Training effect of feedforward neural networks with different inputs and 2pF parameter prediction results for (near) stable nuclei

The dataset contains five Unicode Origin Graph (.opju) files, which can be opened with the Origin 2021 software. Details of the five data files are as follows:1、 Fig.2-data.opjuThe file named "Fig.2-data.opju" contains the original information from Figure 2 in the associated article. It contains two workbooks. One is the density distributions and error bands information for the training set, and the other is for the validation set. Each workbook contains several data tables (more tables in the training set and less tables in the validation set). Each data table is named by the nuclei in the corresponding data set (training set or validation set), and they contain density information and error band information for the corresponding nuclei. Each data table has six columns and one thousand rows. Columns 1 through 6 are: the radial coordinate r of the nucleus, in fm; The charge density distribution predicted by FNN-3I; Error band value of charge density predicted by FNN-3I; The charge density distribution predicted by FNN-4I; Error band value of charge density predicted by FNN-4I; The value of charge density distribution obtained by experiment (2pF model). Each row of data pairs should be the density value or error band value at r. The practice of obtaining these data can be found in the associated article.2、 Fig.3-data.opjuThe file named "Fig.3-data.opju" contains the original information from Figure 3 in the associated article. It contains two workbooks. One is the density distributions information for the training set, and the other is for the validation set. Each workbook contains several data tables (more tables in the training set and less tables in the validation set). Each data table is named by the nuclei in the corresponding data set (training set or validation set). Each data table has six columns and one thousand rows. Columns 1 through 6 are: the radial coordinate r of the nucleus, in fm; The average charge density distribution value obtained by density averaging method; The average charge density distribution value obtained by parameter averaging method; The value of the charge density distribution derived from the network with the smallest loss function; The value of the charge density distribution derived from the network with the largest loss function; The value of charge density distribution obtained by experiment (2pF model). Each row of data pairs should be the density value at r. The practice of obtaining these data can be found in the associated article.3、 Fig.4-data.opjuThe file named "Fig.4-data.opju" contains the original information from Figure 4 in the associated article. It contains two workbooks. One is the density distributions and error bands information for the training set, and the other is for the validation set. Each workbook contains several data tables (more tables in the training set and less tables in the validation set). Each data table is named by the nuclei in the corresponding data set (training set or validation set). Each data table has six columns and one thousand rows. Columns 1 through 6 are: the radial coordinate r of the nucleus, in fm; The average charge density distribution value obtained by density averaging method; The error band value obtained by density averaging method; The average charge density distribution value obtained by parameter averaging method; The error band value obtained by parameter average method; The value of charge density distribution obtained by experiment (2pF model). Each row of data pairs should be the density value or error band value at r. The practice of obtaining these data can be found in the associated article.4、 Fig.5 6 7-data.opjuThe file named "Fig.5 6 7-data.opju" contains the original information from Figures 5 6 and 7 in the associated article. It consists of two tables, one containing the training result information for the training set and the other containing the prediction results for the prediction set. The table related to the training set has 15 columns and 86 rows, in which each column represents: proton number; Neutron number; The experimental value of parameter c; The experimental value of parameter z; The predicted value of parameter c; The predicted value of parameter z; Experimental value of charge radius R (2pF model); The predicted value of charge radius R; The difference between the experimental value and the predicted value of charge radius R; Experimental values of the second moment of charge (2pF model); The predicted value of the second moment of charge; The difference of the second moment of charge; Experimental values of fourth moment of charge (2pF model); The predicted value of fourth moment of charge; The difference of the fourth moment of charge. Each row represents the data of a nucleus. The table associated with the prediction set has 284 rows in 7 columns, each of which contains data on a single nucleus. The seven columns represent the number of protons; Neutron number; Mass number; The predicted value of parameter c; The predicted value of parameter z; The predicted value of charge radius R; The experimental value of the charge radius R.5、 Fig.8-data.opjuThe file named "Fig.8-data.opju" contains the original information from Figure 8 in the associated article. It has only one workbook, which contains several tables, each containing information about a calcium isotope and named by that isotope. Each table has 3 columns and 1000 rows. The three columns are: radial coordinate r, unit is fm; Charge density distribution value; Charge density error band value. Each row corresponds to a drawing coordinate point and error band.