generate (not round) 2 decimal numbers

Hello, in the first part of more complicated calculation I need some restriction to "rand" or so. My point is a reduction of unnecessary calculations and occurence of crowding results by simplification of the random number choice to 2 decimals in the first place. Is it possible?

Thank you

[José-Luis]

What do you mean by simplify calculations? If you are using the double format anyway, I very much doubt that the computational requirements will differ much. A probably better approach is only to round when you have your final result.

[P K]

Even if done in parts I am dealing with bigger set of numbers, if you imagine the set it selects from, it is better to start with 2decimal numbers than 10decimal or what "rand" realy does. When I put in 1e7 Xs to not run out of RAM, it just does not works for me well now

[José-Luis]

The you need to use a format other than double. I am afraid "reducing" the number of decimals will do nothing for you, the number will still be stored as a double and will continue to use the same amount of memory.

Please explain what you are trying to achieve in order to help you better.

Answer 1

[KSSV]

Are you looking for something like this?

sprintf('%3.2f',rand(1,1))

Answer 2

[Stephen23]

Here are two methods of providing random numbers rounded to two decimal places:

>> randi([0,100],5)/100
ans =
    0.8200    0.0900    0.1500    0.1400    0.6600
    0.9100    0.2800    0.9800    0.4200    0.0300
    0.1200    0.5500    0.9600    0.9200    0.8500
    0.9200    0.9600    0.4900    0.8000    0.9400
    0.6300    0.9700    0.8000    0.9600    0.6800
>> round(100*rand(5))/100
ans =
    0.7600    0.7100    0.8200    0.4400    0.4900
    0.7400    0.0300    0.6900    0.3800    0.4500
    0.3900    0.2800    0.3200    0.7700    0.6500
    0.6600    0.0500    0.9500    0.8000    0.7100
    0.1700    0.1000    0.0300    0.1900    0.7500

Note that you also need to understand the behaviors of floating point numbers:

http://www.mathworks.com/matlabcentral/answers/57444-faq-why-is-0-3-0-2-0-1-not-equal-to-zero

[Walter Roberson]

Note: if memory space is a problem then you can use

randi([0,100], 5, 'uint8')

and adjust the rest of your algorithm to expect these to be 100 times too large. It probably will not help much in practice... it will probably lead to more problems than it is worth.