diffcb 0.1.9__tar.gz → 0.1.13__tar.gz

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round-41-p100-batch-benchmark.log

@@ -0,0 +1,263 @@
+[{"stream_name":"stdout","time":0.72082766,"data":"GPU compute capability: 6.0\n"}
+,{"stream_name":"stdout","time":0.720915936,"data":"P100/sm_60 detected — reinstalling torch cu118 + numpy\u003c2 ...\n"}
+,{"stream_name":"stdout","time":12.29011906,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 819.1/819.1 MB 2.2 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":20.188534158,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 23.2/23.2 MB 74.8 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":20.393213602,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 875.6/875.6 kB 50.3 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":20.742501212,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 13.1/13.1 MB 103.4 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":28.449990434,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 728.5/728.5 MB 2.2 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":37.681499644,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 417.9/417.9 MB 2.0 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":40.965091507,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 168.4/168.4 MB 8.5 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":43.634749524,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 58.1/58.1 MB 6.4 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":45.176572974,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 128.2/128.2 MB 14.9 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":47.647797995,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 204.1/204.1 MB 9.2 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":49.71645701,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 135.3/135.3 MB 14.0 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":50.35822416,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 99.1/99.1 kB 8.2 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":50.857135576,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 3.6 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":50.912674846,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 9.1 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":50.934839999,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 12.4 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":50.988030724,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 50.7 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":51.067073057,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 96.3 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":51.115947709,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 34.3 MB/s eta 0:00:00\n"}
+,{"stream_name":"stderr","time":148.347635937,"data":"ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n"}
+,{"stream_name":"stderr","time":148.347702967,"data":"bigframes 2.39.0 requires google-cloud-bigquery-storage\u003c3.0.0,\u003e=2.30.0, which is not installed.\n"}
+,{"stream_name":"stderr","time":148.347707843,"data":"google-adk 1.29.0 requires google-cloud-bigquery-storage\u003e=2.0.0, which is not installed.\n"}
+,{"stream_name":"stderr","time":148.347711415,"data":"datasets 4.8.5 requires fsspec[http]\u003c=2026.2.0,\u003e=2023.1.0, but you have fsspec 2026.4.0 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347715137,"data":"ydata-profiling 4.18.4 requires numba\u003c0.63,\u003e=0.60, but you have numba 0.65.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.34771847,"data":"ydata-profiling 4.18.4 requires numpy\u003c2.4,\u003e=1.22, but you have numpy 2.4.6 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347721905,"data":"google-colab 1.0.0 requires jupyter-server==2.14.0, but you have jupyter-server 2.12.5 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347725079,"data":"google-colab 1.0.0 requires pandas==2.2.2, but you have pandas 2.3.3 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347728254,"data":"dopamine-rl 4.1.2 requires gym\u003c=0.25.2, but you have gym 0.26.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347731603,"data":"torchaudio 2.10.0+cu128 requires torch==2.10.0, but you have torch 2.2.2+cu118 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347734774,"data":"torchvision 0.25.0+cu128 requires torch==2.10.0, but you have torch 2.2.2+cu118 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347741174,"data":"tensorflow 2.19.0 requires numpy\u003c2.2.0,\u003e=1.26.0, but you have numpy 2.4.6 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347744502,"data":"cudf-cu12 26.2.1 requires numba\u003c0.62.0,\u003e=0.60.0, but you have numba 0.65.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347747699,"data":"cudf-cu12 26.2.1 requires numba-cuda[cu12]\u003c0.23.0,\u003e=0.22.2, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":148.347750951,"data":"gcsfs 2025.3.0 requires fsspec==2025.3.0, but you have fsspec 2026.4.0 which is incompatible.\n"}
+,{"stream_name":"stdout","time":150.482489986,"data":"     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 61.0/61.0 kB 2.6 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":150.695599091,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 18.0/18.0 MB 82.9 MB/s eta 0:00:00\n"}
+,{"stream_name":"stderr","time":155.828845789,"data":"ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n"}
+,{"stream_name":"stderr","time":155.828881699,"data":"bigframes 2.39.0 requires google-cloud-bigquery-storage\u003c3.0.0,\u003e=2.30.0, which is not installed.\n"}
+,{"stream_name":"stderr","time":155.828887555,"data":"datasets 4.8.5 requires fsspec[http]\u003c=2026.2.0,\u003e=2023.1.0, but you have fsspec 2026.4.0 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828891839,"data":"kaggle-environments 1.29.3 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828917192,"data":"ydata-profiling 4.18.4 requires numba\u003c0.63,\u003e=0.60, but you have numba 0.65.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828920738,"data":"cesium 0.12.4 requires numpy\u003c3.0,\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828923579,"data":"google-colab 1.0.0 requires jupyter-server==2.14.0, but you have jupyter-server 2.12.5 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.82892648199999,"data":"google-colab 1.0.0 requires pandas==2.2.2, but you have pandas 2.3.3 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828929251,"data":"dask-cuda 26.2.0 requires cuda-core==0.3.*, but you have cuda-core 1.0.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828940533,"data":"dask-cuda 26.2.0 requires numba-cuda\u003c0.23.0,\u003e=0.22.1, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.82894367,"data":"dopamine-rl 4.1.2 requires gym\u003c=0.25.2, but you have gym 0.26.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828946415,"data":"distributed-ucxx-cu12 0.48.0 requires numba-cuda[cu12]\u003c0.23.0,\u003e=0.22.1, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828949465,"data":"jaxlib 0.7.2 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828952233,"data":"cupy-cuda12x 14.0.1 requires numpy\u003c2.6,\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828955059,"data":"moviepy 1.0.3 requires decorator\u003c5.0,\u003e=4.0.2, but you have decorator 5.3.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828957902,"data":"opencv-python 4.13.0.92 requires numpy\u003e=2; python_version \u003e= \"3.9\", but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828960993,"data":"tobler 0.14.0 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828963733,"data":"cuml-cu12 26.2.0 requires numba\u003c0.62.0,\u003e=0.60.0, but you have numba 0.65.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828966622,"data":"cuml-cu12 26.2.0 requires numba-cuda[cu12]\u003c0.23.0,\u003e=0.22.1, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828971153,"data":"jax 0.7.2 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828974087,"data":"ucxx-cu12 0.48.0 requires numba-cuda[cu12]\u003c0.23.0,\u003e=0.22.1, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828976916,"data":"opencv-python-headless 4.13.0.92 requires numpy\u003e=2; python_version \u003e= \"3.9\", but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828979888,"data":"rasterio 1.5.0 requires numpy\u003e=2, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828982645,"data":"torchvision 0.25.0+cu128 requires torch==2.10.0, but you have torch 2.2.2+cu118 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828985412,"data":"pytensor 2.38.2 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828988168,"data":"xarray-einstats 0.10.0 requires numpy\u003e=2.0, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828991002,"data":"cudf-cu12 26.2.1 requires numba\u003c0.62.0,\u003e=0.60.0, but you have numba 0.65.1 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828996899,"data":"cudf-cu12 26.2.1 requires numba-cuda[cu12]\u003c0.23.0,\u003e=0.22.2, but you have numba-cuda 0.30.2 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.828999937,"data":"opencv-contrib-python 4.13.0.92 requires numpy\u003e=2; python_version \u003e= \"3.9\", but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stderr","time":155.829002839,"data":"shap 0.51.0 requires numpy\u003e=2, but you have numpy 1.26.4 which is incompatible.\n"}
+,{"stream_name":"stdout","time":155.925133898,"data":"torch 2.2.2+cu118 installed.\n"}
+,{"stream_name":"stdout","time":155.925171623,"data":"Installing diffcb from PyPI ...\n"}
+,{"stream_name":"stdout","time":157.403641053,"data":"   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 53.3/53.3 kB 3.7 MB/s eta 0:00:00\n"}
+,{"stream_name":"stdout","time":159.507991459,"data":"diffcb installed from PyPI.\n"}
+,{"stream_name":"stdout","time":159.509152224,"data":"Rscript found at /usr/bin/Rscript — installing multimode package ...\n"}
+,{"stream_name":"stderr","time":160.991413058,"data":"also installing the dependencies ‘FNN’, ‘kernlab’, ‘mclust’, ‘multicool’, ‘mvtnorm’, ‘pracma’, ‘diptest’, ‘ks’, ‘rootSolve’\n"}
+,{"stream_name":"stderr","time":160.991441812,"data":"\n"}
+,{"stream_name":"stdout","time":288.702787794,"data":"R multimode install attempted.\n"}
+,{"stream_name":"stdout","time":288.702857898,"data":"Re-launching with DCB_KAGGLE_READY=1 ...\n"}
+,{"stream_name":"stdout","time":409.358928657,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359004858,"data":"Round 41 — P100 fp64 Verification: forward_batched B1+B2 throughput\n"}
+,{"stream_name":"stdout","time":409.359014614,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359020345,"data":"diffcb         : 0.1.11\n"}
+,{"stream_name":"stdout","time":409.359025697,"data":"PyTorch        : 2.2.2+cu118\n"}
+,{"stream_name":"stdout","time":409.359030708,"data":"CUDA available : True\n"}
+,{"stream_name":"stdout","time":409.359035236,"data":"GPU            : Tesla P100-PCIE-16GB  (sm_6.0)\n"}
+,{"stream_name":"stdout","time":409.359039809,"data":"forward_batched: YES (B1+B2 present)\n"}
+,{"stream_name":"stdout","time":409.359044279,"data":"Rscript        : /usr/bin/Rscript\n"}
+,{"stream_name":"stdout","time":409.359048853,"data":"R+multimode    : YES\n"}
+,{"stream_name":"stdout","time":409.359053245,"data":"\n"}
+,{"stream_name":"stdout","time":409.35906211,"data":"\n"}
+,{"stream_name":"stdout","time":409.359065497,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359068597,"data":"Part 1: fp64 vs fp32 single-distribution baseline (DCBLayer forward)\n"}
+,{"stream_name":"stdout","time":409.359071928,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359074992,"data":"\n"}
+,{"stream_name":"stdout","time":409.359078619,"data":"         n     dtype          ms  note\n"}
+,{"stream_name":"stdout","time":409.35909309,"data":"--------------------------------------------------\n"}
+,{"stream_name":"stdout","time":409.359098661,"data":"     1,000   float64      14.339ms\n"}
+,{"stream_name":"stdout","time":409.35910381,"data":"     1,000   float32      11.177ms\n"}
+,{"stream_name":"stdout","time":409.359108961,"data":"    10,000   float64      77.492ms\n"}
+,{"stream_name":"stdout","time":409.359113873,"data":"    10,000   float32      39.110ms\n"}
+,{"stream_name":"stdout","time":409.35913455,"data":"    50,000   float64      38.082ms\n"}
+,{"stream_name":"stdout","time":409.359138465,"data":"    50,000   float32      24.278ms\n"}
+,{"stream_name":"stdout","time":409.359142612,"data":"   100,000   float64      38.839ms\n"}
+,{"stream_name":"stdout","time":409.359145855,"data":"   100,000   float32      24.595ms\n"}
+,{"stream_name":"stdout","time":409.359148866,"data":"   500,000   float64      38.882ms\n"}
+,{"stream_name":"stdout","time":409.359152455,"data":"   500,000   float32      25.510ms\n"}
+,{"stream_name":"stdout","time":409.35915556,"data":" 1,000,000   float64      40.062ms\n"}
+,{"stream_name":"stdout","time":409.359158638,"data":" 1,000,000   float32      25.666ms\n"}
+,{"stream_name":"stdout","time":409.359161641,"data":"\n"}
+,{"stream_name":"stdout","time":409.359164548,"data":"fp64/fp32 ratio (\u003e1 means fp64 is slower):\n"}
+,{"stream_name":"stdout","time":409.35916824,"data":"  n=    1,000  ratio=1.283x\n"}
+,{"stream_name":"stdout","time":409.359183343,"data":"  n=   10,000  ratio=1.981x\n"}
+,{"stream_name":"stdout","time":409.359186914,"data":"  n=   50,000  ratio=1.569x\n"}
+,{"stream_name":"stdout","time":409.359190799,"data":"  n=  100,000  ratio=1.579x\n"}
+,{"stream_name":"stdout","time":409.3591943,"data":"  n=  500,000  ratio=1.524x\n"}
+,{"stream_name":"stdout","time":409.359197857,"data":"  n=1,000,000  ratio=1.561x\n"}
+,{"stream_name":"stdout","time":409.359201197,"data":"\n"}
+,{"stream_name":"stdout","time":409.359204451,"data":"Saved: /kaggle/working/round41_p100_fp64_vs_fp32.csv\n"}
+,{"stream_name":"stdout","time":409.359207733,"data":"\n"}
+,{"stream_name":"stdout","time":409.359211141,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.35921476,"data":"Part 2: forward_batched B1+B2 throughput — K distributions, fp64, CUDA\n"}
+,{"stream_name":"stdout","time":409.35921837,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359222254,"data":"\n"}
+,{"stream_name":"stdout","time":409.3592256,"data":"   K         n    serial_ms    batch_ms    speedup    serial_dps     batch_dps\n"}
+,{"stream_name":"stdout","time":409.359229319,"data":"---------------------------------------------------------------------------\n"}
+,{"stream_name":"stdout","time":409.359232919,"data":"   1    50,000       64.165      11.712      5.478x          15.6          85.4\n"}
+,{"stream_name":"stdout","time":409.359236519,"data":"   4    50,000      257.835      20.254     12.730x          15.5         197.5\n"}
+,{"stream_name":"stdout","time":409.359240116,"data":"   8    50,000      515.684      33.490     15.398x          15.5         238.9\n"}
+,{"stream_name":"stdout","time":409.359245329,"data":"  16    50,000     1035.783      57.476     18.021x          15.4         278.4\n"}
+,{"stream_name":"stdout","time":409.359249085,"data":"  32    50,000     2070.100     104.301     19.847x          15.5         306.8\n"}
+,{"stream_name":"stdout","time":409.359252898,"data":"  64    50,000     4134.924     211.246     19.574x          15.5         303.0\n"}
+,{"stream_name":"stdout","time":409.359256502,"data":"   1   100,000       39.555      11.305      3.499x          25.3          88.5\n"}
+,{"stream_name":"stdout","time":409.359260182,"data":"   4   100,000      161.275      21.432      7.525x          24.8         186.6\n"}
+,{"stream_name":"stdout","time":409.359263746,"data":"   8   100,000      318.401      33.546      9.492x          25.1         238.5\n"}
+,{"stream_name":"stdout","time":409.359272354,"data":"  16   100,000      638.605      59.283     10.772x          25.1         269.9\n"}
+,{"stream_name":"stdout","time":409.359276169,"data":"  32   100,000     1278.956     111.975     11.422x          25.0         285.8\n"}
+,{"stream_name":"stdout","time":409.359279922,"data":"  64   100,000     2586.546     224.367     11.528x          24.7         285.2\n"}
+,{"stream_name":"stdout","time":409.359283666,"data":"\n"}
+,{"stream_name":"stdout","time":409.359287214,"data":"Saved: /kaggle/working/round41_p100_batch_throughput.csv\n"}
+,{"stream_name":"stdout","time":409.359290964,"data":"\n"}
+,{"stream_name":"stdout","time":409.359294412,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359298235,"data":"Part 3: R vs DCB speedup on P100 (fp64, bimodal)\n"}
+,{"stream_name":"stdout","time":409.359301745,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":409.359305302,"data":"\n"}
+,{"stream_name":"stdout","time":409.359308657,"data":"         n    dcb_cpu_ms    dcb_gpu_ms        r_ms   spdup_cpu   spdup_gpu\n"}
+,{"stream_name":"stdout","time":409.359311839,"data":"---------------------------------------------------------------------------\n"}
+,{"stream_name":"stdout","time":410.876480782,"data":"  n=    1,000  R running... 1515.5ms  h_r=1.92690\n"}
+,{"stream_name":"stdout","time":465.795200963,"data":"     1,000       135.443        13.895      1515.5       11.19x      109.06x\n"}
+,{"stream_name":"stdout","time":467.389401749,"data":"  n=   10,000  R running... 1583.3ms  h_r=1.93263\n"}
+,{"stream_name":"stdout","time":489.070690183,"data":"    10,000      6816.680        77.805      1583.3        0.23x       20.35x\n"}
+,{"stream_name":"stdout","time":490.8868209,"data":"  n=  100,000  R running... 1710.0ms  h_r=1.93132\n"}
+,{"stream_name":"stdout","time":514.786554549,"data":"   100,000      2653.420        38.752      1710.0        0.64x       44.13x\n"}
+,{"stream_name":"stdout","time":520.402311799,"data":"  n=1,000,000  R running... 4650.7ms  h_r=1.93536\n"}
+,{"stream_name":"stdout","time":670.076779491,"data":" 1,000,000      2902.096        39.459      4650.7        1.60x      117.86x\n"}
+,{"stream_name":"stdout","time":670.076831944,"data":"\n"}
+,{"stream_name":"stdout","time":670.076839406,"data":"Saved: /kaggle/working/round41_p100_speedup_vs_R.csv\n"}
+,{"stream_name":"stdout","time":670.076844422,"data":"\n"}
+,{"stream_name":"stdout","time":670.076849728,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":670.076854322,"data":"Part 4: Same-sample accuracy vs R (fp64, CPU layer)\n"}
+,{"stream_name":"stdout","time":670.076859076,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":670.076863579,"data":"  [  1/30] n=  1,000 dist= bimodal seed=0 h_r=1.89359 h_dcb=1.89352 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076868321,"data":"  [  2/30] n=  1,000 dist= bimodal seed=1 h_r=1.87362 h_dcb=1.87358 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076873215,"data":"  [  3/30] n=  1,000 dist= bimodal seed=2 h_r=1.94006 h_dcb=1.94005 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076878624,"data":"  [  4/30] n=  1,000 dist= bimodal seed=3 h_r=1.91296 h_dcb=1.91288 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076883023,"data":"  [  5/30] n=  1,000 dist= bimodal seed=4 h_r=1.91158 h_dcb=1.91157 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076887396,"data":"  [  6/30] n=  1,000 dist=gaussian seed=0 h_r=0.31418 h_dcb=0.31417 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076900961,"data":"  [  7/30] n=  1,000 dist=gaussian seed=1 h_r=0.27839 h_dcb=0.27821 overest=-0.07%\n"}
+,{"stream_name":"stdout","time":670.076904881,"data":"  [  8/30] n=  1,000 dist=gaussian seed=2 h_r=0.24442 h_dcb=0.24419 overest=-0.09%\n"}
+,{"stream_name":"stdout","time":670.076908861,"data":"  [  9/30] n=  1,000 dist=gaussian seed=3 h_r=0.21614 h_dcb=0.21593 overest=-0.10%\n"}
+,{"stream_name":"stdout","time":670.076912426,"data":"  [ 10/30] n=  1,000 dist=gaussian seed=4 h_r=0.15119 h_dcb=0.15117 overest=-0.01%\n"}
+,{"stream_name":"stdout","time":670.07692525,"data":"  [ 11/30] n= 10,000 dist= bimodal seed=0 h_r=1.93578 h_dcb=1.93571 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076928845,"data":"  [ 12/30] n= 10,000 dist= bimodal seed=1 h_r=1.93499 h_dcb=1.93494 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076932931,"data":"  [ 13/30] n= 10,000 dist= bimodal seed=2 h_r=1.93374 h_dcb=1.93373 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076936527,"data":"  [ 14/30] n= 10,000 dist= bimodal seed=3 h_r=1.92794 h_dcb=1.92787 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076940159,"data":"  [ 15/30] n= 10,000 dist= bimodal seed=4 h_r=1.93086 h_dcb=1.93085 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076943992,"data":"  [ 16/30] n= 10,000 dist=gaussian seed=0 h_r=0.16972 h_dcb=0.16949 overest=-0.13%\n"}
+,{"stream_name":"stdout","time":670.076948713,"data":"  [ 17/30] n= 10,000 dist=gaussian seed=1 h_r=0.12385 h_dcb=0.12384 overest=-0.01%\n"}
+,{"stream_name":"stdout","time":670.076952714,"data":"  [ 18/30] n= 10,000 dist=gaussian seed=2 h_r=0.14294 h_dcb=0.14283 overest=-0.07%\n"}
+,{"stream_name":"stdout","time":670.076956463,"data":"  [ 19/30] n= 10,000 dist=gaussian seed=3 h_r=0.18045 h_dcb=0.18031 overest=-0.08%\n"}
+,{"stream_name":"stdout","time":670.076960188,"data":"  [ 20/30] n= 10,000 dist=gaussian seed=4 h_r=0.13019 h_dcb=0.12990 overest=-0.22%\n"}
+,{"stream_name":"stdout","time":670.076965783,"data":"  [ 21/30] n=100,000 dist= bimodal seed=0 h_r=1.93205 h_dcb=1.93205 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076969443,"data":"  [ 22/30] n=100,000 dist= bimodal seed=1 h_r=1.93661 h_dcb=1.93660 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076972821,"data":"  [ 23/30] n=100,000 dist= bimodal seed=2 h_r=1.93025 h_dcb=1.93024 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.076976517,"data":"  [ 24/30] n=100,000 dist= bimodal seed=3 h_r=1.92798 h_dcb=1.92797 overest=-0.00%\n"}
+,{"stream_name":"stdout","time":670.07698019,"data":"  [ 25/30] n=100,000 dist= bimodal seed=4 h_r=1.93691 h_dcb=1.93691 overest=+0.00%\n"}
+,{"stream_name":"stdout","time":670.076983892,"data":"  [ 26/30] n=100,000 dist=gaussian seed=0 h_r=0.22108 h_dcb=0.22111 overest=+0.01%\n"}
+,{"stream_name":"stdout","time":670.076987647,"data":"  [ 27/30] n=100,000 dist=gaussian seed=1 h_r=0.11059 h_dcb=0.11063 overest=+0.04%\n"}
+,{"stream_name":"stdout","time":670.076991421,"data":"  [ 28/30] n=100,000 dist=gaussian seed=2 h_r=0.28461 h_dcb=0.28469 overest=+0.03%\n"}
+,{"stream_name":"stdout","time":670.076995212,"data":"  [ 29/30] n=100,000 dist=gaussian seed=3 h_r=0.12143 h_dcb=0.12159 overest=+0.13%\n"}
+,{"stream_name":"stdout","time":670.076998814,"data":"  [ 30/30] n=100,000 dist=gaussian seed=4 h_r=0.14292 h_dcb=0.14281 overest=-0.07%\n"}
+,{"stream_name":"stdout","time":670.077003107,"data":"\n"}
+,{"stream_name":"stdout","time":670.077006452,"data":"Saved: /kaggle/working/round41_p100_accuracy.csv\n"}
+,{"stream_name":"stdout","time":670.07701046,"data":"\n"}
+,{"stream_name":"stdout","time":670.07701425,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":670.077021995,"data":"Round 41 — P100 fp64 Verification Summary\n"}
+,{"stream_name":"stdout","time":670.077026237,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":670.077030158,"data":"diffcb         : 0.1.11\n"}
+,{"stream_name":"stdout","time":670.077034003,"data":"PyTorch        : 2.2.2+cu118\n"}
+,{"stream_name":"stdout","time":670.077037443,"data":"GPU            : 6.0, Tesla P100-PCIE-16GB\n"}
+,{"stream_name":"stdout","time":670.077041008,"data":"forward_batched: YES (B1+B2)\n"}
+,{"stream_name":"stdout","time":670.077044442,"data":"R+multimode    : YES\n"}
+,{"stream_name":"stdout","time":670.07704802,"data":"\n"}
+,{"stream_name":"stdout","time":670.07705142,"data":"── Part 1: fp64 vs fp32 (bimodal, DCBLayer, best-of-5) ──────────────\n"}
+,{"stream_name":"stdout","time":670.077055388,"data":"           n     dtype          ms\n"}
+,{"stream_name":"stdout","time":670.077058863,"data":"──────────────────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077062758,"data":"      50,000   float64     38.0818\n"}
+,{"stream_name":"stdout","time":670.077066308,"data":"      50,000   float32     24.2779\n"}
+,{"stream_name":"stdout","time":670.077069815,"data":"     100,000   float64     38.8386\n"}
+,{"stream_name":"stdout","time":670.077073245,"data":"     100,000   float32     24.5947\n"}
+,{"stream_name":"stdout","time":670.077076812,"data":"   1,000,000   float64     40.0615\n"}
+,{"stream_name":"stdout","time":670.077080331,"data":"   1,000,000   float32     25.6655\n"}
+,{"stream_name":"stdout","time":670.077084128,"data":"\n"}
+,{"stream_name":"stdout","time":670.077087483,"data":"fp64 / fp32 latency ratio:\n"}
+,{"stream_name":"stdout","time":670.077090926,"data":"  n=   50,000  ratio=1.569x  (P100 strong fp64)\n"}
+,{"stream_name":"stdout","time":670.077094476,"data":"  n=  100,000  ratio=1.579x  (P100 strong fp64)\n"}
+,{"stream_name":"stdout","time":670.077098069,"data":"  n=1,000,000  ratio=1.561x  (P100 strong fp64)\n"}
+,{"stream_name":"stdout","time":670.077101597,"data":"\n"}
+,{"stream_name":"stdout","time":670.077105071,"data":"── Part 2: forward_batched B1+B2 speedup (fp64, CUDA) ───────────────\n"}
+,{"stream_name":"stdout","time":670.077109255,"data":"   K         n    serial_ms    batch_ms    speedup     batch_dps\n"}
+,{"stream_name":"stdout","time":670.077113263,"data":"─────────────────────────────────────────────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077118622,"data":"   1    50,000      64.1652     11.7122     5.4785x          85.4\n"}
+,{"stream_name":"stdout","time":670.077122295,"data":"   4    50,000     257.8352     20.2539    12.7302x         197.5\n"}
+,{"stream_name":"stdout","time":670.077125853,"data":"   8    50,000     515.6841     33.4900    15.3981x         238.9\n"}
+,{"stream_name":"stdout","time":670.077131093,"data":"  16    50,000    1035.7828     57.4761    18.0211x         278.4\n"}
+,{"stream_name":"stdout","time":670.07713562,"data":"  32    50,000    2070.0996    104.3010    19.8474x         306.8\n"}
+,{"stream_name":"stdout","time":670.077139355,"data":"  64    50,000    4134.9243    211.2463    19.5739x         303.0\n"}
+,{"stream_name":"stdout","time":670.077143255,"data":"   1   100,000      39.5547     11.3051     3.4988x          88.5\n"}
+,{"stream_name":"stdout","time":670.07714699,"data":"   4   100,000     161.2752     21.4318     7.5250x         186.6\n"}
+,{"stream_name":"stdout","time":670.077158679,"data":"   8   100,000     318.4005     33.5456     9.4916x         238.5\n"}
+,{"stream_name":"stdout","time":670.077162602,"data":"  16   100,000     638.6050     59.2827    10.7722x         269.9\n"}
+,{"stream_name":"stdout","time":670.077165816,"data":"  32   100,000    1278.9561    111.9748    11.4218x         285.8\n"}
+,{"stream_name":"stdout","time":670.077174258,"data":"  64   100,000    2586.5457    224.3666    11.5282x         285.2\n"}
+,{"stream_name":"stdout","time":670.077177802,"data":"\n"}
+,{"stream_name":"stdout","time":670.077180975,"data":"Peak B1+B2 speedup: 19.85x at K=32 n=50,000\n"}
+,{"stream_name":"stdout","time":670.077184333,"data":"10x target: PASS (\u003e=10x)\n"}
+,{"stream_name":"stdout","time":670.077188087,"data":"\n"}
+,{"stream_name":"stdout","time":670.077191611,"data":"── Part 3: DCB CUDA vs R speedup ───────────────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077195726,"data":"           n    dcb_gpu_ms        r_ms   speedup_gpu\n"}
+,{"stream_name":"stdout","time":670.077199377,"data":"────────────────────────────────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077203597,"data":"       1,000       13.8954    1515.485       109.064x\n"}
+,{"stream_name":"stdout","time":670.077207138,"data":"      10,000       77.8050    1583.335        20.350x\n"}
+,{"stream_name":"stdout","time":670.077210624,"data":"     100,000       38.7515    1710.029        44.128x\n"}
+,{"stream_name":"stdout","time":670.077214125,"data":"   1,000,000       39.4587    4650.677       117.862x\n"}
+,{"stream_name":"stdout","time":670.077217714,"data":"\n"}
+,{"stream_name":"stdout","time":670.077221085,"data":"── Part 4: Accuracy (mean overest_pct vs R) ────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077224792,"data":"           n        dist       mean%      min%      max%\n"}
+,{"stream_name":"stdout","time":670.077228465,"data":"───────────────────────────────────────────────────────\n"}
+,{"stream_name":"stdout","time":670.077232667,"data":"       1,000     bimodal      -0.002    -0.004    -0.001\n"}
+,{"stream_name":"stdout","time":670.07723628,"data":"       1,000    gaussian      -0.054    -0.098    -0.003\n"}
+,{"stream_name":"stdout","time":670.077240683,"data":"      10,000     bimodal      -0.002    -0.004    -0.000\n"}
+,{"stream_name":"stdout","time":670.077244261,"data":"      10,000    gaussian      -0.102    -0.223    -0.008\n"}
+,{"stream_name":"stdout","time":670.077247891,"data":"     100,000     bimodal      -0.000    -0.001    +0.001\n"}
+,{"stream_name":"stdout","time":670.077251376,"data":"     100,000    gaussian      +0.027    -0.075    +0.132\n"}
+,{"stream_name":"stdout","time":670.077254913,"data":"\n"}
+,{"stream_name":"stdout","time":670.077258284,"data":"======================================================================\n"}
+,{"stream_name":"stdout","time":670.077261887,"data":"Saved: /kaggle/working/round41_p100_summary.txt\n"}
+,{"stream_name":"stdout","time":670.077265455,"data":"\n"}
+,{"stream_name":"stdout","time":670.077268888,"data":"All done — output files in /kaggle/working/\n"}
+,{"stream_name":"stderr","time":672.421272098,"data":"/usr/local/lib/python3.12/dist-packages/mistune.py:435: SyntaxWarning: invalid escape sequence '\\|'\n"}
+,{"stream_name":"stderr","time":672.42132075,"data":"  cells[i][c] = re.sub('\\\\\\\\\\|', '|', cell)\n"}
+,{"stream_name":"stderr","time":672.642084242,"data":"/usr/local/lib/python3.12/dist-packages/nbconvert/filters/filter_links.py:36: SyntaxWarning: invalid escape sequence '\\_'\n"}
+,{"stream_name":"stderr","time":672.64212271,"data":"  text = re.sub(r'_', '\\_', text) # Escape underscores in display text\n"}
+,{"stream_name":"stderr","time":673.371585114,"data":"[NbConvertApp] Converting notebook __script__.ipynb to html\n"}
+,{"stream_name":"stderr","time":674.607746691,"data":"[NbConvertApp] Writing 407584 bytes to __results__.html\n"}
+]

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round37_s3_convergence_guard.csv

@@ -0,0 +1,67 @@
+n,seed,dist,cosine_fd,mag_ratio,k_proj,path_used,guard_triggered,status
+500,0,gaussian,0.995800,1.012670,10,exact_autograd,False,ok
+500,1,gaussian,0.998325,0.928482,10,exact_autograd,False,ok
+500,2,gaussian,0.891833,0.986902,10,exact_autograd,False,ok
+500,0,bimodal,0.999228,0.987200,10,exact_autograd,False,ok
+500,1,bimodal,0.992311,0.994397,10,exact_autograd,False,ok
+500,2,bimodal,0.999383,1.027159,10,exact_autograd,False,ok
+1000,0,gaussian,0.999568,1.006383,10,exact_autograd,False,ok
+1000,1,gaussian,0.999302,0.953974,10,exact_autograd,False,ok
+1000,2,gaussian,0.999735,0.969581,10,exact_autograd,False,ok
+1000,0,bimodal,0.955931,1.778940,10,exact_autograd,False,ok
+1000,1,bimodal,0.993821,1.177576,10,exact_autograd,False,ok
+1000,2,bimodal,0.998505,1.063358,10,exact_autograd,False,ok
+2000,0,gaussian,-0.006540,0.765355,10,exact_autograd,False,ok
+2000,1,gaussian,0.956470,0.910922,10,exact_autograd,False,ok
+2000,2,gaussian,0.997339,0.964720,10,exact_autograd,False,ok
+2000,0,bimodal,0.972932,0.588275,10,exact_autograd,False,ok
+2000,1,bimodal,0.923467,2.903147,10,exact_autograd,False,ok
+2000,2,bimodal,0.981981,0.612548,10,exact_autograd,False,ok
+5000,0,gaussian,0.870015,0.620875,10,exact_autograd,False,ok
+5000,1,gaussian,0.986924,0.977626,10,exact_autograd,False,ok
+5000,2,gaussian,0.997007,0.975937,10,exact_autograd,False,ok
+5000,0,bimodal,0.952795,1.126758,10,exact_autograd,False,ok
+5000,1,bimodal,0.950633,1.566525,10,exact_autograd,False,ok
+5000,2,bimodal,0.996987,0.820273,10,exact_autograd,False,ok
+8000,0,gaussian,0.988069,1.069129,10,exact_autograd,False,ok
+8000,1,gaussian,0.954147,0.419909,10,exact_autograd,True,ok
+8000,2,gaussian,0.961863,0.564638,10,exact_autograd,True,ok
+8000,0,bimodal,0.948292,1.841476,10,exact_autograd,False,ok
+8000,1,bimodal,0.942435,1.331992,10,exact_autograd,False,ok
+8000,2,bimodal,0.975536,0.694949,10,exact_autograd,False,ok
+9500,0,gaussian,0.972664,0.887930,10,exact_autograd,False,ok
+9500,1,gaussian,0.983723,0.434511,10,exact_autograd,True,ok
+9500,2,gaussian,-0.002076,0.887982,10,exact_autograd,True,ok
+9500,0,bimodal,0.825698,0.511781,10,exact_autograd,False,ok
+9500,1,bimodal,0.958388,1.057657,10,exact_autograd,False,ok
+9500,2,bimodal,0.711021,2.562303,10,exact_autograd,False,ok
+10500,0,gaussian,0.810057,0.766750,10,analytical_large_n,False,ok
+10500,1,gaussian,0.927377,0.313757,10,analytical_large_n,True,ok
+10500,2,gaussian,0.945284,0.387215,10,analytical_large_n,True,ok
+10500,0,bimodal,0.955336,1.145253,10,analytical_large_n,False,ok
+10500,1,bimodal,0.920809,1.756746,10,analytical_large_n,False,ok
+10500,2,bimodal,0.973271,1.071667,10,analytical_large_n,False,ok
+12000,0,gaussian,0.993863,0.955044,10,analytical_large_n,False,ok
+12000,1,gaussian,0.982637,0.328694,10,analytical_large_n,True,ok
+12000,2,gaussian,0.941901,0.328454,10,analytical_large_n,True,ok
+12000,0,bimodal,0.965876,1.096461,10,analytical_large_n,False,ok
+12000,1,bimodal,0.982003,1.071093,10,analytical_large_n,False,ok
+12000,2,bimodal,0.853446,2.692338,10,analytical_large_n,False,ok
+15000,0,gaussian,0.986382,0.820375,10,analytical_large_n,False,ok
+15000,1,gaussian,0.858652,0.292213,10,analytical_large_n,True,ok
+15000,2,gaussian,0.956316,0.315201,10,analytical_large_n,True,ok
+15000,0,bimodal,0.920057,1.640417,10,analytical_large_n,False,ok
+15000,1,bimodal,0.895741,1.751010,10,analytical_large_n,False,ok
+15000,2,bimodal,0.955799,3.184569,10,analytical_large_n,False,ok
+20000,0,gaussian,0.995874,0.656164,10,analytical_large_n,True,ok
+20000,1,gaussian,0.973048,0.196783,10,analytical_large_n,True,ok
+20000,2,gaussian,0.688630,0.221759,10,analytical_large_n,True,ok
+20000,0,bimodal,0.926917,2.560011,10,analytical_large_n,False,ok
+20000,1,bimodal,0.866700,1.475997,10,analytical_large_n,False,ok
+20000,2,bimodal,0.955457,1.909736,10,analytical_large_n,False,ok
+30000,0,gaussian,0.933955,0.580204,10,analytical_large_n,False,ok
+30000,1,gaussian,0.627780,0.131854,10,analytical_large_n,True,ok
+30000,2,gaussian,0.902804,0.341789,10,analytical_large_n,True,ok
+30000,0,bimodal,0.951740,1.581189,10,analytical_large_n,False,ok
+30000,1,bimodal,0.830918,1.191147,10,analytical_large_n,False,ok
+30000,2,bimodal,0.585203,2.159242,10,analytical_large_n,False,ok

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round37_s3_convergence_p3fix.csv

@@ -0,0 +1,25 @@
+n,seed,dist,cosine_fd,mag_ratio,k_proj,path_used,guard_triggered,h_hard,h_smooth,h_gap,status
+40000,0,gaussian,0.987850,0.920893,10,analytical_large_n,True,0.200887,0.200887,0.000000,ok
+40000,1,gaussian,-0.519445,1.030258,10,analytical_large_n,True,0.278578,0.278578,0.000000,ok
+40000,2,gaussian,0.970279,0.302844,10,analytical_large_n,True,0.180560,0.180560,0.000000,ok
+40000,0,bimodal,0.818326,1.587899,10,analytical_large_n,False,1.931522,1.931522,0.000000,ok
+40000,1,bimodal,0.571152,3.871155,10,analytical_large_n,False,1.930529,1.930529,0.000000,ok
+40000,2,bimodal,0.563411,4.205519,10,analytical_large_n,False,1.931129,1.931129,0.000000,ok
+43500,0,gaussian,0.980985,0.692190,10,analytical_large_n,True,0.199173,0.199173,0.000000,ok
+43500,1,gaussian,0.922113,0.191676,10,analytical_large_n,True,0.277210,0.277210,0.000000,ok
+43500,2,gaussian,0.949411,0.381802,10,analytical_large_n,True,0.178196,0.178196,0.000000,ok
+43500,0,bimodal,0.603045,0.360059,10,analytical_large_n,False,1.937251,1.937251,0.000000,ok
+43500,1,bimodal,0.721492,0.075686,10,analytical_large_n,False,1.935287,1.935287,0.000000,ok
+43500,2,bimodal,0.825189,1.416749,10,analytical_large_n,False,1.926904,1.926904,0.000000,ok
+45000,0,gaussian,0.869039,0.963885,10,analytical_large_n,False,0.172047,0.172047,0.000000,ok
+45000,1,gaussian,0.905200,0.243881,10,analytical_large_n,True,0.277211,0.277211,0.000000,ok
+45000,2,gaussian,0.902572,0.375361,10,analytical_large_n,True,0.178175,0.178175,0.000000,ok
+45000,0,bimodal,0.076642,10.958243,10,analytical_large_n,False,1.934259,1.934259,0.000000,ok
+45000,1,bimodal,-0.718661,4.058991,10,analytical_large_n,False,1.932619,1.932619,0.000000,ok
+45000,2,bimodal,0.887982,2.375318,10,analytical_large_n,False,1.930509,1.930509,0.000000,ok
+50000,0,gaussian,-0.097235,3.247465,10,analytical_large_n,False,0.170480,0.170480,0.000000,ok
+50000,1,gaussian,-0.173074,1.464650,10,analytical_large_n,True,0.274689,0.274689,0.000000,ok
+50000,2,gaussian,-0.028264,0.052645,10,analytical_large_n,False,0.118208,0.118208,0.000000,ok
+50000,0,bimodal,-0.048805,0.057015,10,analytical_large_n,False,1.934293,1.934293,0.000000,ok
+50000,1,bimodal,0.130395,0.045658,10,analytical_large_n,False,1.933133,1.933133,0.000000,ok
+50000,2,bimodal,-0.040132,0.046722,10,analytical_large_n,False,1.926262,1.926262,0.000000,ok

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round37_s3_convergence_post.csv

@@ -0,0 +1,73 @@
+n,seed,dist,cosine_fd,mag_ratio,k_proj,path_used,status
+500,0,gaussian,0.995800,1.012670,10,exact_autograd,ok
+500,1,gaussian,0.998325,0.928482,10,exact_autograd,ok
+500,2,gaussian,0.891833,0.986902,10,exact_autograd,ok
+500,0,bimodal,0.999228,0.987200,10,exact_autograd,ok
+500,1,bimodal,0.992311,0.994397,10,exact_autograd,ok
+500,2,bimodal,0.999383,1.027159,10,exact_autograd,ok
+1000,0,gaussian,0.999568,1.006383,10,exact_autograd,ok
+1000,1,gaussian,0.999302,0.953974,10,exact_autograd,ok
+1000,2,gaussian,0.999735,0.969581,10,exact_autograd,ok
+1000,0,bimodal,0.955931,1.778940,10,exact_autograd,ok
+1000,1,bimodal,0.993821,1.177576,10,exact_autograd,ok
+1000,2,bimodal,0.998505,1.063358,10,exact_autograd,ok
+2000,0,gaussian,-0.006540,0.765355,10,exact_autograd,ok
+2000,1,gaussian,0.956470,0.910922,10,exact_autograd,ok
+2000,2,gaussian,0.997339,0.964720,10,exact_autograd,ok
+2000,0,bimodal,0.972932,0.588275,10,exact_autograd,ok
+2000,1,bimodal,0.923467,2.903147,10,exact_autograd,ok
+2000,2,bimodal,0.981981,0.612548,10,exact_autograd,ok
+5000,0,gaussian,0.870015,0.620875,10,exact_autograd,ok
+5000,1,gaussian,0.986924,0.977626,10,exact_autograd,ok
+5000,2,gaussian,0.997007,0.975937,10,exact_autograd,ok
+5000,0,bimodal,0.952795,1.126758,10,exact_autograd,ok
+5000,1,bimodal,0.950633,1.566525,10,exact_autograd,ok
+5000,2,bimodal,0.996987,0.820273,10,exact_autograd,ok
+8000,0,gaussian,0.988069,1.069129,10,exact_autograd,ok
+8000,1,gaussian,0.954147,0.419909,10,exact_autograd,ok
+8000,2,gaussian,0.961863,0.564638,10,exact_autograd,ok
+8000,0,bimodal,0.948292,1.841476,10,exact_autograd,ok
+8000,1,bimodal,0.942435,1.331992,10,exact_autograd,ok
+8000,2,bimodal,0.975536,0.694949,10,exact_autograd,ok
+9500,0,gaussian,0.972664,0.887930,10,exact_autograd,ok
+9500,1,gaussian,0.983723,0.434511,10,exact_autograd,ok
+9500,2,gaussian,-0.002076,0.887982,10,exact_autograd,ok
+9500,0,bimodal,0.825698,0.511781,10,exact_autograd,ok
+9500,1,bimodal,0.958388,1.057657,10,exact_autograd,ok
+9500,2,bimodal,0.711021,2.562303,10,exact_autograd,ok
+10500,0,gaussian,0.810057,0.766750,10,analytical_large_n,ok
+10500,1,gaussian,0.927377,0.313757,10,analytical_large_n,ok
+10500,2,gaussian,0.945284,0.387215,10,analytical_large_n,ok
+10500,0,bimodal,0.955336,1.145253,10,analytical_large_n,ok
+10500,1,bimodal,0.920809,1.756746,10,analytical_large_n,ok
+10500,2,bimodal,0.973271,1.071667,10,analytical_large_n,ok
+12000,0,gaussian,0.993863,0.955044,10,analytical_large_n,ok
+12000,1,gaussian,0.982637,0.328694,10,analytical_large_n,ok
+12000,2,gaussian,0.941901,0.328454,10,analytical_large_n,ok
+12000,0,bimodal,0.965876,1.096461,10,analytical_large_n,ok
+12000,1,bimodal,0.982003,1.071093,10,analytical_large_n,ok
+12000,2,bimodal,0.853446,2.692338,10,analytical_large_n,ok
+15000,0,gaussian,0.986382,0.820375,10,analytical_large_n,ok
+15000,1,gaussian,0.858652,0.292213,10,analytical_large_n,ok
+15000,2,gaussian,0.956316,0.315201,10,analytical_large_n,ok
+15000,0,bimodal,0.920057,1.640417,10,analytical_large_n,ok
+15000,1,bimodal,0.895741,1.751010,10,analytical_large_n,ok
+15000,2,bimodal,0.955799,3.184569,10,analytical_large_n,ok
+20000,0,gaussian,0.995874,0.656164,10,analytical_large_n,ok
+20000,1,gaussian,0.973048,0.196783,10,analytical_large_n,ok
+20000,2,gaussian,0.688630,0.221759,10,analytical_large_n,ok
+20000,0,bimodal,0.926917,2.560011,10,analytical_large_n,ok
+20000,1,bimodal,0.866700,1.475997,10,analytical_large_n,ok
+20000,2,bimodal,0.955457,1.909736,10,analytical_large_n,ok
+30000,0,gaussian,0.933955,0.580204,10,analytical_large_n,ok
+30000,1,gaussian,0.627780,0.131854,10,analytical_large_n,ok
+30000,2,gaussian,0.902804,0.341789,10,analytical_large_n,ok
+30000,0,bimodal,0.951740,1.581189,10,analytical_large_n,ok
+30000,1,bimodal,0.830918,1.191147,10,analytical_large_n,ok
+30000,2,bimodal,0.585203,2.159242,10,analytical_large_n,ok
+50000,0,gaussian,-0.097235,3.247465,10,analytical_large_n,ok
+50000,1,gaussian,-0.173074,1.464650,10,analytical_large_n,ok
+50000,2,gaussian,-0.028264,0.052645,10,analytical_large_n,ok
+50000,0,bimodal,0.136339,0.241660,10,analytical_large_n,ok
+50000,1,bimodal,0.089725,0.060727,10,analytical_large_n,ok
+50000,2,bimodal,0.541547,0.369623,10,analytical_large_n,ok

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round37_s3_convergence_pre.csv

@@ -0,0 +1,73 @@
+n,seed,dist,cosine_fd,mag_ratio,k_proj,path_used,status
+500,0,gaussian,0.9958,1.0127,10,exact_autograd,ok
+500,1,gaussian,0.9983,0.9285,10,exact_autograd,ok
+500,2,gaussian,0.8918,0.9869,10,exact_autograd,ok
+500,0,bimodal,0.9992,0.9872,10,exact_autograd,ok
+500,1,bimodal,0.9923,0.9944,10,exact_autograd,ok
+500,2,bimodal,0.9994,1.0272,10,exact_autograd,ok
+1000,0,gaussian,0.9996,1.0064,10,exact_autograd,ok
+1000,1,gaussian,0.9993,0.9540,10,exact_autograd,ok
+1000,2,gaussian,0.9997,0.9696,10,exact_autograd,ok
+1000,0,bimodal,0.9559,1.7789,10,exact_autograd,ok
+1000,1,bimodal,0.9938,1.1776,10,exact_autograd,ok
+1000,2,bimodal,0.9985,1.0634,10,exact_autograd,ok
+2000,0,gaussian,-0.0065,0.7654,10,exact_autograd,ok
+2000,1,gaussian,0.9565,0.9109,10,exact_autograd,ok
+2000,2,gaussian,0.9973,0.9647,10,exact_autograd,ok
+2000,0,bimodal,0.9729,0.5883,10,exact_autograd,ok
+2000,1,bimodal,0.9235,2.9031,10,exact_autograd,ok
+2000,2,bimodal,0.9820,0.6125,10,exact_autograd,ok
+5000,0,gaussian,0.8700,0.6209,10,exact_autograd,ok
+5000,1,gaussian,0.9869,0.9776,10,exact_autograd,ok
+5000,2,gaussian,0.9970,0.9759,10,exact_autograd,ok
+5000,0,bimodal,0.9528,1.1268,10,exact_autograd,ok
+5000,1,bimodal,0.9506,1.5665,10,exact_autograd,ok
+5000,2,bimodal,0.9970,0.8203,10,exact_autograd,ok
+8000,0,gaussian,0.9881,1.0691,10,exact_autograd,ok
+8000,1,gaussian,0.9541,0.4199,10,exact_autograd,ok
+8000,2,gaussian,0.9619,0.5646,10,exact_autograd,ok
+8000,0,bimodal,0.9483,1.8415,10,exact_autograd,ok
+8000,1,bimodal,0.9424,1.3320,10,exact_autograd,ok
+8000,2,bimodal,0.9755,0.6949,10,exact_autograd,ok
+9500,0,gaussian,0.9727,0.8879,10,exact_autograd,ok
+9500,1,gaussian,0.9837,0.4345,10,exact_autograd,ok
+9500,2,gaussian,-0.0021,0.8880,10,exact_autograd,ok
+9500,0,bimodal,0.8257,0.5118,10,exact_autograd,ok
+9500,1,bimodal,0.9584,1.0577,10,exact_autograd,ok
+9500,2,bimodal,0.7110,2.5623,10,exact_autograd,ok
+10500,0,gaussian,0.8101,0.7668,10,analytical_large_n,ok
+10500,1,gaussian,0.9274,0.3138,10,analytical_large_n,ok
+10500,2,gaussian,0.9453,0.3872,10,analytical_large_n,ok
+10500,0,bimodal,0.9553,1.1460,10,analytical_large_n,ok
+10500,1,bimodal,0.9208,1.7575,10,analytical_large_n,ok
+10500,2,bimodal,0.9733,1.0757,10,analytical_large_n,ok
+12000,0,gaussian,0.9939,0.9551,10,analytical_large_n,ok
+12000,1,gaussian,0.9826,0.3287,10,analytical_large_n,ok
+12000,2,gaussian,0.9419,0.3285,10,analytical_large_n,ok
+12000,0,bimodal,0.9659,1.0972,10,analytical_large_n,ok
+12000,1,bimodal,0.9820,1.0718,10,analytical_large_n,ok
+12000,2,bimodal,0.8534,2.6940,10,analytical_large_n,ok
+15000,0,gaussian,0.9864,0.8204,10,analytical_large_n,ok
+15000,1,gaussian,0.8587,0.2922,10,analytical_large_n,ok
+15000,2,gaussian,0.9563,0.3152,10,analytical_large_n,ok
+15000,0,bimodal,0.9201,1.6414,10,analytical_large_n,ok
+15000,1,bimodal,0.8957,1.7522,10,analytical_large_n,ok
+15000,2,bimodal,0.9558,3.2160,10,analytical_large_n,ok
+20000,0,gaussian,0.9959,0.6562,10,analytical_large_n,ok
+20000,1,gaussian,0.9730,0.1968,10,analytical_large_n,ok
+20000,2,gaussian,0.6886,0.2218,10,analytical_large_n,ok
+20000,0,bimodal,0.9269,2.5615,10,analytical_large_n,ok
+20000,1,bimodal,0.8667,1.4770,10,analytical_large_n,ok
+20000,2,bimodal,0.9555,1.9105,10,analytical_large_n,ok
+30000,0,gaussian,0.9340,0.5802,10,analytical_large_n,ok
+30000,1,gaussian,0.6278,0.1319,10,analytical_large_n,ok
+30000,2,gaussian,0.9028,0.3418,10,analytical_large_n,ok
+30000,0,bimodal,0.9517,1.5823,10,analytical_large_n,ok
+30000,1,bimodal,0.8309,1.1916,10,analytical_large_n,ok
+30000,2,bimodal,0.5852,2.1607,10,analytical_large_n,ok
+50000,0,gaussian,-0.0972,3.2475,10,analytical_large_n,ok
+50000,1,gaussian,-0.1731,1.4646,10,analytical_large_n,ok
+50000,2,gaussian,-0.0283,0.0526,10,analytical_large_n,ok
+50000,0,bimodal,0.1363,0.2417,10,analytical_large_n,ok
+50000,1,bimodal,0.0897,0.0607,10,analytical_large_n,ok
+50000,2,bimodal,0.5415,0.3699,10,analytical_large_n,ok

diffcb-0.1.13/02_projects/01_dcb_proposal/04_analysis/results/round37_s3_guard_clear_stats.txt

@@ -0,0 +1,7 @@
+Round 37 S3 — Guard-Clear Cosine Statistics (P2.2)
+=======================================================
+exact_autograd   (n≤10K, guard_clear)
+  mean cosine = 0.9289  min cosine = -0.0065  n_cells = 32
+analytical_large_n (10K<n≤30K, guard_clear)
+  mean cosine = 0.9086  min cosine = 0.5852  n_cells = 19
+=======================================================