let R be Relation; :: thesis: ( R is strongly-normalizing & R is locally-confluent implies R is confluent )
assume A1: R is strongly-normalizing ; :: thesis: ( not R is locally-confluent or R is confluent )
defpred S1[ object ] means for b, c being object st R reduces c1,b & R reduces c1,c holds
b,c are_convergent_wrt R;
assume A2: for a, b, c being object st [a,b] in R & [a,c] in R holds
b,c are_convergent_wrt R ; :: according to REWRITE1:def 24 :: thesis: R is confluent
A3: for a being object st ( for b being object st [a,b] in R & a <> b holds
S1[b] ) holds
S1[a]
proof
let a be object ; :: thesis: ( ( for b being object st [a,b] in R & a <> b holds
S1[b] ) implies S1[a] )

assume A4: for b being object st [a,b] in R & a <> b holds
S1[b] ; :: thesis: S1[a]
let b, c be object ; :: thesis: ( R reduces a,b & R reduces a,c implies b,c are_convergent_wrt R )
assume that
A5: R reduces a,b and
A6: R reduces a,c ; :: thesis: b,c are_convergent_wrt R
consider p being RedSequence of R such that
A7: a = p . 1 and
A8: b = p . (len p) by A5;
A9: len p >= 0 + 1 by NAT_1:13;
consider q being RedSequence of R such that
A10: a = q . 1 and
A11: c = q . (len q) by A6;
A12: len q >= 0 + 1 by NAT_1:13;
per cases ( len p = 1 or len q = 1 or ( len p <> 1 & len q <> 1 ) ) ;
end;
end;
A29: R is co-well_founded by A1;
A30: for a being object st a in field R holds
S1[a] from REWRITE1:sch 2(A29, A3);
given a0, b0 being object such that A31: a0,b0 are_divergent_wrt R and
A32: not a0,b0 are_convergent_wrt R ; :: according to REWRITE1:def 22 :: thesis: contradiction
consider c0 being object such that
A33: ( R reduces c0,a0 & R reduces c0,b0 ) by A31;
( a0 <> c0 or b0 <> c0 ) by A32, Th38;
then c0 in field R by A33, Th18;
hence contradiction by A32, A33, A30; :: thesis: verum