:: Mazur-Ulam Theorem
:: by Artur Korni{\l}owicz
::
:: Received December 21, 2010
:: Copyright (c) 2010-2016 Association of Mizar Users
:: (Stowarzyszenie Uzytkownikow Mizara, Bialystok, Poland).
:: This code can be distributed under the GNU General Public Licence
:: version 3.0 or later, or the Creative Commons Attribution-ShareAlike
:: License version 3.0 or later, subject to the binding interpretation
:: detailed in file COPYING.interpretation.
:: See COPYING.GPL and COPYING.CC-BY-SA for the full text of these
:: licenses, or see http://www.gnu.org/licenses/gpl.html and
:: http://creativecommons.org/licenses/by-sa/3.0/.
environ
vocabularies FUNCT_1, NORMSP_1, VECTMETR, PRE_TOPC, ARYTM_1, RUSUB_4, XREAL_0,
CARD_1, XXREAL_0, RELAT_1, ARYTM_3, FUNCT_2, SEQ_2, MEMBERED, ORDINAL2,
MAZURULM, BINOP_1, VALUED_1, SUPINF_2, COMPLEX1, SUBSET_1, XXREAL_2,
XBOOLE_0, NUMBERS, MEMBER_1, SUPINF_1, TARSKI, RLVECT_1, CFCONT_1, NAT_1,
PARTFUN1, REAL_1, NEWTON, TOPMETR, BORSUK_1, TOPS_1, XXREAL_1, URYSOHN1,
METRIC_1, STRUCT_0, SEQ_1, FUNCOP_1, RCOMP_1, XCMPLX_0;
notations TARSKI, XBOOLE_0, SUBSET_1, ORDINAL1, FUNCT_1, RELSET_1, PARTFUN1,
FUNCT_2, FUNCOP_1, NUMBERS, XCMPLX_0, XXREAL_0, XXREAL_1, XXREAL_3,
XREAL_0, REAL_1, XXREAL_2, MEMBERED, VALUED_1, SEQ_1, SEQ_2, COMPLEX1,
MEMBER_1, SUPINF_1, NEWTON, MEASURE6, STRUCT_0, PRE_TOPC, TOPS_1, TOPS_2,
BORSUK_1, METRIC_1, TOPMETR, ALGSTR_0, RLVECT_1, NORMSP_0, NORMSP_1,
NDIFF_1, NFCONT_1, CONNSP_3, URYSOHN1;
constructors TOPS_2, REAL_1, SUPINF_2, INTEGRA2, EXTREAL1, BINOP_2, NFCONT_1,
NEWTON, CONNSP_3, URYSOHN1, TOPS_1, TOPMETR, PCOMPS_1, MEASURE6, NDIFF_1,
COMSEQ_2;
registrations RELSET_1, XREAL_0, ORDINAL1, STRUCT_0, MEMBERED, FUNCT_1,
FUNCT_2, XBOOLE_0, NORMSP_1, FUNCTOR1, NUMBERS, SEQ_4, XXREAL_2,
MEMBER_1, XXREAL_0, RLVECT_1, TOPS_1, TOPMETR, METRIC_1, XXREAL_1,
VALUED_1, VALUED_0, FUNCOP_1, BORSUK_1, INT_1, NEWTON, XXREAL_3, NAT_1,
SEQ_1, SEQ_2;
requirements REAL, NUMERALS, SUBSET, BOOLE, ARITHM;
definitions TARSKI, XBOOLE_0, FUNCT_2, XXREAL_2, NORMSP_1, TOPS_1;
equalities XCMPLX_0, STRUCT_0, RLVECT_1, METRIC_1, CONNSP_3;
expansions TARSKI, NORMSP_1, TOPS_1;
theorems FUNCT_2, RLVECT_1, FUNCT_1, NORMSP_1, COMPLEX1, TOPS_2, RELAT_1,
XXREAL_2, XREAL_1, XXREAL_0, MEMBERED, URYSOHN2, MEMBER_1, GROUP_6,
PENCIL_2, XREAL_0, INTEGRA2, VECTSP_1, NDIFF_1, NFCONT_1, XCMPLX_1,
NEWTON, BORSUK_1, XBOOLE_0, CONNSP_3, TOPMETR, GOBOARD6, FRECHET,
XXREAL_1, MEASURE6, SEQ_2, FUNCOP_1, VALUED_1, URYSOHN1, SEQ_4, TREAL_1,
TOPS_3, NAT_1, INT_1, JORDAN5A, ORDINAL1;
schemes FUNCT_2, NAT_1;
begin
:: Mazur-Ulam Theorem
:: cluster bijective isometric -> Affine Function of E,F;
:: is a consequence of two registrations:
:: cluster bijective isometric -> midpoints-preserving Function of E,F;
:: cluster isometric midpoints-preserving -> Affine Function of E,F;
reserve E,F,G for RealNormSpace;
reserve f for Function of E,F;
reserve g for Function of F,G;
reserve a,b,c for Point of E;
reserve t for Real;
registration
cluster I[01] -> closed for SubSpace of R^1;
coherence by TOPMETR:20,TREAL_1:2;
end;
reconsider D = DYADIC as Subset of I[01] by BORSUK_1:40,URYSOHN2:28;
Lm1: Cl D = [#]I[01]
proof
reconsider P = [. 0,1 .] as Subset of R^1 by MEMBERED:3,TOPMETR:17;
A1: I[01] = R^1 | P by TOPMETR:19,20;
reconsider D1 = D as Subset of R^1 | P by TOPMETR:19,20;
A2: Cl D = (Cl Up(D1)) /\ P by A1,CONNSP_3:30;
thus Cl D c= [#]I[01];
let x be object;
assume
A3: x in [#]I[01];
reconsider p = x as Point of RealSpace by A3,BORSUK_1:40;
now
let r be Real such that
A4: r > 0;
A5: p-r < p-0 by A4,XREAL_1:10;
A6: p <= 1 by A3,BORSUK_1:43;
A7: Ball(p,r) = ].p-r, p+r.[ by FRECHET:7;
per cases;
suppose r <= p;
then r-r <= p-r by XREAL_1:9;
then consider c being Real such that
A8: c in D and
A9: p-r < c and
A10: c < p by A5,A6,URYSOHN2:24;
p+0 < p+r by A4,XREAL_1:8;
then c < p+r by A10,XXREAL_0:2;
then c in Ball(p,r) by A9,A7,XXREAL_1:4;
hence Ball(p,r) meets D by A8,XBOOLE_0:3;
end;
suppose r > p;
then
A11: p-r < p-p by XREAL_1:10;
0 <= p by A3,BORSUK_1:43;
then 0 in Ball(p,r) by A4,A11,A7,XXREAL_1:4;
hence Ball(p,r) meets D by URYSOHN2:27,XBOOLE_0:3;
end;
end;
then x in Cl Up(D1) by GOBOARD6:92,TOPMETR:def 6;
hence x in Cl D by A3,A2,BORSUK_1:40,XBOOLE_0:def 4;
end;
theorem
DYADIC is dense Subset of I[01]
proof
D is dense
by Lm1;
hence thesis;
end;
theorem Th2:
Cl DYADIC = [. 0,1 .]
proof
reconsider W = D as Subset of R^1 by TOPMETR:17;
thus Cl DYADIC = Cl W by JORDAN5A:24
.= [. 0,1 .] by Lm1,BORSUK_1:40,TOPS_3:55;
end;
theorem Th3:
a + a = 2*a
proof
1*a = a by RLVECT_1:def 8;
hence a + a = (1+1)*a by RLVECT_1:def 6
.= 2*a;
end;
theorem Th4:
a + b - b = a
proof
thus a + b - b = a + (b - b) by RLVECT_1:28
.= a + 0.E by RLVECT_1:15
.= a;
end;
registration
let A be bounded_above real-membered set;
let r be non negative Real;
cluster r**A -> bounded_above;
coherence
proof
per cases;
suppose A1: A <> {};
A c= REAL by MEMBERED:3;
hence thesis by A1,INTEGRA2:9;
end;
suppose A = {};
hence thesis;
end;
end;
end;
registration
let A be bounded_above real-membered set;
let r be non positive Real;
cluster r**A -> bounded_below;
coherence
proof
per cases;
suppose A1: A <> {};
A c= REAL by MEMBERED:3;
hence thesis by A1,INTEGRA2:10;
end;
suppose A = {};
hence thesis;
end;
end;
end;
registration
let A be bounded_below real-membered set;
let r be non negative Real;
cluster r**A -> bounded_below;
coherence
proof
per cases;
suppose A1: A <> {};
A c= REAL by MEMBERED:3;
hence thesis by A1,INTEGRA2:11;
end;
suppose A = {};
hence thesis;
end;
end;
end;
registration
let A be bounded_below non empty real-membered set;
let r be non positive Real;
cluster r**A -> bounded_above;
coherence
proof
A c= REAL by MEMBERED:3;
hence thesis by INTEGRA2:12;
end;
end;
theorem Th5:
for f being Real_Sequence holds f + (NAT --> t) = t + f
proof
let f be Real_Sequence;
let n be Element of NAT;
A1: (NAT --> t).n = t by FUNCOP_1:7;
thus (f + (NAT --> t)).n = f.n + (NAT --> t).n by VALUED_1:1
.= (f + t).n by A1,VALUED_1:2;
end;
theorem Th6:
for r being Element of REAL holds lim (NAT --> r) = r
proof
let r be Element of REAL;
thus lim (NAT --> r) = (NAT --> r).0 by SEQ_4:26
.= r by FUNCOP_1:7;
end;
theorem Th7:
for f being convergent Real_Sequence holds
lim (t + f) = t + lim f
proof
let f be convergent Real_Sequence;
reconsider r = t as Element of REAL by XREAL_0:def 1;
f + (NAT --> t) = t + f by Th5;
hence lim (t + f) = lim (NAT --> r) + lim f by SEQ_2:6
.= t + lim f by Th6;
end;
registration
let f be convergent Real_Sequence;
let t;
cluster t + f -> convergent for Real_Sequence;
coherence
proof
reconsider r = t as Element of REAL by XREAL_0:def 1;
f + (NAT --> r) = t + f by Th5;
hence thesis;
end;
end;
theorem Th8:
for f be Real_Sequence holds f (#) (NAT --> a) = f * a
proof
let f be Real_Sequence;
let n be Element of NAT;
A1: (NAT --> a).n = a by FUNCOP_1:7;
thus (f (#) (NAT --> a)).n = f.n * (NAT --> a).n by NDIFF_1:def 2
.= (f * a).n by A1,NDIFF_1:def 3;
end;
theorem Th9:
lim (NAT --> a) = a
proof
thus lim (NAT --> a) = (NAT --> a).0 by NDIFF_1:18
.= a by FUNCOP_1:7;
end;
theorem Th10:
for f being convergent Real_Sequence holds
lim (f * a) = (lim f) * a
proof
let f be convergent Real_Sequence;
f (#) (NAT --> a) = f * a by Th8;
hence lim (f * a) = lim f * lim (NAT --> a) by NDIFF_1:14,18
.= (lim f) * a by Th9;
end;
registration
let f be convergent Real_Sequence;
let E,a;
cluster f * a -> convergent;
coherence
proof
f (#) (NAT --> a) = f * a by Th8;
hence thesis by NDIFF_1:13,18;
end;
end;
definition
let E,F be non empty NORMSTR, f be Function of E,F;
attr f is isometric means :Def1:
for a,b being Point of E holds ||. f.a - f.b .|| = ||. a - b .||;
end;
definition
let E,F be non empty RLSStruct, f be Function of E,F;
attr f is Affine means
for a,b being Point of E, t being Real st 0 <= t & t <= 1 holds
f.((1-t)*a+t*b) = (1-t)*(f.a) + t*(f.b);
attr f is midpoints-preserving means
for a,b being Point of E holds f.(1/2*(a+b)) = 1/2*(f.a+f.b);
end;
registration
let E be non empty NORMSTR;
cluster id E -> isometric;
coherence;
end;
registration
let E be non empty RLSStruct;
cluster id E -> midpoints-preserving Affine;
coherence;
end;
registration
let E be non empty NORMSTR;
cluster bijective isometric midpoints-preserving Affine for UnOp of E;
existence
proof
take id E;
thus thesis;
end;
end;
theorem Th11:
f is isometric & g is isometric implies g*f is isometric
proof
assume that
A1: f is isometric and
A2: g is isometric;
set h = g*f;
let a,b;
h.a = g.(f.a) & h.b = g.(f.b) by FUNCT_2:15;
hence ||. h.a - h.b .|| = ||. f.a - f.b .|| by A2
.= ||. a - b .|| by A1;
end;
registration
let E; let f,g be isometric UnOp of E;
cluster g*f -> isometric for UnOp of E;
coherence by Th11;
end;
Lm2:
now
let E,F,f;
assume
A1: f is bijective;
set g = f/";
let a be Point of F;
rng f = [#]F by A1,FUNCT_2:def 3;
then
A2: g/" = f by A1,TOPS_2:51;
A3: g = f qua Function" by A1,TOPS_2:def 4;
A4: g is bijective by A1,PENCIL_2:12;
f = g qua Function" by A2,A4,TOPS_2:def 4;
hence f.(g.a) = a by A1,A3,FUNCT_2:26;
end;
theorem Th12:
f is bijective isometric implies f/" is isometric
proof
assume that
A1: f is bijective and
A2: f is isometric;
set g = f/";
let a,b be Point of F;
f.(g.a) = a & f.(g.b) = b by A1,Lm2;
hence thesis by A2;
end;
registration
let E; let f be bijective isometric UnOp of E;
cluster f/" -> isometric;
coherence by Th12;
end;
theorem Th13:
f is midpoints-preserving & g is midpoints-preserving implies
g*f is midpoints-preserving
proof
assume that
A1: f is midpoints-preserving and
A2: g is midpoints-preserving;
set h = g*f;
let a,b;
A3: h.a = g.(f.a) & h.b = g.(f.b) by FUNCT_2:15;
thus h.(1/2*(a+b)) = g.(f.(1/2*(a+b))) by FUNCT_2:15
.= g.(1/2*((f.a)+(f.b))) by A1
.= 1/2*(h.a+h.b) by A3,A2;
end;
registration
let E; let f,g be midpoints-preserving UnOp of E;
cluster g*f -> midpoints-preserving for UnOp of E;
coherence by Th13;
end;
Lm3:
now
let E,F,f;
assume
A1: f is bijective;
then
A2: rng f = [#]F by FUNCT_2:def 3;
dom f = [#]E by FUNCT_2:def 1;
hence (f/")*f = id E by A1,A2,TOPS_2:52;
end;
theorem Th14:
f is bijective midpoints-preserving implies f/" is midpoints-preserving
proof
assume that
A1: f is bijective and
A2: f is midpoints-preserving;
set g = f/";
let a,b be Point of F;
A3: g*f = id E by A1,Lm3;
f.(g.a) = a & f.(g.b) = b by A1,Lm2;
hence g.(1/2*(a+b)) = g.(f.(1/2*((f/".a)+(f/".b)))) by A2
.= (g*f).(1/2*((f/".a)+(f/".b))) by FUNCT_2:15
.= 1/2*(g.a+g.b) by A3;
end;
registration
let E; let f be bijective midpoints-preserving UnOp of E;
cluster f/" -> midpoints-preserving;
coherence by Th14;
end;
theorem Th15:
f is Affine & g is Affine implies g*f is Affine
proof
assume that
A1: f is Affine and
A2: g is Affine;
set h = g*f;
let a,b,t such that
A3: 0 <= t & t <= 1;
A4: h.a = g.(f.a) & h.b = g.(f.b) by FUNCT_2:15;
thus h.((1-t)*a+t*b) = g.(f.((1-t)*a+t*b)) by FUNCT_2:15
.= g.((1-t)*f.a+t*f.b) by A1,A3
.= (1-t)*(h.a) + t*(h.b) by A2,A3,A4;
end;
registration
let E; let f,g be Affine UnOp of E;
cluster g*f -> Affine for UnOp of E;
coherence by Th15;
end;
theorem Th16:
f is bijective Affine implies f/" is Affine
proof
assume that
A1: f is bijective and
A2: f is Affine;
set g = f/";
let a,b be Point of F;
let t such that
A3: 0 <= t & t <= 1;
A4: g*f = id E by A1,Lm3;
f.(g.a) = a & f.(g.b) = b by A1,Lm2;
hence g.((1-t)*a+t*b) = g.(f.((1-t)*(g.a)+t*(g.b))) by A3,A2
.= (g*f).((1-t)*(g.a)+t*(g.b)) by FUNCT_2:15
.= (1-t)*g.a+t*g.b by A4;
end;
registration
let E; let f be bijective Affine UnOp of E;
cluster f/" -> Affine;
coherence by Th16;
end;
definition
let E be non empty RLSStruct, a be Point of E;
func a-reflection -> UnOp of E means :Def4:
for b being Point of E holds it.b = 2*a - b;
existence
proof
deffunc F(Point of E) = 2*a - $1;
ex f being UnOp of E st for x being Point of E holds f.x = F(x)
from FUNCT_2:sch 4;
hence thesis;
end;
uniqueness
proof
let f1,f2 be UnOp of E such that
A1: for b being Point of E holds f1.b = 2*a - b and
A2: for b being Point of E holds f2.b = 2*a - b;
let b be Point of E;
thus f1.b = 2*a - b by A1
.= f2.b by A2;
end;
end;
theorem Th17:
(a-reflection) * (a-reflection) = id E
proof
set R = a-reflection;
let b;
thus (R*R).b = R.(R.b) by FUNCT_2:15
.= 2*a - R.b by Def4
.= 2*a - (2*a - b) by Def4
.= 2*a - 2*a + b by RLVECT_1:29
.= (0.E) + b by RLVECT_1:5
.= b
.= (id E).b;
end;
registration
let E,a;
cluster a-reflection -> bijective;
coherence
proof
set R = a-reflection;
R*R = id E by Th17;
hence R is one-to-one onto by FUNCT_2:23;
end;
end;
theorem Th18: :: a is the only fixed point of R
(a-reflection).a = a & for b st (a-reflection).b = b holds a = b
proof
set R = a-reflection;
thus R.a = 2*a - a by Def4
.= a + a - a by Th3
.= a + (a - a) by RLVECT_1:28
.= a + 0.E by RLVECT_1:15
.= a;
let b;
assume R.b = b;
then
A1: R.b + b = 2*b by Th3;
R.b = 2*a - b by Def4;
then R.b + b = 2*a - (b - b) by RLVECT_1:29
.= 2*a - 0.E by RLVECT_1:15
.= 2*a;
hence thesis by A1,RLVECT_1:36;
end;
theorem Th19:
(a-reflection).b - a = a - b
proof
set R = a-reflection;
A1: 1*a = a by RLVECT_1:def 8;
R.b = 2*a - b by Def4;
hence R.b - a = 2*a - a - b by VECTSP_1:34
.= (2-1)*a - b by A1,RLVECT_1:35
.= a - b by RLVECT_1:def 8;
end;
theorem Th20:
||. (a-reflection).b - a .|| = ||. b - a .||
proof
(a-reflection).b - a = a - b by Th19;
hence thesis by NORMSP_1:7;
end;
theorem Th21:
(a-reflection).b - b = 2 * (a - b)
proof
2*a - b - b = 2*a - (b + b) by RLVECT_1:27
.= 2*a - 2*b by Th3
.= 2*(a-b) by RLVECT_1:34;
hence thesis by Def4;
end;
theorem Th22:
||. (a-reflection).b - b .|| = 2 * ||. b - a .||
proof
A1: (a-reflection).b - b = 2 * (a - b) by Th21;
A2: |.2.| = 2 by COMPLEX1:43;
||. a - b .|| = ||. b - a .|| by NORMSP_1:7;
hence thesis by A1,A2,NORMSP_1:def 1;
end;
theorem Th23:
(a-reflection)/" = a-reflection
proof
set R = a-reflection;
A1: rng R = [#]E by FUNCT_2:def 3;
A2: R/" = (R qua Function)" by TOPS_2:def 4;
R*R = id E by Th17;
hence thesis by A1,A2,FUNCT_2:30;
end;
registration
let E,a;
cluster a-reflection -> isometric;
coherence
proof
set R = a-reflection;
let b,c;
R.b = 2*a - b & R.c = 2*a - c by Def4;
then R.b - R.c = 2*a - b - 2*a + c by RLVECT_1:29
.= 2*a - 2*a - b + c by VECTSP_1:34
.= 0.E - b + c by RLVECT_1:15
.= c - b;
hence ||. R.b - R.c .|| = ||. b - c .|| by NORMSP_1:7;
end;
end;
deffunc W(RealNormSpace,Point of $1,Point of $1) =
{g where g is UnOp of $1: g is bijective isometric &
g.$2 = $2 & g.$3 = $3};
deffunc l(RealNormSpace,Point of $1,Point of $1) =
{||. g.(1/2*($2+$3)) - (1/2*($2+$3)) .||
where g is UnOp of $1: g in W($1,$2,$3)};
Lm4:
now
let E,a,b;
let l be real-membered set such that
A1: l = l(E,a,b);
set z = 1/2*(a+b);
thus 2 * ||. a - z .|| is UpperBound of l
proof
let x be ExtReal;
assume x in l;
then consider g1 being UnOp of E such that
A2: x = ||. g1.z - z .|| and
A3: g1 in W(E,a,b) by A1;
consider g being UnOp of E such that
A4: g1 = g and
g is bijective and
A5: g is isometric and
A6: g.a = a and g.b = b by A3;
A7: ||. g.z - a .|| = ||. z - a .|| by A5,A6
.= ||. a - z .|| by NORMSP_1:7;
||. g.z - z .|| <= ||. g.z - a .|| + ||. a - z .|| by NORMSP_1:10;
hence thesis by A2,A4,A7;
end;
end;
Lm5:
now
let E,a,b;
let h be UnOp of E such that
A1: h in W(E,a,b);
set z = 1/2*(a+b);
set R = z-reflection;
set gs = R*(h/")*R*h;
consider g being UnOp of E such that
A2: g = h and
A3: g is bijective and
A4: g is isometric and
A5: g.a = a and
A6: g.b = b by A1;
A7: g/" = (g qua Function)" by A3,TOPS_2:def 4;
then
A8: g/" is bijective by A3,GROUP_6:63;
A9: 2*z = (2*(1/2))*(a+b) by RLVECT_1:def 7
.= a + b by RLVECT_1:def 8;
then
A10: 2*z - a = b by Th4;
A11: 2*z - b = a by A9,Th4;
A12: dom g = [#]E by FUNCT_2:def 1;
A13: (g/").b = b by A7,A6,A3,A12,FUNCT_1:32;
A14: (g/").a = a by A7,A5,A3,A12,FUNCT_1:32;
A15: gs.a = (R*(g/")*R).a by A5,A2,FUNCT_2:15
.= (R*(g/")).(R.a) by FUNCT_2:15
.= (R*(g/")).b by A10,Def4
.= R.((g/").b) by FUNCT_2:15
.= 2*z - b by A13,Def4
.= a by A9,Th4;
gs.b = (R*(g/")*R).b by A6,A2,FUNCT_2:15
.= (R*(g/")).(R.b) by FUNCT_2:15
.= (R*(g/")).a by A11,Def4
.= R.((g/").a) by FUNCT_2:15
.= 2*z - a by A14,Def4
.= b by A9,Th4;
hence gs in W(E,a,b) by A2,A3,A4,A8,A15;
end;
Lm6:
now
let E,a,b;
let l be non empty bounded_above Subset of REAL such that
A1: l = l(E,a,b);
thus sup l is UpperBound of 2**l
proof
set z = 1/2*(a+b);
set R = z-reflection;
let x be ExtReal;
assume x in 2**l;
then consider w being Element of ExtREAL such that
A2: x = 2*w and
A3: w in l by MEMBER_1:188;
consider h being UnOp of E such that
A4: w = ||. h.z - z .|| and
A5: h in W(E,a,b) by A3,A1;
consider g being UnOp of E such that
A6: g = h and
A7: g is bijective and
A8: g is isometric and g.a = a & g.b = b by A5;
A9: R*(g/"*(R*g)) = R*(g/")*(R*g) by RELAT_1:36
.= R*(g/")*R*g by RELAT_1:36;
A10: R.(g/".(R.(g.z))) = R.(g/".((R*g).z)) by FUNCT_2:15
.= R.((g/"*(R*g)).z) by FUNCT_2:15
.= (R*(g/"*(R*g))).z by FUNCT_2:15;
A11: g/" = (g qua Function)" by A7,TOPS_2:def 4;
R*(g/")*R*g in W(E,a,b) by A5,A6,Lm5;
then
A12: ||. (R*(g/")*R*g).z - z .|| in l by A1;
reconsider d = 2 as R_eal by XXREAL_0:def 1;
A13: g/".(g.z) = z by A7,A11,FUNCT_2:26;
2 * ||. g.z - z .|| = ||. R.(g.z) - g.z .|| by Th22
.= ||. g/".(R.(g.z)) - g/".(g.z) .|| by A7,A8,Def1
.= ||. R.(g/".(R.(g.z))) - z .|| by A13,Th20;
hence x <= sup l by A2,A4,A9,A10,A12,A6,XXREAL_2:4;
end;
end;
Lm7:
now
let E,a,b;
let l be real-membered set such that
A1: l = l(E,a,b);
let g be UnOp of E such that
A2: g in W(E,a,b);
set z = 1/2*(a+b);
set R = z-reflection;
A3: l c= REAL
by XREAL_0:def 1;
(id E).a = a & (id E).b = b;
then id E in W(E,a,b);
then
A4: ||. (id E).z - z .|| in l by A1;
2 * ||. a - z .|| is UpperBound of l by A1,Lm4;
then reconsider A = l as non empty bounded_above Subset of REAL
by A3,A4,XXREAL_2:def 10;
set lambda = sup A;
reconsider d = 2 as R_eal by XXREAL_0:def 1;
A5: d*sup A = sup(2**A) by URYSOHN2:18;
A6: ||. g.z - z .|| in A by A1,A2;
lambda is UpperBound of 2**A by A1,Lm6;
then sup(2**A) <= lambda by XXREAL_2:def 3;
then 2*lambda - lambda <= lambda - lambda by A5,XREAL_1:9;
then ||. g.z - z .|| = 0 by A6,XXREAL_2:4;
hence g.z = z by NORMSP_1:6;
end;
theorem Th24:
f is isometric implies f is_continuous_on dom f
proof
assume
A1: f is isometric;
set X = dom f;
for s1 being sequence of E st rng s1 c= X & s1 is convergent &
lim s1 in X holds f/*s1 is convergent & f/.(lim s1) = lim (f/*s1)
proof
let s1 be sequence of E;
assume that
A2: rng s1 c= X and
A3: s1 is convergent and lim s1 in X;
consider a such that
A4: for r being Real st 0 < r
ex m being Nat st for n being Nat st m <= n
holds ||. s1.n - a .|| < r by A3;
A5: a = lim s1 by A3,A4,NORMSP_1:def 7;
A6: for r being Real st 0 < r ex m being Nat st
for n being Nat st m <= n holds ||. (f/*s1).n - f.a .|| < r
proof
let r be Real;
assume 0 < r;
then consider m being Nat such that
A7: for n being Nat st m <= n holds ||. s1.n - a .|| < r by A4;
take m;
let n be Nat;
A8: n in NAT by ORDINAL1:def 12;
assume m <= n;
then
A9: ||. s1.n - a .|| < r by A7;
A10: ||. f.(s1.n) - f.a .|| = ||. s1.n - a .|| by A1;
f/*s1 = f*s1 by A2,FUNCT_2:def 11;
hence ||. (f/*s1).n - f.a .|| < r by A9,A10,FUNCT_2:15,A8;
end;
thus f/*s1 is convergent
by A6;
hence f/.(lim s1) = lim (f/*s1) by A5,A6,NORMSP_1:def 7;
end;
hence thesis by NFCONT_1:18;
end;
Lm8:
(1-t)*a + t*b = a + t*(b-a)
proof
thus (1-t)*a + t*b = 1*a - t*a + t*b by RLVECT_1:35
.= a - t*a + t*b by RLVECT_1:def 8
.= a + t*b - t*a by RLVECT_1:def 3
.= a + (t*b - t*a) by RLVECT_1:28
.= a + t*(b-a) by RLVECT_1:34;
end;
Lm9:
now
let E,F,f,a,b;
let n,j be Nat;
set m = 2|^(n+(1 qua Complex) qua Nat);
set k = 2|^n;
set x = k-j;
assume
A1: f is midpoints-preserving;
assume j <= k;
then x in NAT by INT_1:5;
then reconsider x as Nat;
take x;
thus x = k-j;
set z = 1-j/m - 1/2;
A2: k <> 0 by NEWTON:83;
A3: 2*k = m by NEWTON:6;
A4: (1/2)*(1/k) = 1/(2*k) by XCMPLX_1:102;
x/m = (1*k)/m - j/m
.= 1/2 - j/m by A2,A3,XCMPLX_1:91
.= z;
then
A5: z*a + j/m*b = 1/m*(x*a) + 1/m*j*b by RLVECT_1:def 7
.= 1/m*(x*a) + 1/m*(j*b) by RLVECT_1:def 7
.= 1/m*(x*a+j*b) by RLVECT_1:def 5
.= 1/2*(1/k*(x*a+j*b)) by A4,A3,RLVECT_1:def 7;
a+j/m*(b-a) = a+(j/m*b-j/m*a) by RLVECT_1:34
.= a+j/m*b-j/m*a by RLVECT_1:def 3
.= a-j/m*a+j/m*b by RLVECT_1:def 3
.= 1*a-j/m*a+j/m*b by RLVECT_1:def 8
.= (1-j/m)*a+j/m*b by RLVECT_1:35
.= (1-j/m)*a + z*a - z*a + j/m*b by Th4
.= (1-j/m)*a - z*a + z*a + j/m*b by RLVECT_1:def 3
.= (1-j/m-z)*a + z*a + j/m*b by RLVECT_1:35
.= 1/2*a + (z*a + j/m*b) by RLVECT_1:def 3
.= 1/2*(a+(1/k*(x*a+j*b))) by A5,RLVECT_1:def 5;
hence f.(a+j/m*(b-a)) = 1/2*(f.a+f.(1/k*(x*a+j*b))) by A1;
end;
Lm10:
now
let E,F,f,a,b;
let n,j be Nat;
set m = 2|^(n+(1 qua Complex)qua Nat);
set k = 2|^n;
set x = k+j-m;
assume
A1: f is midpoints-preserving;
A2: 2*k = m by NEWTON:6;
assume j >= k;
then k+k <= k+j by XREAL_1:6;
then x in NAT by A2,INT_1:5;
then reconsider x as Nat;
take x;
thus x = k+j-m;
set z = j/m - (1/2 qua Complex);
A3: k <> 0 by NEWTON:83;
A4: m <> 0 by NEWTON:83;
A5: m/m = 1 by A4,XCMPLX_1:60;
then
A6: 1-j/m = m/m - j/m
.= 1/m*(m-j);
A7: k/m = (1*k)/(2*k) by NEWTON:6
.= 1/2 by A3,XCMPLX_1:91;
A8: (1/2)*(1/k) = 1/(2*k) by XCMPLX_1:102;
x/m = (k+j)/m - m/m
.= k/m+j/m - m/m
.= z by A5,A7;
then
A9: z*b + (1-j/m)*a = 1/m*((m-j)*a) + 1/m*x*b by A6,RLVECT_1:def 7
.= 1/m*((m-j)*a) + 1/m*(x*b) by RLVECT_1:def 7
.= 1/m*((m-j)*a+x*b) by RLVECT_1:def 5
.= 1/2*(1/k*((m-j)*a+x*b)) by A8,A2,RLVECT_1:def 7;
a+j/m*(b-a) = a+(j/m*b-j/m*a) by RLVECT_1:34
.= a+j/m*b-j/m*a by RLVECT_1:def 3
.= a-j/m*a+j/m*b by RLVECT_1:def 3
.= 1*a-j/m*a+j/m*b by RLVECT_1:def 8
.= (1-j/m)*a+j/m*b by RLVECT_1:35
.= j/m*b + z*b - z*b + (1-j/m)*a by Th4
.= j/m*b - z*b + z*b + (1-j/m)*a by RLVECT_1:def 3
.= (j/m-z)*b + z*b + (1-j/m)*a by RLVECT_1:35
.= 1/2*b + (z*b + (1-j/m)*a) by RLVECT_1:def 3
.= 1/2*(b+(1/k*((m-j)*a+x*b))) by A9,RLVECT_1:def 5;
hence f.(a+j/m*(b-a)) = 1/2*(f.b+f.(1/k*((m-j)*a+x*b))) by A1;
end;
Lm11:
now
let E,F,f,a,b;
let t be Nat;
assume
A1: f is midpoints-preserving;
thus for n being Nat st t <= 2|^n holds
f.((1-t/(2|^n))*a + t/(2|^n)*b) = (1-t/(2|^n))*f.a + t/(2|^n)*f.b
proof
defpred P[Nat] means
for w being Nat st w <= 2|^$1 holds
f.((1-w/(2|^$1))*a + w/(2|^$1)*b) =
(1-w/(2|^$1))*f.a + w/(2|^$1)*f.b;
A2: P[0]
proof
let t be Nat such that
A3: t <= 2|^0;
A4: 2|^0 = 1 by NEWTON:4;
per cases by A3,A4,NAT_1:25;
suppose
A5: t = 1;
then f.((1-t)*a + t*b) = f.(0.E + t*b) by RLVECT_1:10
.= f.(t*b)
.= f.b by A5,RLVECT_1:def 8
.= t*(f.b) by A5,RLVECT_1:def 8
.= 0.F + t*(f.b)
.= (1-t)*(f.a) + t*(f.b) by A5,RLVECT_1:10;
hence thesis by A4;
end;
suppose
A6: t = 0;
then f.((1-t)*a + t*b) = f.(1*a + 0.E) by RLVECT_1:10
.= f.(1*a)
.= f.a by RLVECT_1:def 8
.= (1-t)*(f.a) by A6,RLVECT_1:def 8
.= (1-t)*(f.a) + 0.F
.= (1-t)*(f.a) + t*(f.b) by A6,RLVECT_1:10;
hence thesis by A4;
end;
end;
A7: for n being Nat st P[n] holds P[n+(1 qua Complex)]
proof
let n be Nat;
set m = 2|^n, k = 2|^(n+(1 qua Complex));
assume
A8: P[n];
let t be Nat such that
A9: t <= k;
A10: k = m*2 by NEWTON:6;
A11: m <> 0 by NEWTON:83;
A12: 1/2*(t/m) = (1*t)/(2*m) by XCMPLX_1:76
.= t/k by NEWTON:6;
per cases;
suppose
A13: t <= m;
then consider x being Nat such that
A14: x = m-t and
A15: f.(a+t/k*(b-a)) = 1/2*(f.a+f.(1/m*(x*a+t*b))) by A1,Lm9;
A16: x/m = m/m-t/m by A14
.= 1-t/m by A11,XCMPLX_1:60;
A17: 1/m*(x*a+t*b) = 1/m*(x*a) + 1/m*(t*b) by RLVECT_1:def 5
.= 1/m*(x*a) + t/m*b by RLVECT_1:def 7
.= x/m*a + t/m*b by RLVECT_1:def 7;
thus f.((1-t/k)*a + t/k*b) = f.(a+t/k*(b-a)) by Lm8
.= 1/2*f.a + 1/2*f.(1/m*(x*a+t*b)) by A15,RLVECT_1:def 5
.= 1/2*f.a + 1/2*((1-t/m)*f.a + t/m*f.b) by A16,A13,A17,A8
.= 1/2*f.a + (1/2*((1-t/m)*f.a) + 1/2*(t/m*f.b)) by RLVECT_1:def 5
.= 1/2*f.a + 1/2*((1-t/m)*f.a) + 1/2*(t/m*f.b) by RLVECT_1:def 3
.= 1/2*f.a + 1/2*(1-t/m)*f.a + 1/2*(t/m*f.b) by RLVECT_1:def 7
.= (1/2+1/2*(1-t/m))*f.a + 1/2*(t/m*f.b) by RLVECT_1:def 6
.= (1-t/k)*f.a + t/k*f.b by A12,RLVECT_1:def 7;
end;
suppose t >= m;
then consider x being Nat such that
A18: x = m+t-k and
A19: f.(a+t/k*(b-a)) = 1/2*(f.b+f.(1/m*((k-t)*a+x*b))) by A1,Lm10;
set w = t-m;
A20: w <= 2*m-m by A9,A10,XREAL_1:9;
A21: m/m = 1 by A11,XCMPLX_1:60;
A22: k/m = (2*m)/(1*m) by NEWTON:6
.= 2/1 by A11,XCMPLX_1:91;
A23: 1-t/m = m/m - t/m by A21
.= -(t/m - m/m)
.= -(t-m)/m;
A24: 1/m*(k-t) = k/m - t/m
.= 1 + 1 - t/m by A22
.= 1 - ((t-m)/m) by A23
.= 1-w/m;
1/m*x = w/m by A18,A10;
then 1/m*(k-t)*a+1/m*x*b
= (1-w/m)*a + w/m*b by A24;
then
A25: 1/2*f.(1/m*(k-t)*a+1/m*x*b)
= 1/2*((1-w/m)*f.a + w/m*f.b) by A20,A18,A10,A8;
A26: 1/2*(1-w/m)
= 1/2*(1-(t/m-m/m))
.= 1 - 1/2*(t/m) by A21
.= 1 - (1*t)/(2*m) by XCMPLX_1:76
.= 1-t/k by NEWTON:6;
thus f.((1-t/k)*a + t/k*b) = f.(a+t/k*(b-a)) by Lm8
.= 1/2*f.b + 1/2*f.(1/m*((k-t)*a+x*b)) by A19,RLVECT_1:def 5
.= 1/2*f.b + 1/2*f.(1/m*((k-t)*a)+1/m*(x*b)) by RLVECT_1:def 5
.= 1/2*f.b + 1/2*f.(1/m*(k-t)*a+1/m*(x*b)) by RLVECT_1:def 7
.= 1/2*f.b + 1/2*f.(1/m*(k-t)*a+1/m*x*b) by RLVECT_1:def 7
.= 1/2*f.b + 1/2*((1-w/m)*f.a + w/m*f.b)
by A25
.= 1/2*f.b + (1/2*((1-w/m)*f.a) + 1/2*(w/m*f.b)) by RLVECT_1:def 5
.= 1/2*((1-w/m)*f.a) + (1/2*f.b + 1/2*(w/m*f.b)) by RLVECT_1:def 3
.= 1/2*((1-w/m)*f.a) + (1/2*(f.b+(w/m*f.b))) by RLVECT_1:def 5
.= 1/2*((1-w/m)*f.a) + (1/2*(1*f.b+(w/m*f.b))) by RLVECT_1:def 8
.= 1/2*((1-w/m)*f.a) + 1/2*((1+w/m)*f.b) by RLVECT_1:def 6
.= 1/2*((1-w/m)*f.a) + 1/2*(1+w/m)*f.b by RLVECT_1:def 7
.= (1-t/k)*f.a + t/k*f.b by A26,RLVECT_1:def 7;
end;
end;
for n being Nat holds P[n] from NAT_1:sch 2(A2,A7);
hence thesis;
end;
end;
registration
let E,F;
cluster bijective isometric -> midpoints-preserving for Function of E,F;
coherence
proof
let f be Function of E,F such that
A1: f is bijective and
A2: f is isometric;
let a,b be Point of E;
set W = W(E,a,b);
set l = l(E,a,b);
set z = 1/2 * (a + b);
set z1 = 1/2 * (f.a + f.b);
set R = z-reflection;
set R1 = z1-reflection;
set h = R * (f/") * R1 * f;
now
let x be object;
assume x in l;
then ex g being UnOp of E st x = ||. g.z - z .|| & g in W(E,a,b);
hence x is real;
end;
then reconsider l as real-membered set by MEMBERED:def 3;
A3: rng f = [#]F by A1,FUNCT_2:def 3;
A4: f/" = (f qua Function)" by A1,TOPS_2:def 4;
then
A5: f/" is bijective by A1,GROUP_6:63;
R * (f/") is onto by A5,FUNCT_2:27;
then R * (f/") * R1 is onto by FUNCT_2:27;
then
A6: h is onto by A1,FUNCT_2:27;
f/" is isometric by A1,A2,Th12;
then R * (f/") is isometric by Th11;
then R * (f/") * R1 is isometric by Th11;
then
A7: h is isometric by A2,Th11;
A8: 2*z = (2*(1/2))*(a+b) by RLVECT_1:def 7
.= a + b by RLVECT_1:def 8;
then
A9: 2*z - a = b by Th4;
A10: 2*z - b = a by A8,Th4;
A11: 2*z1 = (2*(1/2))*(f.a+f.b) by RLVECT_1:def 7
.= f.a + f.b by RLVECT_1:def 8;
then
A12: 2*z1 - f.a = f.b by Th4;
A13: 2*z1 - f.b = f.a by A11,Th4;
A14: h.a = (R * (f/") * R1).(f.a) by FUNCT_2:15
.= (R * (f/")).(R1.(f.a)) by FUNCT_2:15
.= (R * (f/")).(f.b) by A12,Def4
.= R.((f/").(f.b)) by FUNCT_2:15
.= R.b by A4,A1,FUNCT_2:26
.= a by A10,Def4;
h.b = (R * (f/") * R1).(f.b) by FUNCT_2:15
.= (R * (f/")).(R1.(f.b)) by FUNCT_2:15
.= (R * (f/")).(f.a) by A13,Def4
.= R.((f/").(f.a)) by FUNCT_2:15
.= R.a by A4,A1,FUNCT_2:26
.= b by A9,Def4;
then
A15: h in W by A4,A1,A6,A7,A14;
rng R = [#]E by FUNCT_2:def 3;
then
A16: R/" * R = id dom R by TOPS_2:52
.= id E by FUNCT_2:def 1;
A17: f * (f/") = id F by A1,A3,TOPS_2:52;
A18: R/" * h = R/" * (R * (f/") * (R1 * f)) by RELAT_1:36
.= R/" * (R * (f/" * (R1 * f))) by RELAT_1:36
.= R/" * R * (f/" * (R1 * f)) by RELAT_1:36
.= R/" * R * (f/") * (R1 * f) by RELAT_1:36
.= R/" * R * (f/") * R1 * f by RELAT_1:36
.= (f/") * R1 * f by A16,FUNCT_2:17;
A19: f * (f/" * R1 * f) = f * (f/" * (R1 * f)) by RELAT_1:36
.= f * (f/") * (R1 * f) by RELAT_1:36
.= R1 * f by A17,FUNCT_2:17;
l = l(E,a,b);
then h.z = z by A15,Lm7;
then (R/"*h).z = R/".z by FUNCT_2:15;
then
A20: (f*(f/" * R1 * f)).z = f.(R/".z) by A18,FUNCT_2:15
.= (f*(R/")).z by FUNCT_2:15;
R1.(f.z) = (R1*f).z by FUNCT_2:15
.= (f*R).z by A20,A19,Th23
.= f.(R.z) by FUNCT_2:15
.= f.z by Th18;
hence thesis by Th18;
end;
end;
registration
let E,F;
cluster isometric midpoints-preserving -> Affine for Function of E,F;
coherence
proof
let f such that
A1: f is isometric and
A2: f is midpoints-preserving;
let a,b,t;
assume 0 <= t & t <= 1;
then t in [. 0,1 .] by XXREAL_1:1;
then consider s being Real_Sequence such that
A3: rng s c= DYADIC and
A4: s is convergent and
A5: lim s = t by Th2,MEASURE6:64;
A6: dom f = the carrier of E by FUNCT_2:def 1;
set stb = s * b;
set 1t = 1+(-s);
set s1ta = 1t * a;
set s1 = s1ta + stb;
set za = 1t * f.a;
set zb = s * f.b;
A7: f is_continuous_on dom f by A1,Th24;
A8: -s is convergent by A4;
A9: s1 is convergent by A4,NORMSP_1:19;
A10: lim s1ta = lim 1t * a by A8,Th10;
A11: lim -s = -lim s by A4,SEQ_2:10;
A12: lim 1t
= 1 + lim -s by A8,Th7
.= 1 - lim s by A11
.= 1-t by A5;
lim stb = lim s * b by A4,Th10;
then
A13: lim s1 = (1-t)*a + t*b by A5,A10,A12,A4,NORMSP_1:25;
A14: lim za = (1-t)*(f.a) by A8,A12,Th10;
A15: lim zb = t*(f.b) by A4,A5,Th10;
A16: f/*s1 = za+zb
proof
let n be Element of NAT;
A17: 1t.n = 1 + ((-s).n) by VALUED_1:2
.= 1 + -s.n by VALUED_1:8;
dom s = NAT by FUNCT_2:def 1;
then s.n in rng s by FUNCT_1:3;
then consider m being Nat such that
A18: s.n in dyadic(m) by A3,URYSOHN1:def 2;
consider i being Nat such that
A19: i <= 2|^m & s.n = i/(2|^m qua Complex)
by A18,URYSOHN1:def 1;
reconsider t = i as Nat;
A20: f.((1-t/(2|^m))*a + t/(2|^m)*b) = (1-t/(2|^m))*f.a + t/(2|^m)*f.b
by A19,Lm11,A2;
rng s1 c= dom f by A6;
hence (f/*s1).n = f.(s1.n) by FUNCT_2:108
.= f.(s1ta.n + stb.n) by NORMSP_1:def 2
.= f.((1-s.n) * a + stb.n) by A17,NDIFF_1:def 3
.= f.((1-s.n) * a + s.n * b) by NDIFF_1:def 3
.= (1-s.n) * f.a + s.n * f.b by A19,A20
.= 1t.n * f.a + zb.n by A17,NDIFF_1:def 3
.= za.n + zb.n by NDIFF_1:def 3
.= (za+zb).n by NORMSP_1:def 2;
end;
A21: rng s1 c= the carrier of E;
thus f.((1-t)*a+t*b) = f/.((1-t)*a+t*b)
.= lim (f/*s1) by A6,A7,A21,A9,A13,NFCONT_1:18
.= (1-t)*(f.a) + t*(f.b) by A14,A15,A4,A16,NORMSP_1:25;
end;
end;