Simulated Resistive Drift of Memristive Devices
- Citation Author(s):
-
Waleed El-Geresy
(Imperial College London)
Christos Papavassiliou
(Imperial College London)
Deniz Gündüz
(Imperial College London)
- Submitted by:
- Waleed El-Geresy
- Last updated:
- DOI:
- 10.21227/edzz-7d60
- Data Format:
68 views
Abstract
A dataset of simulated resistive drift series for an illustrative stochastic memristor.
Dataset Description
The memristor has an equilibrium resistance of approximately 500kΩ.
5000 series are generated with starting resistances sampled uniformly from the range [100Ω, 750kΩ].
Each series consists of 1001 datapoints, with the first (zeroth) point corresponding to the initial resistance, and subsequent points sampled at subsequent timesteps.
Dataset Creation
The simulations were conducted using the event-based memristor model in [1], with parameters as follows, as specified in [2]:
N = 1.8 × 10^6
n_thresh = 0
g_parallel = 1 × 10^-6
g_step = 1 × 10^-8
V_a = 0.256
V_off = 0.2533
References
[1] Waleed El-Geresy, Christos Papavassiliou, Deniz Gündüz (2024), Event-Based Simulation of Stochastic Memristive Devices for Neuromorphic Computing, arXiv:2407.04718 [cs.ET]
[2] Waleed El-Geresy, Christos Papavassiliou, Deniz Gündüz (2024), Delay Conditioned Generative Modelling of Resistive Drift in Memristors, arXiv:2408.01539 [cs.ET]
Instructions:
Usage
The dataset is supplied as a numpy array of dimension (5000, 1001), and can be loaded using np.load(...).
