search

arnab37 / Total-Electron-Ionization-Cross-Sections

It is created to know the total electron ionization cross section ratios of organic molecule w.r.t. n-hexane
Other
1 stars 0 forks source link

Array indices must be positive integers or logical values #1

Open actionjay10 opened 1 year ago

actionjay10 commented 1 year ago

clc; clear all; close all;

AA=load('Make_a_txt_file_after_copying_the_descriptors.txt'); AA=AA';

load BoseWestmorelandNet.mat Y=zeros(size(AA,2),500);

for k=1:500 Ys=BoseWestmorelandNet{k}(AA); Y(:,k)=Ys; end

for i=1:size(AA,2) TEICS70_out(i,1)=mean(Y(i,:)); %%%%% n-hexane's cross section was assigned to be 20.0 x 10^-16 cm^2 previously. TEICS70_out(i,2)=std(Y(i,:))/sqrt(500)*1.6479; %%%% 1.6479 is the multipler for the 95% confidence level from t-stat end

xlswrite('Output.xlsx',TEICS70_out);

Giving error "Array indices must be positive integers or logical values" on line 15, Ys=BoseWestmorelandNet{k}(AA);

arnab37 commented 1 year ago

I think the error is appearing because of the mismatch in Matlab version. You can retrain the NN object using the following subroutine: GroupNN.txt, you need to convert it to GroupNN.m before running it. You may also adopt it to the newer Matlab version.

Also, we found new errors in the training data: 2-Buten-1-ol, number of cis corrections: 0.5 (corrected value = 0) perfluorobuta-1,3diene, number of cis corrections: 0.5 (corrected value = 0) Nitric Acid: Number of N(sp) atoms: 0.5 (corrected value = 0)

GroupsNN.txt Prediction.txt Sens_Dummy.txt