This readme file was generated on 2024/10/22 by Dary Lu


-------------------
GENERAL INFORMATION
-------------------

Title of Dataset: Data from: Deep learning for accelerated all-dielectric metasurface design


Author Contact Information (Name, Institution, Email, ORCID)

	Principal Investigator: Willie John Padilla
	Institution: Duke University 
	Email: willie.padilla@duke.edu
	ORCID: https://orcid.org/0000-0001-7734-8847

*Date of data collection (single date, range, approximate date): 2019 

*Funding and grant numbers: Department of Energy (DOE) (DE-SC0014372); Alfred P. Sloan Foundation through the Duke University Energy Data Analytics Fellowship; Duke University Energy Initiative.

--------------------
DATA & FILE OVERVIEW
--------------------

File list (filenames, directory structure (for zipped files) and brief description of all data files):
data_file_*.csv (from 1 to 22)

*Relationship between files, if important for context:  
Each file is a subset of the overall dataset. Together, they cover the full range of simulated spectra.


--------------------------
METHODOLOGICAL INFORMATION
--------------------------

*Software- or Instrument-specific information needed to interpret the data, including software version numbers, packages or other dependencies: CST 2019 

*Environmental/experimental conditions: Simulations were performed on a computer to generate data.

--------------------------
DATA-SPECIFIC INFORMATION 
--------------------------
 
Our work here focuses on a supercell ADM comprised of four cylindrical unit cells. Each unit-cell cylinder is parameterized by a 
radius and a height (two parameters), yielding a 8-dimensional parameterization, x, for the geometry of the supercell. 
We select silicon (Si) for the metasurface material and, with no loss of generality, embed it within vacuum for the purposes of faster training set acquisition. 
After that, we collected the training dataset by running different x-values through the simulator to obtain their corresponding s values. 

To avoid introducing bias towards any particular geometry, we randomly sampled the full geometric parameter space. 
A total of approximately 21,000 simulated spectra were collected. Each CSV file contains data structured as follows:

Columns 1-4: Radiuses of the unit cells
Columns 5-8: Heights of the unit cells
Subsequent columns: Corresponding spectra of the metasurfaces

-------------------------
USE and ACCESS INFORMATION 
--------------------------
Data License:Creative Commons CC0 1.0 Universal

Other Rights Information: 

To cite the data: Nadell, C. C., Huang, B., Malof, J. M., & Padilla, W. J. (2024). Data from: Deep learning for accelerated all-dielectric metasurface design. Duke Research Data Repository. https://doi.org/10.7924/r44j0qd59