polished

all material

all material

almost finished

more on paper

more on the paper

more on paper

minor

more on paper

more on paper

more on paper

more on the paper

more

more on paper

more

paper

merged

more on paper

more

more on the paper

more on paper

more on the lmcs paper

updated to Isabelle 28 Aug

Use same constructor names as Lambda, remove copies of FCB, remove [eqvt].

Add lmcs-paper to hgignore

Add 'no-brackets' to avoid '[| |]' in papers.

Comment out examples with 'True' that do not work because function still does not work

Update to new Isabelle

a bit more on the paper

made same changes as in main branch

more on the lmcs paper

a little tuning on the paper

more on the intro and correct style-files

uodated to new Isabelle (15. Aug)

updated for new Isabelle (11. Aug.)

merged

started lmcs paper (isabelle make lmcs)

update to 'termination (eqvt)'.

tuned

completed the eqvt-proofs for functions; they are stored under the name function_name.eqvt and added to the eqvt-list

temporary fix

Add an ".hgignore" file

merged

merged

merged

added termination file

preliminary version of automatically generation the eqvt-lemmas for functions defined with nominal_primrec

generated the partial eqvt-theorem for functions

added a flag (eqvt) to termination proofs arising fron nominal_primrecs

moved eqvt for Option.map

some tuning

direct definition of height using bn

defined a function directly over a nominal datatype with bn

more one the NBE example

some improvements to the NBE example

slight tuning

use eqvt_at_perm

Remove copy of FCB and cleanup

Experiment with permuting eqvt_at

combinators for local theories and lists

merged

some more experiments with let and bns

some code refactoring

merged

code refactoring; introduced a record for raw_dt_info

more on NBE

more on the NBE function

a little further with NBE

Setup eqvt_at for first goal

attempt for NBE

added some relatively simple examples from paper by Norrish

changed bind to binds in specifications; bind will cause trouble with Monad_Syntax

side commment for future use

made the tests go through again

added another example which seems difficult to define

added a tactic "all_trivials" which simplifies all trivial constructor cases and leaves the others untouched.

merged

all FCB lemmas

exported various FCB-lemmas to a separate file

Express trans_db with Option.map and Option.bind. Possibly mbind is a copy of bind?

Define a version of aux only for same binders. Completeness is fine.

Move If / Let with 'True' to the end of Lambda

merged

tuned

added an example that recurses over two arguments; the interesting proof-obligation is not yet done

Let with a different invariant; not true.

Add non-working Lambda_F_T using FCB2

Added non-working CPS3 using FCB2

Change CPS1 to FCB2

Did the proofs of height and subst for Let with list-like binders. Having apply_assns allows proving things by alpha_bn

side-by-side tests of lets with single assignment; deep-binder case works if the recursion is avoided using an auxiliary function

Exhaust Issue

clarified a sentence

more code refactoring

combined distributed data for alpha in alpha_result (partially done)

moved Classical and Let temporarily into a section where "sorry" is allowed; this makes all test go through

added a warning if a theorem is already declared as equivariant

merged

Prove bn injectivity and experiment more with Let

some experiments

trying new fcb in let/subst

Leftover only inj and eqvt

eapply fcb ok

Removed Inl and Inr

relaxed type in fcb

fcb with explicit bn function

added let-rec example

Experiments with res

proved the fcb also for sets (no restriction yet)

copied all work to Lambda.thy; had to derive a special version of fcb1 for concrete atom

streamlined the fcb-proof and made fcb1 a special case of fcb

completed proof in Classical; the fcb lemma works for any list of atoms (despite what was written earlier)

fcb for multible (list) binders; at the moment all of them have to have the same sort (at-class); this should also work for set binders, but not yet for restriction.

renamed ds to dset (disagreement set)

added small lemma about disagreement set

merge

New-style fcb for multiple binders.

equality of lst_binder and a few helper lemmas [l]lst. T = [l]lst. S <-> T = S

only one of the fcb precondistions are needed (probably the same with the perm-conditions)

another change to the fcb2; this is needed in order to get all proofs through in Lambda.thy

did all cases, except the multiple binder case

an alternative FCB for Abs_lst1; seems simpler but not as simple as I thought; not sure whether it generalises to multiple binders.

except for the interated binder case, finished definition in Calssical.thy

Make examples work with non-precompiled image

Remove Lambda_add.thy

The examples in Lambda_add can be defined by nominal_function directly

Theory name changes for JEdit

More usual names for substitution properties

Second Fixed Point Theorem

Speed-up the completeness proof.

the simplifier can simplify "sort (atom a)" if a is a concrete atom type declared with atom_decl

added file

expanded the example

fixed nasty bug with type variables in nominal_datatypes; this included to be careful with the output of the inductive and function package

tuned

deleted some dead code

some rudimentary infrastructure for storing data about nominal datatypes

constants with the same names

Quotients/TODO addtion

Minor

merge

spelling

merge

Abs_set_fcb

function for let-rec

TODO/minor

Move lst_fcb to Nominal2_Abs

More minor TODOs

Update TODO

Let/minor

Update Quotient/TODO and remove some attic code

merge

little on cps2

got rid of the boolean flag in the raw_equivariance function

Fix

merge

CPS3 can be defined with eqvt_rhs.

moved for the moment CPS translations into the example directory

merged

added eqvt_at and invariant for boths sides of the equations

Added the CPS translation experiments. CPS1 comes with all the proofs, CPS2,3 just have the function and need eqvt_rhs to finish the obligations.

added a test that every function must be of pt-sort

all tests work again

added size-lemmas to simplifier; as a result termination can be proved by the standard lexicographic_order method

added an abstract

added a stub for function paper; "isabelle make fnpaper"

merge

one TODO and one Problem?

merge

Some TODOs

merge

TypeSchemes work with 'default'.

tuned some proofs

fixed the problem when giving a complex default-term; the fundef lemmas in Nominal_Base were not general enough

tuned

Move working examples before non-working ones

Optimized proofs and removed some garbage.

merge

Slightly modify fcb for list1 and put in common place.

Experiments with Let

Eval can be defined with additional freshness

Minor simplification

abs_res_fcb will be enough to finish the multiple-recursive proof, if we have a working 'default'.

More experiments with 'default'

Finished the proof with the invariant

Issue with 'default'

merged

merged

using the option "default" the function package allows one to give an explicit default value

Simpler proof of TypeSchemes/substs

Simplify fcb_res

FCB for res binding and simplified proof of subst for type schemes

Simplify ln-trans proof

cbvs can be easily defined without an invariant

defined the "count-bound-variables-occurences" function which has an accumulator like trans

merged

remove garbage (proofs that assumes the invariant outside function)

Proof of trans with invariant

Testing invariant in Lambda_F_T

cleaned ups a bit the examples with the invariant framework; exported nominal_function_config datatype into separate structure and file

fixed problem with earlier commit about nominal_function_common; added facility for specifying an invariant - added a definition of frees_set which need a finiteness invariant

slightly stronger property in fundef_ex_prop

added an option for an invariant (at the moment only a stub)

added a more general lemma fro fundef_ex1

Trying the induction on the graph

Finish and test the locale approach

FiniteSupp precondition in the function is enough to get rid of completeness obligationss

recursion combinator inside a locale

merge

F for lambda used to define translation to locally nameless

typo

removed dead code

finished the missing obligations

merged

a test with a recursion combinator defined on top of nominal_primrec

Use FCB to simplify proof

merge

Remove SMT

hopefully final fix for ho-functions

first test to fix the problem with free variables

proved subst for All constructor in type schemes.

DB translation using index; easier to reason about.

Problem: free variables in the goal

fixed previous commit

equivariance of db_trans

fixed the problem with cps-like functions

DeBruijn translation in a simplifier friendly way

map_term can be defined when equivariance is assumed

map_term is not a function the way it is defined

Defined translation from nominal to de-Bruijn; with a freshness condition for the lambda case.

Simple eqvt proofs with perm_simps for clarity

tuned last commit

functions involving if and case do not throw exceptions anymore; but eqvt_at assumption has now a precondition

updated to new Isabelle

added eq_iff and distinct lemmas of nominal datatypes to the simplifier

more on slides

added slides for copenhagen

added a problem with inductive_cases (reported by Randy)

misc

made the subtyping work again