FURY
FURY copied to clipboard
szaghi
Fortran Units (environment) for Reliable phYsical math
Implement a methid to simplify uom references, e.g. ``` fortran type(uom) :: u1 type(uom) :: u2 u1 = uom('m.s-1.kg.m-2') u2 = u1%simplify() print*, u2%stringify() ! Output ! m-1.s.1.kg ```
Once qreal/qinter classes are ready, some bultin procedures are worth to be overloaded: - [ ] sum - [ ] max/min - [ ] max_val/min_val - [ ] max_loc/min_loc
If a physical quantity has an associated estimation of error, a more reliable computation should automatically propagate the errors. Possible pseudo code ``` fortran type(qreal) :: q1 type(qreal) :: q2...
In order to promote Van Snyder's proposal, it could be of some help to prepare a journal publication to submit to a Software Eng. related journal. To this aim, many...
Like Pint module teaches, as well as the comments of experts like Glen, _context_ concept is very useful. Remember to think to it.
Add examples like "FURY within FFT" computations... is it viable to "wrap" procedures like Pint does?
Compuations with dimensional quantities have an overhead, please quantify it!
Presently only real quantity is provided, support also integer one.
Not sure if it is possible to use array syntax with `type(qreal)`... tests are need.
UOM definitions is based on string parsing, thus it is very trivial to load systems definitions from file... Probably, it is better to integrate [json-fortran](https://github.com/jacobwilliams/json-fortran) to read JSON input files.