chore: add tests for reservoir functions

[cdc-as81]

No description provided.

[github-actions]

Benchmark Results

Hyperfine

  Group                  Bench                     Mean [ms]  Min [ms]  Max [ms]       Relative
  ---------------------  ---------------------  ------------  --------  --------  -------------
  periodic_counts_bench  no_periodic_reports       7.3 ± 0.3       7.1       9.3           1.00
  periodic_counts_bench  with_periodic_reports    12.9 ± 1.1      12.3      19.8    1.76 ± 0.17
  large_sir              baseline                  2.8 ± 0.0       2.8       2.9           1.00
  large_sir              ixa                     136.9 ± 1.2     135.1     142.1   48.13 ± 0.70
  large_sir              ixa_no_queries           36.7 ± 1.0      35.3      40.7   12.90 ± 0.38

Criterion

Regressions:
  Group          Bench                                                         Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  -------  --------  --------
  examples       example-basic-infection                                       4.100%    3.792%    4.382%
  indexing       with_query_results_indexed_multi-property                     0.385%    0.006%    0.659%
  indexing       with_query_results_single_indexed_property                    0.358%    0.035%    0.573%
  indexing       query_people_multiple_individually_indexed_properties         1.126%    1.021%    1.234%
  sample_people  sampling_multiple_l_reservoir                                 2.775%    2.627%    2.938%
  sample_people  sampling_multiple_known_length                                0.867%    0.568%    1.155%
  sample_people  sampling_single_l_reservoir                                   0.700%    0.562%    0.838%
  sample_people  sampling_single_known_length                                  1.253%    0.884%    1.605%
  large_dataset  bench_query_population_multi_unindexed                        0.600%    0.266%    0.850%
  large_dataset  bench_query_population_multi_indexed                          0.913%    0.581%    1.269%

Improvements:
  Group          Bench                                                         Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  -------  --------  --------
  examples       example-births-deaths                                        -0.446%   -0.704%   -0.162%
  indexing       query_people_count_single_indexed_property                   -8.195%   -9.385%   -6.770%
  indexing       query_people_count_indexed_multi-property                    -0.640%   -0.879%   -0.433%
  indexing       with_query_results_multiple_individually_indexed_properties  -2.174%   -3.129%   -1.145%
  indexing       query_people_indexed_multi-property                          -2.898%   -3.622%   -2.009%
  indexing       query_people_count_multiple_individually_indexed_properties  -0.601%   -0.877%   -0.384%
  large_dataset  bench_query_population_derived_property                      -0.667%   -0.846%   -0.448%
  large_dataset  bench_query_population_indexed_property                      -1.859%   -2.172%   -1.426%

Unchanged:
  Group          Bench                                                         Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  -------  --------  --------
  indexing       query_people_single_indexed_property                         -0.083%   -0.190%    0.038%
  large_dataset  bench_query_population_property                              -0.244%   -0.663%    0.114%

[RobertJacobsonCDC]

This isn't what I had in mind when I opened #646. I added some more detail in a comment on the issue. The idea is to test a broad range of inputs to make sure the outputs have the expected distribution.

But this is good work and is worth merging.

• Double check the critical value for χ²₀.₉₉₉ for df=99. I have found that LLMs happily and confidently give wrong critical values.
• Leave #646 open. I can self-assign it and loop in one of our statisticians to help me with the mathematics.

[github-actions]

Benchmark Results

Hyperfine

  Group                  Bench                     Mean [ms]  Min [ms]  Max [ms]       Relative
  ---------------------  ---------------------  ------------  --------  --------  -------------
  periodic_counts_bench  no_periodic_reports       7.3 ± 0.2       7.1       8.1           1.00
  periodic_counts_bench  with_periodic_reports    12.7 ± 0.4      12.4      14.8    1.74 ± 0.07
  large_sir              baseline                  3.0 ± 0.3       2.8       4.7           1.00
  large_sir              ixa                     137.1 ± 1.9     134.7     143.7   45.11 ± 4.81
  large_sir              ixa_no_queries           36.7 ± 1.7      34.9      42.9   12.09 ± 1.40

Criterion

Regressions:
  Group          Bench                                                          Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  --------  --------  --------
  examples       example-births-deaths                                          1.507%    0.838%    2.144%
  indexing       query_people_count_single_indexed_property                    14.550%   13.102%   15.710%
  indexing       query_people_multiple_individually_indexed_properties          2.203%    1.277%    3.074%
  sample_people  sampling_multiple_l_reservoir                                  0.554%    0.493%    0.618%
  sample_people  sampling_single_l_reservoir                                    1.160%    0.910%    1.312%

Improvements:
  Group          Bench                                                          Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  --------  --------  --------
  indexing       query_people_count_indexed_multi-property                     -2.611%   -2.989%   -2.209%
  indexing       query_people_single_indexed_property                         -15.192%  -15.380%  -14.919%
  indexing       query_people_indexed_multi-property                          -14.561%  -15.383%  -13.797%
  indexing       query_people_count_multiple_individually_indexed_properties   -0.482%   -0.613%   -0.346%
  sample_people  sampling_multiple_known_length                                -0.396%   -0.448%   -0.345%
  sample_people  sampling_single_known_length                                  -2.202%   -2.389%   -2.018%
  large_dataset  bench_query_population_derived_property                       -3.346%   -3.736%   -2.837%
  large_dataset  bench_query_population_multi_unindexed                        -2.942%   -3.242%   -2.654%
  large_dataset  bench_query_population_multi_indexed                          -3.169%   -3.698%   -2.654%

Unchanged:
  Group          Bench                                                          Change  CI Lower  CI Upper
  -------------  -----------------------------------------------------------  --------  --------  --------
  examples       example-basic-infection                                        0.406%   -0.301%    1.112%
  indexing       with_query_results_indexed_multi-property                      0.169%   -0.129%    0.607%
  indexing       with_query_results_single_indexed_property                     0.290%   -0.181%    0.754%
  indexing       with_query_results_multiple_individually_indexed_properties   -0.726%   -1.723%    0.289%
  large_dataset  bench_query_population_property                               -0.108%   -0.549%    0.251%
  large_dataset  bench_query_population_indexed_property                       -0.276%   -0.746%    0.194%