src/HOL/Hahn_Banach/Hahn_Banach_Ext_Lemmas.thy
changeset 31795 be3e1cc5005c
parent 29252 ea97aa6aeba2
child 32962 69916a850301
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/src/HOL/Hahn_Banach/Hahn_Banach_Ext_Lemmas.thy	Wed Jun 24 21:46:54 2009 +0200
     1.3 @@ -0,0 +1,280 @@
     1.4 +(*  Title:      HOL/Hahn_Banach/Hahn_Banach_Ext_Lemmas.thy
     1.5 +    Author:     Gertrud Bauer, TU Munich
     1.6 +*)
     1.7 +
     1.8 +header {* Extending non-maximal functions *}
     1.9 +
    1.10 +theory Hahn_Banach_Ext_Lemmas
    1.11 +imports Function_Norm
    1.12 +begin
    1.13 +
    1.14 +text {*
    1.15 +  In this section the following context is presumed.  Let @{text E} be
    1.16 +  a real vector space with a seminorm @{text q} on @{text E}. @{text
    1.17 +  F} is a subspace of @{text E} and @{text f} a linear function on
    1.18 +  @{text F}. We consider a subspace @{text H} of @{text E} that is a
    1.19 +  superspace of @{text F} and a linear form @{text h} on @{text
    1.20 +  H}. @{text H} is a not equal to @{text E} and @{text "x\<^sub>0"} is
    1.21 +  an element in @{text "E - H"}.  @{text H} is extended to the direct
    1.22 +  sum @{text "H' = H + lin x\<^sub>0"}, so for any @{text "x \<in> H'"}
    1.23 +  the decomposition of @{text "x = y + a \<cdot> x"} with @{text "y \<in> H"} is
    1.24 +  unique. @{text h'} is defined on @{text H'} by @{text "h' x = h y +
    1.25 +  a \<cdot> \<xi>"} for a certain @{text \<xi>}.
    1.26 +
    1.27 +  Subsequently we show some properties of this extension @{text h'} of
    1.28 +  @{text h}.
    1.29 +
    1.30 +  \medskip This lemma will be used to show the existence of a linear
    1.31 +  extension of @{text f} (see page \pageref{ex-xi-use}). It is a
    1.32 +  consequence of the completeness of @{text \<real>}. To show
    1.33 +  \begin{center}
    1.34 +  \begin{tabular}{l}
    1.35 +  @{text "\<exists>\<xi>. \<forall>y \<in> F. a y \<le> \<xi> \<and> \<xi> \<le> b y"}
    1.36 +  \end{tabular}
    1.37 +  \end{center}
    1.38 +  \noindent it suffices to show that
    1.39 +  \begin{center}
    1.40 +  \begin{tabular}{l}
    1.41 +  @{text "\<forall>u \<in> F. \<forall>v \<in> F. a u \<le> b v"}
    1.42 +  \end{tabular}
    1.43 +  \end{center}
    1.44 +*}
    1.45 +
    1.46 +lemma ex_xi:
    1.47 +  assumes "vectorspace F"
    1.48 +  assumes r: "\<And>u v. u \<in> F \<Longrightarrow> v \<in> F \<Longrightarrow> a u \<le> b v"
    1.49 +  shows "\<exists>xi::real. \<forall>y \<in> F. a y \<le> xi \<and> xi \<le> b y"
    1.50 +proof -
    1.51 +  interpret vectorspace F by fact
    1.52 +  txt {* From the completeness of the reals follows:
    1.53 +    The set @{text "S = {a u. u \<in> F}"} has a supremum, if it is
    1.54 +    non-empty and has an upper bound. *}
    1.55 +
    1.56 +  let ?S = "{a u | u. u \<in> F}"
    1.57 +  have "\<exists>xi. lub ?S xi"
    1.58 +  proof (rule real_complete)
    1.59 +    have "a 0 \<in> ?S" by blast
    1.60 +    then show "\<exists>X. X \<in> ?S" ..
    1.61 +    have "\<forall>y \<in> ?S. y \<le> b 0"
    1.62 +    proof
    1.63 +      fix y assume y: "y \<in> ?S"
    1.64 +      then obtain u where u: "u \<in> F" and y: "y = a u" by blast
    1.65 +      from u and zero have "a u \<le> b 0" by (rule r)
    1.66 +      with y show "y \<le> b 0" by (simp only:)
    1.67 +    qed
    1.68 +    then show "\<exists>u. \<forall>y \<in> ?S. y \<le> u" ..
    1.69 +  qed
    1.70 +  then obtain xi where xi: "lub ?S xi" ..
    1.71 +  {
    1.72 +    fix y assume "y \<in> F"
    1.73 +    then have "a y \<in> ?S" by blast
    1.74 +    with xi have "a y \<le> xi" by (rule lub.upper)
    1.75 +  } moreover {
    1.76 +    fix y assume y: "y \<in> F"
    1.77 +    from xi have "xi \<le> b y"
    1.78 +    proof (rule lub.least)
    1.79 +      fix au assume "au \<in> ?S"
    1.80 +      then obtain u where u: "u \<in> F" and au: "au = a u" by blast
    1.81 +      from u y have "a u \<le> b y" by (rule r)
    1.82 +      with au show "au \<le> b y" by (simp only:)
    1.83 +    qed
    1.84 +  } ultimately show "\<exists>xi. \<forall>y \<in> F. a y \<le> xi \<and> xi \<le> b y" by blast
    1.85 +qed
    1.86 +
    1.87 +text {*
    1.88 +  \medskip The function @{text h'} is defined as a @{text "h' x = h y
    1.89 +  + a \<cdot> \<xi>"} where @{text "x = y + a \<cdot> \<xi>"} is a linear extension of
    1.90 +  @{text h} to @{text H'}.
    1.91 +*}
    1.92 +
    1.93 +lemma h'_lf:
    1.94 +  assumes h'_def: "h' \<equiv> \<lambda>x. let (y, a) =
    1.95 +      SOME (y, a). x = y + a \<cdot> x0 \<and> y \<in> H in h y + a * xi"
    1.96 +    and H'_def: "H' \<equiv> H + lin x0"
    1.97 +    and HE: "H \<unlhd> E"
    1.98 +  assumes "linearform H h"
    1.99 +  assumes x0: "x0 \<notin> H"  "x0 \<in> E"  "x0 \<noteq> 0"
   1.100 +  assumes E: "vectorspace E"
   1.101 +  shows "linearform H' h'"
   1.102 +proof -
   1.103 +  interpret linearform H h by fact
   1.104 +  interpret vectorspace E by fact
   1.105 +  show ?thesis
   1.106 +  proof
   1.107 +    note E = `vectorspace E`
   1.108 +    have H': "vectorspace H'"
   1.109 +    proof (unfold H'_def)
   1.110 +      from `x0 \<in> E`
   1.111 +      have "lin x0 \<unlhd> E" ..
   1.112 +      with HE show "vectorspace (H + lin x0)" using E ..
   1.113 +    qed
   1.114 +    {
   1.115 +      fix x1 x2 assume x1: "x1 \<in> H'" and x2: "x2 \<in> H'"
   1.116 +      show "h' (x1 + x2) = h' x1 + h' x2"
   1.117 +      proof -
   1.118 +	from H' x1 x2 have "x1 + x2 \<in> H'"
   1.119 +          by (rule vectorspace.add_closed)
   1.120 +	with x1 x2 obtain y y1 y2 a a1 a2 where
   1.121 +          x1x2: "x1 + x2 = y + a \<cdot> x0" and y: "y \<in> H"
   1.122 +          and x1_rep: "x1 = y1 + a1 \<cdot> x0" and y1: "y1 \<in> H"
   1.123 +          and x2_rep: "x2 = y2 + a2 \<cdot> x0" and y2: "y2 \<in> H"
   1.124 +          unfolding H'_def sum_def lin_def by blast
   1.125 +	
   1.126 +	have ya: "y1 + y2 = y \<and> a1 + a2 = a" using E HE _ y x0
   1.127 +	proof (rule decomp_H') txt_raw {* \label{decomp-H-use} *}
   1.128 +          from HE y1 y2 show "y1 + y2 \<in> H"
   1.129 +            by (rule subspace.add_closed)
   1.130 +          from x0 and HE y y1 y2
   1.131 +          have "x0 \<in> E"  "y \<in> E"  "y1 \<in> E"  "y2 \<in> E" by auto
   1.132 +          with x1_rep x2_rep have "(y1 + y2) + (a1 + a2) \<cdot> x0 = x1 + x2"
   1.133 +            by (simp add: add_ac add_mult_distrib2)
   1.134 +          also note x1x2
   1.135 +          finally show "(y1 + y2) + (a1 + a2) \<cdot> x0 = y + a \<cdot> x0" .
   1.136 +	qed
   1.137 +	
   1.138 +	from h'_def x1x2 E HE y x0
   1.139 +	have "h' (x1 + x2) = h y + a * xi"
   1.140 +          by (rule h'_definite)
   1.141 +	also have "\<dots> = h (y1 + y2) + (a1 + a2) * xi"
   1.142 +          by (simp only: ya)
   1.143 +	also from y1 y2 have "h (y1 + y2) = h y1 + h y2"
   1.144 +          by simp
   1.145 +	also have "\<dots> + (a1 + a2) * xi = (h y1 + a1 * xi) + (h y2 + a2 * xi)"
   1.146 +          by (simp add: left_distrib)
   1.147 +	also from h'_def x1_rep E HE y1 x0
   1.148 +	have "h y1 + a1 * xi = h' x1"
   1.149 +          by (rule h'_definite [symmetric])
   1.150 +	also from h'_def x2_rep E HE y2 x0
   1.151 +	have "h y2 + a2 * xi = h' x2"
   1.152 +          by (rule h'_definite [symmetric])
   1.153 +	finally show ?thesis .
   1.154 +      qed
   1.155 +    next
   1.156 +      fix x1 c assume x1: "x1 \<in> H'"
   1.157 +      show "h' (c \<cdot> x1) = c * (h' x1)"
   1.158 +      proof -
   1.159 +	from H' x1 have ax1: "c \<cdot> x1 \<in> H'"
   1.160 +          by (rule vectorspace.mult_closed)
   1.161 +	with x1 obtain y a y1 a1 where
   1.162 +            cx1_rep: "c \<cdot> x1 = y + a \<cdot> x0" and y: "y \<in> H"
   1.163 +          and x1_rep: "x1 = y1 + a1 \<cdot> x0" and y1: "y1 \<in> H"
   1.164 +          unfolding H'_def sum_def lin_def by blast
   1.165 +	
   1.166 +	have ya: "c \<cdot> y1 = y \<and> c * a1 = a" using E HE _ y x0
   1.167 +	proof (rule decomp_H')
   1.168 +          from HE y1 show "c \<cdot> y1 \<in> H"
   1.169 +            by (rule subspace.mult_closed)
   1.170 +          from x0 and HE y y1
   1.171 +          have "x0 \<in> E"  "y \<in> E"  "y1 \<in> E" by auto
   1.172 +          with x1_rep have "c \<cdot> y1 + (c * a1) \<cdot> x0 = c \<cdot> x1"
   1.173 +            by (simp add: mult_assoc add_mult_distrib1)
   1.174 +          also note cx1_rep
   1.175 +          finally show "c \<cdot> y1 + (c * a1) \<cdot> x0 = y + a \<cdot> x0" .
   1.176 +	qed
   1.177 +	
   1.178 +	from h'_def cx1_rep E HE y x0 have "h' (c \<cdot> x1) = h y + a * xi"
   1.179 +          by (rule h'_definite)
   1.180 +	also have "\<dots> = h (c \<cdot> y1) + (c * a1) * xi"
   1.181 +          by (simp only: ya)
   1.182 +	also from y1 have "h (c \<cdot> y1) = c * h y1"
   1.183 +          by simp
   1.184 +	also have "\<dots> + (c * a1) * xi = c * (h y1 + a1 * xi)"
   1.185 +          by (simp only: right_distrib)
   1.186 +	also from h'_def x1_rep E HE y1 x0 have "h y1 + a1 * xi = h' x1"
   1.187 +          by (rule h'_definite [symmetric])
   1.188 +	finally show ?thesis .
   1.189 +      qed
   1.190 +    }
   1.191 +  qed
   1.192 +qed
   1.193 +
   1.194 +text {* \medskip The linear extension @{text h'} of @{text h}
   1.195 +  is bounded by the seminorm @{text p}. *}
   1.196 +
   1.197 +lemma h'_norm_pres:
   1.198 +  assumes h'_def: "h' \<equiv> \<lambda>x. let (y, a) =
   1.199 +      SOME (y, a). x = y + a \<cdot> x0 \<and> y \<in> H in h y + a * xi"
   1.200 +    and H'_def: "H' \<equiv> H + lin x0"
   1.201 +    and x0: "x0 \<notin> H"  "x0 \<in> E"  "x0 \<noteq> 0"
   1.202 +  assumes E: "vectorspace E" and HE: "subspace H E"
   1.203 +    and "seminorm E p" and "linearform H h"
   1.204 +  assumes a: "\<forall>y \<in> H. h y \<le> p y"
   1.205 +    and a': "\<forall>y \<in> H. - p (y + x0) - h y \<le> xi \<and> xi \<le> p (y + x0) - h y"
   1.206 +  shows "\<forall>x \<in> H'. h' x \<le> p x"
   1.207 +proof -
   1.208 +  interpret vectorspace E by fact
   1.209 +  interpret subspace H E by fact
   1.210 +  interpret seminorm E p by fact
   1.211 +  interpret linearform H h by fact
   1.212 +  show ?thesis
   1.213 +  proof
   1.214 +    fix x assume x': "x \<in> H'"
   1.215 +    show "h' x \<le> p x"
   1.216 +    proof -
   1.217 +      from a' have a1: "\<forall>ya \<in> H. - p (ya + x0) - h ya \<le> xi"
   1.218 +	and a2: "\<forall>ya \<in> H. xi \<le> p (ya + x0) - h ya" by auto
   1.219 +      from x' obtain y a where
   1.220 +          x_rep: "x = y + a \<cdot> x0" and y: "y \<in> H"
   1.221 +	unfolding H'_def sum_def lin_def by blast
   1.222 +      from y have y': "y \<in> E" ..
   1.223 +      from y have ay: "inverse a \<cdot> y \<in> H" by simp
   1.224 +      
   1.225 +      from h'_def x_rep E HE y x0 have "h' x = h y + a * xi"
   1.226 +	by (rule h'_definite)
   1.227 +      also have "\<dots> \<le> p (y + a \<cdot> x0)"
   1.228 +      proof (rule linorder_cases)
   1.229 +	assume z: "a = 0"
   1.230 +	then have "h y + a * xi = h y" by simp
   1.231 +	also from a y have "\<dots> \<le> p y" ..
   1.232 +	also from x0 y' z have "p y = p (y + a \<cdot> x0)" by simp
   1.233 +	finally show ?thesis .
   1.234 +      next
   1.235 +	txt {* In the case @{text "a < 0"}, we use @{text "a\<^sub>1"}
   1.236 +          with @{text ya} taken as @{text "y / a"}: *}
   1.237 +	assume lz: "a < 0" then have nz: "a \<noteq> 0" by simp
   1.238 +	from a1 ay
   1.239 +	have "- p (inverse a \<cdot> y + x0) - h (inverse a \<cdot> y) \<le> xi" ..
   1.240 +	with lz have "a * xi \<le>
   1.241 +          a * (- p (inverse a \<cdot> y + x0) - h (inverse a \<cdot> y))"
   1.242 +          by (simp add: mult_left_mono_neg order_less_imp_le)
   1.243 +	
   1.244 +	also have "\<dots> =
   1.245 +          - a * (p (inverse a \<cdot> y + x0)) - a * (h (inverse a \<cdot> y))"
   1.246 +	  by (simp add: right_diff_distrib)
   1.247 +	also from lz x0 y' have "- a * (p (inverse a \<cdot> y + x0)) =
   1.248 +          p (a \<cdot> (inverse a \<cdot> y + x0))"
   1.249 +          by (simp add: abs_homogenous)
   1.250 +	also from nz x0 y' have "\<dots> = p (y + a \<cdot> x0)"
   1.251 +          by (simp add: add_mult_distrib1 mult_assoc [symmetric])
   1.252 +	also from nz y have "a * (h (inverse a \<cdot> y)) =  h y"
   1.253 +          by simp
   1.254 +	finally have "a * xi \<le> p (y + a \<cdot> x0) - h y" .
   1.255 +	then show ?thesis by simp
   1.256 +      next
   1.257 +	txt {* In the case @{text "a > 0"}, we use @{text "a\<^sub>2"}
   1.258 +          with @{text ya} taken as @{text "y / a"}: *}
   1.259 +	assume gz: "0 < a" then have nz: "a \<noteq> 0" by simp
   1.260 +	from a2 ay
   1.261 +	have "xi \<le> p (inverse a \<cdot> y + x0) - h (inverse a \<cdot> y)" ..
   1.262 +	with gz have "a * xi \<le>
   1.263 +          a * (p (inverse a \<cdot> y + x0) - h (inverse a \<cdot> y))"
   1.264 +          by simp
   1.265 +	also have "\<dots> = a * p (inverse a \<cdot> y + x0) - a * h (inverse a \<cdot> y)"
   1.266 +	  by (simp add: right_diff_distrib)
   1.267 +	also from gz x0 y'
   1.268 +	have "a * p (inverse a \<cdot> y + x0) = p (a \<cdot> (inverse a \<cdot> y + x0))"
   1.269 +          by (simp add: abs_homogenous)
   1.270 +	also from nz x0 y' have "\<dots> = p (y + a \<cdot> x0)"
   1.271 +          by (simp add: add_mult_distrib1 mult_assoc [symmetric])
   1.272 +	also from nz y have "a * h (inverse a \<cdot> y) = h y"
   1.273 +          by simp
   1.274 +	finally have "a * xi \<le> p (y + a \<cdot> x0) - h y" .
   1.275 +	then show ?thesis by simp
   1.276 +      qed
   1.277 +      also from x_rep have "\<dots> = p x" by (simp only:)
   1.278 +      finally show ?thesis .
   1.279 +    qed
   1.280 +  qed
   1.281 +qed
   1.282 +
   1.283 +end