type:

list[Any]

set_config(experiment_dir: str, config: dict[str, Any]) → None

Generate all the necessary data files based on the given configuration.

See OptimizerABC.set_config() for more information about how to form the config dictionary and for general usage information.

Parameters:

• experiment_dir (str) – Output directory for the configuration file

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

train(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

No training step for this algorithm.

Returns:

List will always be empty.

Return type:

list[Any]

cyrxnopt.OptimizerNMSimplex module

class cyrxnopt.OptimizerNMSimplex.OptimizerNMSimplex(venv: NestedVenv)

Bases: OptimizerABC

get_config() → list[dict[str, Any]]

Gets the configuration options available for this optimizer.

See OptimizerABC.get_config() for more information about the config descriptions returned by this method and for general usage information.

Returns:

List of configuration options with option name, data type, and information about which values are allowed/defaulted.

Return type:

list[dict[str, Any]]

predict(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Find the desired optimum of the provided objective function.

Parameters:

• prev_param (list[Any]) – Parameters provided from the previous prediction, provide an empty list for the first call

• yield_value (float) – Result from the previous prediction

• experiment_dir (str) – Output directory for the optimizer algorithm

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

• obj_func (Optional[Callable[..., float]], optional) – Objective function to optimize, defaults to None

Returns:

The next suggested reaction to perform

Return type:

list[Any]

set_config(experiment_dir: str, config: dict[str, Any]) → None

Sets the configuration for this instance of the optimizer.

See OptimizerABC.set_config() for more information about how to form the config dictionary and for general usage information.

Parameters:

• experiment_dir (str) – Output directory for the configuration file

• config (dict[str, Any]) – Configuration options for this optimizer instance

train(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable | None = None) → list[Any]

No training step for this algorithm.

Returns:

List will always be empty.

Return type:

list[Any]

cyrxnopt.OptimizerSQSnobFit module

class cyrxnopt.OptimizerSQSnobFit.OptimizerSQSnobFit(venv: NestedVenv)

Bases: OptimizerABC

get_config() → list[dict[str, Any]]

Get the configuration options available for this optimizer.

See OptimizerABC.get_config() for more information about the config descriptions returned by this method and for general usage information.

Returns:

List of configuration options with option name, data type, and information about which values are allowed/defaulted.

Return type:

list[dict[str, Any]]

predict(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable[[...], float] | None = None) → list[Any]

Find the desired optimum of the provided objective function.

Parameters:

• prev_param (list[Any]) – Parameters provided from the previous prediction, provide an empty list for the first call

• yield_value (float) – Result from the previous prediction

• experiment_dir (str) – Output directory for the optimizer algorithm

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

• obj_func (Optional[Callable[..., float]], optional) – Objective function to optimize, defaults to None

Returns:

The next suggested reaction to perform

Return type:

list[Any]

set_config(experiment_dir: str, config: dict[str, Any]) → None

Set the configuration for this instance of the optimizer.

See OptimizerABC.set_config() for more information about how to form the config dictionary and for general usage information.

Parameters:

• experiment_dir (str) – Output directory for the configuration file

• config (dict[str, Any]) – CyRxnOpt-level config for the optimizer

train(prev_param: list[Any], yield_value: float, experiment_dir: str, config: dict[str, Any], obj_func: Callable | None = None) → list[Any]

No training step for this algorithm.

Returns:

List will always be empty.

Return type:

list[Any]

Module contents