Unnecessary rounding when reading an xlsx

Question

Emily Barber 2020-7-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/572347-unnecessary-rounding-when-reading-an-xlsx

评论： Adam Danz 2020-7-29

I currently need to input xlsx data into matlab to a precision of 7 decimal places, and then interpolate data inbetween those points. However, I'm relatively sure that the data is only being inputted to 4 decimal places, as the interp1 function is giving back the error: "The grid vectors must contain unique points" (it uses griddedInterpolant to solve the answer). This would happen if the program is rounding to 4 decimals, as I have some data that is 0.0000100, 0.0000106, etc.. Everything I see online, though, says that MATLab imports up to 15 decimal places. Is there something else that could be wrong here?

(Additionally, I have the format set to long)

Thanks.

6 个评论
显示 4更早的评论隐藏 4更早的评论

Emily Barber 2020-7-29

在 MATLAB Online 中打开

The code I used for this purpose. I believe I included all variables in this shortened version but if not, they are all present in the total code

        EElas=xlsread('U238Elastic');     % Read in data
            nRow = size(EElas, 1);        % Removes any not-unique 2D points
            c = cell(1, nRow);
               for iRow = 1:nRow
                    c{iRow} = unique(EElas(iRow, :), 'stable');
               end
            EE = cat(1, c{:});
            EElasInput=EE(:,1);           % Seperates out into inputs and outputs
            EElasOutput=EE(:,2);
            
            E1=10^6
            
                wElas=ismember(E1,EElasInput);  % If it's already a point, no need to interpolate
                if wElas==1
                    v=find(EElasInput==E1);
                    SigEl=EElasOutput(v);
                else                            % If not, try interpolating
                    SigEl=interp1(EElasInput,EElasOutput,E1)*10^-28;   % The error occurs in this line
                     if isnan(SigEl)            % If it still doesn't work, extrapolate
                         SigEl=interp1(EElasInput,EElasOutput,E1,'linear','extrap')*10^-28;
                     end
                end

Adam Danz 2020-7-29

编辑：Adam Danz 2020-7-29

在 MATLAB Online 中打开

Clearly the xlsread output is greater than 4dp precision

format long
>> EElas(1:10,:)
ans =
000010000000000  35.397709999999996
000010625000000  34.404020000000003
000011250000000  33.496060000000000
000011875000000  32.662320000000001
000012500000000  31.893370000000001
000013750000000  30.519690000000001
000015000000000  29.325990000000001
000016250000000  28.276740000000000
000017500000000  27.345490000000002
000018750000000  26.512090000000001

I don't understand the point of the iRow loop. If any of the rows contained duplicate values, the code would break just after the loop when you contatenate the "c" values since at least one cell would only contain 1 value.

I haven't looked too deeply at the rest of the code but I ran it and it does not produce an error. So I'm a but lost as to how I can help. My answer was written before seeing your updated code so my answer may not be entirely relevant. However, I do recommend using readmatrix instead of xlsread.

Emily Barber 2020-7-29

I copied exactly what I sent you into a function and it worked, which means that something's likely wrong with my code as a whole (as it didn't work once placed in the function). I'll try to keep working with it, thank you for your help.

Adam Danz 2020-7-29

Feel free to continue to discussion if you find the source the problem and have trouble fixing it. If the problem is greatly different from the conversation so far, it would be better to start a new question and you can provide a link to it here in the comment section if you'd like.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Adam Danz 2020-7-29

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/572347-unnecessary-rounding-when-reading-an-xlsx#answer_472567

编辑：Adam Danz 2020-7-29

在 MATLAB Online 中打开

Thanks for providing the raw data.

I'm not seeing what you're seeing. My data exporation steps are below.

% Read-in the data
x = readmatrix('U238Elastic.xlsx'); 
% Check out the first few rows
x(1:10,:)
% RESULT (format: shortg)
%     ans =
%             1e-05       35.398
%        1.0625e-05       34.404
%         1.125e-05       33.496
%        1.1875e-05       32.662
%          1.25e-05       31.893
%         1.375e-05        30.52
%           1.5e-05       29.326
%         1.625e-05       28.277
%          1.75e-05       27.345
%         1.875e-05       26.512

Clearly the precision is greater than 4dp.

Are there any duplicate rows?

% Produce index of unique rows of x
[~, unqIdx] = unique(x,'rows');
isUnq = ismember(1:size(x,1), unqIdx); 
% Compute the percentage of rows that are unique
percentUnq = sum(isUnq)/size(x,1)*100
% RESULT
%     percentUnq =
%        100

It seems that all of the coordinates are unique. Maybe you're reading in the data differently or perhaps you're looking at a different pair of values than this.