Investigation on the Impact of Memristive Device Variability on Uncoupled Memristive Cellular Nonlinear Networksdoi:10.26165/JUELICH-DATA/R3LHENJülich DATA2026-09-251Salwa, Yasmeen Neyaz, 2026, "Investigation on the Impact of Memristive Device Variability on Uncoupled Memristive Cellular Nonlinear Networks", https://doi.org/10.26165/JUELICH-DATA/R3LHEN, Jülich DATA, V1Investigation on the Impact of Memristive Device Variability on Uncoupled Memristive Cellular Nonlinear Networksdoi:10.26165/JUELICH-DATA/R3LHENSalwa, Yasmeen NeyazJülich DATASalwa, Yasmeen NeyazSalwa, Yasmeen Neyaz2026-04-28EngineeringPhysicsMemristor, M-CNN, EDGE detection<h2>Description</h2> <p> The dataset contains all raw data and Python scripts (.py) used to create the figures in the manuscript "Investigation on the Impact of Memristive Device Variability on Uncoupled Memristive Cellular Nonlinear Networks". The manuscript is not yet publicly available; we will add a reference once it is published. </p> <p> It includes experimental data of each Local Rule (LR1–LR3) for five tested memristive devices (D1–D5) for the EDGE detection operation of a Memristive Cellular Nonlinear Network (M-CNN). It also includes simulated fitted data from Cadence for the corresponding experimental results. Furthermore, the dataset includes simulation results of EDGE detection of a diamond-shaped 16×16 image under three scenarios: observed experimental device-to-device (D2D) variability, global ±20% RC deviation with D2D variability, and local ±2% RC mismatch with D2D variability. </p> <p> The dataset also includes simulation results for four additional, harder 16×16 test images, stored as task files: task1 is a comb structure, task2 a binarized digit "3", task3 a periodic pinhole array and task4 a single-pixel hole. We tested these under the same three scenarios, with results in the subfolders <code>D2D_only</code>, <code>global_20pct</code>, and <code>2pct_dev_plusd2d</code>. The subfolder <code>4pct_pinholes</code> contains a 4% boundary error case for the pinhole image. </p> <p> The data is structured in subfolders according to the individual figures. The Python scripts read the raw data files and reproduce the figures. Update the file paths in the scripts before execution. </p> <p> <h2>Folders</h2> <ul> <li><code>Figure 4a Expt_and_sim_data_LocalRules/</code>: experimental and simulated transient data per Local Rule and device</li> <li><code>Figure 5_Diamond_shape_edge_detection/</code>: EDGE detection of the diamond-shaped image under the three scenarios</li> <li><code>Figure 3_Extended_test_images/</code>: EDGE detection of the four additional test images (task1–task4)</li> <li><code>D2D_only/</code>: observed experimental D2D variability</li> <li><code>global_20pct/</code>: global ±20% RC with D2D variability</li> <li><code>2pct_dev_plusd2d/</code>: local ±2% RC with D2D variability</li> <li><code>4pct_pinholes/</code>: 4% boundary error case for the pinhole image</li> </ul> </p>CC0 Waiver1776762958797_mcnnlr001n_Ch4_LR1.csvtext/csv20pctglobal_rc.csvtext/csv2pctchange.csvtext/csv4_tasks_deviations.pytext/x-pythoncomb_expected_output.scsapplication/scvp-cv-responsecomb_stimulus.scsapplication/scvp-cv-responsed2d_task1.csvtext/csvd2d_task2.csvtext/csvd2d_task3.csvtext/csvd2d_task4.csvtext/csvdiamond_expected_output.scsapplication/scvp-cv-responsediamond_stimulus.scsapplication/scvp-cv-responsedigit3-thick_expected_output.scsapplication/scvp-cv-responsedigit3-thick_stimulus.scsapplication/scvp-cv-responseError_4_pct_pinhole_code.pytext/x-pythonglobal_task1.csvtext/csvglobal_task2.csvtext/csvglobal_task3.csvtext/csvglobal_task4.csvtext/csvlocal10pct.csvtext/csvlocal15pct.csvtext/csvlocal20pct.csvtext/csvlocal_2pct_rx1633.scsapplication/scvp-cv-responselocal_2pct_task1.csvtext/csvlocal_2pct_task2.csvtext/csvlocal_2pct_task3.csvtext/csvlocal_2pct_task4.csvtext/csvlocal3pct.csvtext/csvlocal_4pct_task1.csvtext/csvlocal_4pct_task2.csvtext/csvlocal_4pct_task3.csvtext/csvlocal_4pct_task4.csvtext/csvlocal5pct.csvtext/csvLR1_D1_simulation.csvtext/csvLR1_D1_x12y30.csvtext/csvLR1_D2_simulation.csvtext/csvLR1_D2_x13y30(1).csvtext/csvLR1_d3_simulation (2).csvtext/csvLR1_D3_x11y30.csvtext/csvLR1_D4_simulation.csvtext/csvLR1_D4_x17y30.csvtext/csvLR1_D5_simulation.csvtext/csvLR2_D1_simulation.csvtext/csvLR2_D1_VX.csvtext/csvLR2_D2_simulation.csvtext/csvLR2_D2_VX.csvtext/csvLR2_D3_simulation.csvtext/csvLR2_D3_VX.csvtext/csvLR2_D4_simulation.csvtext/csvLR2_D4_VX_return.csvtext/csvLR2_D5_simulation.csvtext/csvLR3_D1_simulation.csvtext/csvLR3_D1_x19y30_2.csvtext/csvLR3_D2_simulation.csvtext/csvLR3_D3_ninit_120m.csvtext/csvLR3_D3_x22y30.csvtext/csvLR3_D4_simulation.csvtext/csvLR3_D4_x23y30.csvtext/csvLR3_D5_simulation1.csvtext/csvLR3_D5_x24y30.csvtext/csvLRthreebetween_Ch4_LR3.csvtext/csvoutput_of_images_16_16.csvtext/csvpinholes_4_expected_output.scsapplication/scvp-cv-responsepinholes_4_stimulus.scsapplication/scvp-cv-responseplot_fitting_curves.pytext/x-pythonpy_plot_edge_detection.pytext/x-pythonsingle_hole_expected_output.scsapplication/scvp-cv-responsesingle_hole_stimulus.scsapplication/scvp-cv-responsew20mcnndatanew_Ch3_LR2.csvtext/csv