Stream: Archive Mirror: Isabelle Users Mailing List

Topic: [isabelle] Recdef or Inductive


view this post on Zulip Email Gateway (Aug 17 2022 at 14:49):

From: Peter Lammich <Views@gmx.de>
-------- Original-Nachricht --------
if the type of the argument is given constructor based, i.e. s.th like
datatype C = Name X | Var Y | Pair C C
then you should use a primrec - definition, this will define the measure-function for you :)

primrec Calculate
"Calculate (Name n) = 0"
"Calculate (Var n) = 0"
"Calculate (Pair a b) = Calculate a + Calculate b"

view this post on Zulip Email Gateway (Aug 18 2022 at 09:30):

From: Temesghen Kahsai <lememta@gmail.com>
Hi all

I would like to define function by induction "Calculate" ,that takes
a Term (that could be a Name, Variable or a Pair) and return a Nat.
Ex. Calculate of Name is 0
Calculate of Var is 0
Calculate of Pair(m,n) is Calculate (m) + Calculate (n)

Somthing similar to the Fibonacci function. Should I use Recdef or
Inductive? if I have to use Recdef how I should map the Term (like:
measure function λn. n in Fibonacci)?
Thanks for any advice

-T

view this post on Zulip Email Gateway (Aug 18 2022 at 09:32):

From: Tjark Weber <tjark.weber@gmx.de>
Temesghen,

your example is simple enough to be defined with "primrec". Here's how:

datatype Term = Name
| Var
| Pair Term Term

consts
Calculate :: "Term => nat"

primrec
"Calculate Name = 0"
"Calculate Var = 0"
"Calculate (Pair t1 t2) = Calculate t1 + Calculate t2"

If you want to define a more complicated function using "recdef", you will
need to provide an appropriate measure function of type "Term => nat" (or,
less conveniently, a termination set of type "(Term \<times> Term) set").
Note that the datatype package already creates a default "size" function for
each datatype. Use

thm "Term.size"

to view the size function's definition for your Term datatype. See Sections
2.4.3 and 3.5 of the Isabelle/HOL Tutorial [1] for further details.

Best,
Tjark

[1] http://isabelle.in.tum.de/dist/Isabelle/doc/tutorial.pdf


Last updated: Nov 21 2024 at 12:39 UTC