Struct OptimizerBenchmarks 

pub struct OptimizerBenchmarks { /* private fields */ }

Core optimizer benchmark suite

Provides fundamental benchmarking operations including step performance, convergence analysis, memory scaling, and sparse gradient handling.

Implementations

impl OptimizerBenchmarks

pub fn new() -> Self

Create a new benchmark suite with default configuration

pub fn with_config(config: BenchmarkConfig) -> Self

Create a new benchmark suite with custom configuration

pub fn config(&self) -> &BenchmarkConfig

Get the current configuration

pub fn set_config(&mut self, config: BenchmarkConfig)

Update the configuration

pub fn benchmark_step_performance<O: Optimizer>( &self, optimizer: O, problem_size: usize, ) -> OptimizerResult<BenchmarkResult>

Benchmark optimizer step performance

Measures the raw computational performance of the optimizer’s step operation across multiple iterations to get reliable timing statistics.

Arguments

• optimizer - Optimizer to benchmark
• problem_size - Number of parameters in the optimization problem

Returns

Benchmark results with timing statistics

pub fn benchmark_quadratic_convergence<O: Optimizer>( &self, optimizer: O, dimension: usize, ) -> OptimizerResult<BenchmarkResult>

Benchmark convergence on quadratic function

Tests how quickly the optimizer converges on a simple quadratic optimization problem. This provides insight into convergence behavior and rate.

Arguments

• optimizer - Optimizer to benchmark
• dimension - Dimensionality of the optimization problem

Returns

Benchmark results with convergence metrics

pub fn benchmark_sparse_gradients<O: Optimizer>( &self, optimizer: O, total_params: usize, sparsity: f32, ) -> OptimizerResult<BenchmarkResult>

Benchmark sparse gradient handling

Tests optimizer performance when dealing with sparse gradients, which is common in many machine learning scenarios.

Arguments

• optimizer - Optimizer to benchmark
• total_params - Total number of parameters
• sparsity - Fraction of gradients that are non-zero (0.0 to 1.0)

Returns

Benchmark results for sparse gradient performance

pub fn run_comprehensive_benchmarks<O: Optimizer + Clone>( &self, optimizer: O, ) -> OptimizerResult<Vec<BenchmarkResult>>

Run a comprehensive set of core benchmarks

Executes multiple benchmark scenarios to provide a complete performance profile of the optimizer including step performance, convergence, and sparse gradient handling.

Arguments

• optimizer - Optimizer to benchmark (must be cloneable)

Returns

Vector of benchmark results covering different scenarios

pub fn print_results(&self, results: &[BenchmarkResult])

Print benchmark results in a formatted table

Displays benchmark results in an easy-to-read tabular format with timing, convergence, and performance metrics.

Trait Implementations

impl Default for OptimizerBenchmarks

fn default() -> Self

Returns the “default value” for a type.