Memristor Models

Python implementations and simulations of HP Labs Ion Drift and Yakopcic memristor models.

Files

Frontend

• simulate.py runs experiment based on simulation of the HP Labs ion drift memristor model [1] or of the Yakopcic memristor model [2]. The latter can be used to match characterisations for devices from literature [3-5]. The model is simulated, noise is added to the current measurement, and this noisy measurement is used to regress the model parameters to obtain a model capable of reproducing the original data.

  • -e selects the experiment to run [hp_sine/hp_pulsed/oblea_sine/oblea_pulsed/miao/jo].
  • -s selects the solver to use to simulate the model [LSODA/EU/RK4]; more than one can be passed, and the experiment will be run for each one.
  • --video generates a video of the simulation of the memristor model.

• data_import.py reads the .csv files containing the I-V measurements and outputs them as pickled files in the . /pickles folder, and as plots in the ./plots folder. The data for our memristive device is available on request.

• hp_labs_interactive.py launches a matplotlib-based GUI that lets the user experiment in real time with changing the HP Labs ion drift memristor model [1] parameters.

• yakopcic_interactive.py launches a matplotlib-based GUI that lets the user experiment in real time with changing the Yakopcic model [2] parameters.

• fit_interactive.py launches a Tk-based GUI application that lets the user load a pickled file created by data_import.py, and fit the updated Yakopcic model [6] to the real device's behaviour.

  

• fit_yakopcic.py loads a pickled file created by data_import.py, and attempts to reproduce the semi-automated model parameter fitting procedure outlined in [6].

• load_and_fit.py loads a pickled file created by data_import.py, and attempts a blind regression on model the Yakopcic model parameters.

Backend

• models.py contains implementation of the HP Labs ion drift memristor model [1], Yakopcic generalised memristor model [2], and an updated version of Yakopcic's model [6].

• experiments.py contains the experiment definitions used by simulate.py.

• functions.py contains various helper functions used throughout the code.

References

[1] Yang, J. J. et al. Memristive switching mechanism for metal/oxide/metal nanodevices. Nat Nanotechnol 3, 429–433 (2008).

[2] Yakopcic, C., Taha, T. M., Subramanyam, G., Pino, R. E. & Rogers, S. A Memristor Device Model. Ieee Electr Device L 32, 1436–1438 (2011).

[3] Oblea, A. S., Timilsina, A., Moore, D. & Campbell, K. A. Silver Chalcogenide Based Memristor Devices. 2010 Int Jt Conf Neural Networks Ijcnn 1, 1–3 (2010).

[4] Miao, F. et al. Anatomy of a Nanoscale Conduction Channel Reveals the Mechanism of a High‐Performance Memristor. Adv Mater 23, 5633–5640 (2011).

[5] C. Yakopcic, T. M. Taha, G. Subramanyam, and R. E. Pino, “Generalized Memristive Device SPICE Model and its Application in Circuit Design,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 32(8) August, 2013 pp. 1201-1214.

[6] Yakopcic, C. et al. Memristor Model Optimization Based on Parameter Extraction From Device Characterization Data. Ieee T Comput Aid D 39, 1084–1095 (2020).