The exact solution for 38.59 in base 15 is the repeating decimal 28.8cb3b3b3b3b3b3b3..., which means the given solution 28.8cb3b3b3b3b42d4a contains some "junk digits" at the end; it's not really fair to ask for those junk digits as opposed to any others. Also, it looks like trailing zeros are supposed to be removed, but some of the base 2 solutions contain a trailing zero.

Tim,
You are correct. Thanks for the comment. Fixed the errors you detected as well as a few others. Should be good now.

Hi David, nice problem but there are still I believe a few precision issues remaining in the testsuite. For example the number "7285.4" as a float can only be precisely encoded up to ~1e-12 (since eps(7285.4)=9.0949e-13), so its base-15 representation cannot be defined with 16-decimal places (as that would require >1e-19 precision). I believe perhaps you could fix that simply entering the input decimal numbers using chars, which would avoid any limited-precision issues? (or by relaxing the requirements to fewer decimal places)

Changed and fixed the problem. Input is now cell array of character arrays.