Module temporal 

Polars-first temporal filtering operations for event camera data

This module provides time-based filtering functionality using Polars DataFrames and LazyFrames for maximum performance and memory efficiency. All operations work directly with Polars expressions and avoid unnecessary conversions.

Philosophy

This implementation follows a strict Polars-first approach:

• Input: LazyFrame (from events_to_dataframe)
• Processing: Polars expressions and transformations
• Output: LazyFrame (convertible to Vec/numpy only when needed)

Performance Benefits

• Lazy evaluation: Operations are optimized and executed only when needed
• Vectorized operations: All filtering uses SIMD-optimized Polars operations
• Memory efficiency: No intermediate Vec allocations
• Query optimization: Polars optimizes the entire filtering pipeline

Example

use polars::prelude::*;
use evlib::ev_filtering::temporal::*;

// Convert events to LazyFrame once
let events_df = events_to_dataframe(&events)?.lazy();

// Apply temporal filtering with Polars expressions
let filtered = apply_temporal_filter(events_df, &TemporalFilter::time_window(1.0, 5.0))?;

Structs

TemporalFilter

Temporal filtering configuration optimized for Polars operations

Constants

COL_POLARITY

COL_T

COL_X

Polars column names for event data

COL_Y

Functions

apply_temporal_filter

Apply temporal filtering using Polars expressions

calculate_event_rates

Calculate event rates over time windows using Polars

filter_by_time_df

Filter events by time window - DataFrame-native version (recommended)

filter_time_window

Filter events by time window using Polars expressions

get_temporal_statistics

Get temporal statistics using Polars aggregations