Example Sweep 2 Float 1 Cat Over Time

"""Auto-generated example: 2 Float, 1 Categorical (over time)."""

import random
import math
import bencher as bn
from datetime import datetime, timedelta

class CompressionCodec(bn.ParametrizedSweep):
    """Compression ratio across block size, entropy, and codec."""

    block_size = bn.FloatSweep(default=4096, bounds=[512, 65536], doc="Block size in bytes")
    entropy = bn.FloatSweep(default=0.5, bounds=[0.0, 1.0], doc="Input data entropy")
    codec = bn.StringSweep(["zlib", "lz4", "zstd"], doc="Compression codec")

    ratio = bn.ResultFloat(units="x", doc="Compression ratio")

    _time_offset = 0.0

    def benchmark(self):
        codec_eff = {"zlib": 1.0, "lz4": 0.7, "zstd": 1.1}[self.codec]
        self.ratio = codec_eff * (1.0 - 0.7 * self.entropy) * (1.0 + 0.3 * math.log2(self.block_size / 512))
        self.ratio += random.gauss(0, 0.1 * 0.3)
        self.ratio += self._time_offset * 10


def example_sweep_2_float_1_cat_over_time(run_cfg: bn.BenchRunCfg | None = None) -> bn.Bench:
    """2 Float, 1 Categorical (over time)."""
    if run_cfg is None:
        run_cfg = bn.BenchRunCfg()
    benchable = CompressionCodec()
    bench = benchable.to_bench(run_cfg)
    _base_time = datetime(2000, 1, 1)
    for i, offset in enumerate([0.0, 0.5, 1.0]):
        benchable._time_offset = offset
        run_cfg.clear_cache = True
        run_cfg.clear_history = i == 0
        res = bench.plot_sweep(
            "over_time",
            input_vars=['block_size', 'entropy', 'codec'],
            result_vars=['ratio'],
            description='A 2 float + 1 categorical parameter sweep tracked over time. Setting over_time=True records multiple time snapshots that can be scrubbed via a slider. Each call to plot_sweep with a new time_src appends a snapshot to the history. This is designed for nightly benchmarks or CI pipelines where you want to track how metrics evolve across commits, releases, or environmental changes. Use clear_history=True on the first snapshot to reset, and clear_cache=True to force re-evaluation. A 2D float sweep produces a heatmap. Additional categorical variables create faceted heatmaps, one per category combination.',
            post_description="The time slider lets you scrub through snapshots. The 'All Time Points (aggregated)' tab pools all snapshots into one view, smoothing out per-snapshot noise to reveal long-term trends.",
            run_cfg=run_cfg,
            time_src=_base_time + timedelta(seconds=i),
        )

    return bench


if __name__ == "__main__":
    bn.run(example_sweep_2_float_1_cat_over_time, subsampling_divisions=4, over_time=True)