run_optimization
boulderopal.run_optimization(graph, cost_node_name, output_node_names, max_iteration_count=None, target_cost=None, cost_tolerance=None, optimization_count=4, cost_history_scope=HistoryScope.NONE, seed=None)
Perform gradient-based deterministic optimization of generic real-valued functions.
Use this function to determine a choice of variables that minimizes the value of a scalar real-valued cost function of those variables. You express that cost function as a graph describing how the input variables are transformed into the cost value.
Note that this function will pick a different initial guess for each optimization run. You can provide initial values (or a list of them of the same length) for the optimization variables in the graph, and the optimizer will start with those values. If the optimization_count is larger than the number of provided initial values, the optimizer will use random values after consuming all available initial values.
Parameters
- graph (Graph) – The graph describing the cost C(v) and outputs {Fj(v)} as functions of the optimizable input variables v. The graph must contain nodes with names s (giving the cost function) and {sj} (giving the output functions).
- cost_node_name (str) – The name s of the real-valued scalar graph node that defines the cost function C(v) to be minimized.
- output_node_names (str or list [ str ]) – The names {sj} of the graph nodes that define the output functions {Fj(v)}. The function evaluates these using the optimized variables and returns them in the output.
- max_iteration_count (int or None , optional) – The maximum number of cost evaluations to perform. You can set this as an early stop condition for the optimizer. Defaults to None, which means that the optimizer runs until it converges.
- target_cost (float or None , optional) – A target value of the cost that you can set as an early stop condition for the optimizer. If the cost becomes equal or smaller than this value, the optimization halts. Defaults to None, meaning that the optimizer runs until it converges.
- cost_tolerance (float or None , optional) – The tolerance tol to check whether the cost function has converged. You can set this as an early stop condition for the optimizer. The optimizer stops when the improvement of the cost function over the iterations is below the tolerance. That is, it stops when (Cj−Cj+1)/max(∣Cj∣,∣Cj+1∣,1)≤tol. Defaults to None, meaning the optimizer handles the tolerance automatically. You can try different values depending on your optimization problem. A recommended starting value is about 1e-9.
- optimization_count (int , optional) – The number N of independent randomly seeded optimizations to perform. The function returns the results from the best optimization (the one with the lowest cost). Defaults to 4. Depending on the landscape of the optimization problem, a larger value will help in finding lower costs, at the expense of prolonging computation time.
- cost_history_scope (HistoryScope , optional) – Configuration for the scope of the returned cost history data. Use this to select if you want the history data to be returned. Defaults to no cost history data returned.
- seed (int or None , optional) – Seed for the random number generator used in the optimizer. If set, must be a non-negative integer. Use this option to generate deterministic results from the optimizer. Note that if your graph contains nodes generating random samples, you need to also set seeds for those nodes to ensure a reproducible result.
Returns
A dictionary containing the optimization result, with the following keys:
cost
: The minimum cost function value C(voptimized)
achieved across all optimizations.
output
: The dictionary giving the value of each requested output node, evaluated
at the optimized variables, namely
{sj:Fj(voptimized)}.
The keys of the dictionary are the names {sj} of the output nodes.
cost_history
: The evolution of the cost function across all optimizations and iterations.
metadata
: Metadata associated with the calculation.
No guarantees are made about the contents of this metadata dictionary;
the contained information is intended purely to help interpret the results of the
calculation on a one-off basis.
Return type
dict
SEE ALSO
boulderopal.closed_loop.optimize
: Run a closed-loop optimization to find a minimum of the given cost function.
boulderopal.closed_loop.step
: Perform a single step in a closed-loop optimization.
boulderopal.execute_graph
: Evaluate generic functions.
boulderopal.run_gradient_free_optimization
: Perform model-based optimization without using gradient values.
boulderopal.run_stochastic_optimization
: Perform gradient-based stochastic optimization of generic real-valued functions.
Notes
Given a cost function C(v) of variables v, this function computes an estimate voptimized of argminvC(v), namely the choice of variables v that minimizes C(v). The function then calculates the values of arbitrary output functions {Fj(voptimized)} with that choice of variables.
This function represents the cost and output functions as nodes of a graph. This graph defines the input optimization variables v, and how these variables are transformed into the corresponding cost and output quantities. You build the graph from primitive nodes defined in the graph of the Boulder Opal package. Each such node, which can be identified by a name, represents a function of the previous nodes in the graph (and thus, transitively, a function of the input variables). You can use any named scalar real-valued node s as the cost function, and any named nodes {sj} as outputs.
After you provide a cost function C(v) (via a graph), this function runs an optimization process N times, each with random initial variables, to identify N local minima of the cost function, and then takes the variables corresponding to the best such minimum as voptimized.
A common use-case for this function is to determine controls for a quantum system that yield an optimal gate: the variables v parameterize the controls to be optimized, and the cost function C(v) is the operational infidelity describing the quality of the resulting gate relative to a target gate. When combined with the node definitions in the Boulder Opal package, which make it convenient to define such cost functions, this function provides a highly configurable framework for quantum control that encapsulates other common tools such as gradient ascent pulse engineering 1 and chopped random basis (CRAB) optimization 2.
The optimizer will carry out the gradient-based optimization until it converges to a minimum. You can set up early stop conditions such as a target cost or a maximum number of cost evaluations via the target_cost and max_iteration_count parameters. If both are provided, the optimizer will terminate when either of the two conditions is satisfied.
References
- [1] N. Khaneja, T. Reiss, C. Kehlet, T. Schulte-Herbrüggen, and S. J. Glaser, Journal of Magnetic Resonance 172, 2 (2005).
- [2] P. Doria, T. Calarco, and S. Montangero, Physical Review Letters 106, 190501 (2011).
Examples
Perform a simple optimization with variables that are initialized to given values.
>>> graph = bo.Graph()
>>> variables = graph.optimization_variable(
... 2, -1, 1, initial_values=np.array([0.6, 0.]), name="variables"
... )
>>> x, y = variables[0], variables[1]
>>> cost = (x - 0.5) ** 2 + (y - 0.1) ** 2
>>> cost.name = "cost"
>>> result = bo.run_optimization(
... graph=graph, cost_node_name="cost", output_node_names="variables"
... )
>>> result["cost"], result["output"]
(0.0, {'variables': {'value': array([0.5, 0.1])}})
See also the How to optimize controls in arbitrary quantum systems using graphs user guide.