In principle i agree with Marcus Muller about arbitrary enforced rules, but i can also see there might be system constraints that prevent the use of a period.
More information on how it will be used will help people give useful answers, for example you seem to prefer using a single character to represent it, which may or may not be ideal depending on the usecases.
You also mentioned "The decimal number is often times preceded by text and an underscore.", but I have no idea if that preceeding text may also contain an integer. For example 10packflowers or 10xflowers to repesent a bundle of goods sold as one item as opposed to a singular flower.
I also dont know if there is limit on the amount of characters.
lastly I don't really know what sort of answer youre looking for
as youve already given some possible solutions, with no mention of
why they haven't statisfied your requirements.
To actually answer your question.
Personally I generally lean towards longer representations to lower the chance of it already existing in the text. eg "decPoint" or even a full "decimalPoint", or "_DP" depending on how much i expect the users to rely on intuition.
Alternatively if you want a single letter from your provided solutions i like the use of 'd' for decimal point, or just an underscore. Float i think will be confusing for anyone who isnt already a programmer. I think underscore is the mostly likely to be intuited by a non-programmer, but i still lean towards 'decPoint', or 'decimalPoint'