Is there a way to retain the precision of the zero?

Question

h s 2017-2-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/327020-is-there-a-way-to-retain-the-precision-of-the-zero

编辑： Stephen23 2017-2-26

I have a large string array with 0.000000, but when I convert it to a matrix through 'num2str' it is converted into 0. I have tried format long, etc. but it only affects the numbers other than zero.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Stephen23 2017-2-26

编辑：Stephen23 2017-2-26

在 MATLAB Online 中打开

"I have tried format long, etc. but it only affects the numbers other than zero"

Changing the format makes absolutely no difference to the precision of the variable stored in memory, only to how they are displayed. Here is an example:

>> X = 2.009
X =
                     2.009
>> format bank
>> X
X =
          2.01
>> X*100
ans =
        200.90

The last step above shows that format bank does not change the precision of the stored variable whatsoever: it still retains all digits of precision that it had when it was defined.

Zero is stored with exactly the same precision as any other values of the same class. You cannot change its stored precision for one class, only its displayed precision.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2017-2-26

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/327020-is-there-a-way-to-retain-the-precision-of-the-zero#answer_256400

If the string representation is just '0.000000', the str2num result double(0) is nnumerically as precise as it’s going to get! It has the full precision internally.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 2

Walter Roberson 2017-2-26

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/327020-is-there-a-way-to-retain-the-precision-of-the-zero#answer_256401

在 MATLAB Online 中打开

Assuming that numbers are not in scientific format,

num_decimals = cellfun(@length, regexprep(YourCell, '^\d*\.', ''));
as_num = str2double(YourCell);

Now, as_num is the numeric value of YourCell, and num_decimals is the number of digits after the decimal point associated with each entry.

If at some point you need to print them back out with the original number of decimal places, then

delimiter = ' ';
fmt = [repmat( ['%.*f', delimiter], 1, size(as_num,2)-1), '%.*f\n'];
temp = [ num2cell(num_decimals(:)) .'; num2cell(as_num(:)) .'];
fprintf( fmt, temp{:} )

Remember, the number of 0's printed out is a matter of display, not of what is actually stored. What is actually stored is binary rather than decimal.