READ ME File for dataset 'Convolutional Neural Networks for Mode On-Demand High Finesse Optical Resonator Design'

Dataset DOI: 10.5258/SOTON/D2774

ReadMe Author: Peter Horak, University of Southampton

This dataset supports the publication: 
	"Convolutional Neural Networks for Mode On-Demand High Finesse Optical Resonator Design" 
	D.V. Karpov, S. Kurdiumov, and P. Horak
	accepted for publication in Scientific Reports (6/9/2023)


Contents
+++++++++

This dataset contains the data of Figures 3 and 4, and the numerical model to recreate Figures 5-7of the associated publication


The following data files are plain text files in CSV format (comma-serparated values), readable by any text editor, spreadsheet editor, or numerical software (we use Microsoft Excel). All data descriptions and units are given as headers in the csv files.

figure3.csv: data for Fig. 3, true and predicted amplitude and period (in um) of the validation dataset 

figure4.csv: data for Fig.4: mean error of the training dataset and of the validation dataset versus epoch number


"one peak" and "two peaks" directories:
Data in these directories contain the fully trained neural network structures and weights. Figures 5, 7 where generated from "one peak", Figure 6 from "two peaks".

The model is stored in the standard Keras/Tensorflow format: The model architecture, and training configuration (including the optimizer, losses, and metrics) are stored in saved_model.pb. The weights are saved in the variables/ directory.

For example, the "one peak" model can be loaded in Python by:
	from tensorflow import keras
	model = keras.models.load_model("one peak")

The predicted pair of (Amplitude, Period) for a target field E (cooperativity C on a spatial grid in a 100x150 array, formatted to Keras/Tensorflow style) is then obtained as:
	predictions = model.predict(E)


Additional information:
+++++++++++++++++++++++

Geographic location of data collection: University of Southampton, U.K.

Related projects:
EPSRC Hub in Quantum Computing and Simulation (EP/T001062/1)

Dataset available under a CC BY 4.0 licence

Publisher: University of Southampton, U.K.

Date: September 2023