Increasing precision when parsing float values using textread()

Hello,

I am using textread() function to parse text values; I have the following line which works well:

[V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11  V12  V13  V14] = textread(file_1,'%f %f %f %f %f %f %f %f %f %f %f %f %f %f','headerlines',5);

The problem I am facing is that float values in the original text contain 6 digits after decimal point, while textread() returns only 4 digits after the decimal point.

For example:

Original value: -0.785834

Value returned by textread(): -0.7858

Please, suggest how to get the full precision of floats when using textread()

Thanks in advance

[Iain]

Are you sure that the value returned is EXACTLY -0.7858, or is that just what matlab is displaying?

Answer 1

[Stephen23]

I just created a new text file containing only the text '0.1234567890123456789', and named it 'temp.txt'. Then I ran this:

>> A = textread('temp.txt','%f');
>> fprintf('%.20f\n',A)
0.12345678901234568000

This shows that textread has read that number up to the double precision limit (~17 sigfig).

The "problem" is likely to be how MATLAB displays numbers up to four decimal digits, much to the confusion of all MATLAB learners, who wonder where their digits have gone to. Don't worry, those digits are still there! Have a look at format for more information on how to change the displaying of numbers.

Also as Adam mentions, the documentation reccomends using textscan instead of textread.

[Kamen]

Thanks, indeed the problem was with formatting when displaying the result!

Answer 2

[Adam]

The help page (R2014b) suggests you should use textscan now instead of textread, but I assume the following works for either while textread still exists.

You can specify the precision you want for your floats by e.g.

%.6f

to specify the number of values after the decimal place.

[Kamen]

Thanks for the great help, it worked!