None of the file formats supported by imwrite() are able to write floating point numbers for the components. PPM files support raw binary of up to 16 bits per channel, but the smallest floating point format that MATLAB supports is 32 bits per channel. All other file formats for color images scale the RGB values by 2^N (where N is the number of bits per channel) and convert the result to integers. Depending on the file format chosen, 8 or 16 bits per channel are typical. A file format that only supports 8 bits per channel would have a minimum distinguishable RGB value of 1/256 which is about 0.004 . Therefore, unless you use a file format with more than 8 bits per channel, your values of 0.01 will be converted into binary 0.
imshow() is not able to display HSV images directly.
None of the file formats supported by imwrite() is intended to store HSV images; on the other hand, the lossless formats do not care what the components represent... just don't ask to display the resulting file. Beware that if you use a lossy image format such as JPEG that there will be assumptions about the meaning of the components, and so storing HSV values and allowing those to be compressed could potentially result in significant image alteration.
Please see rgb2hsv() and hsv2rgb().
