<TeXmacs|1.0.2.2>

<style|<tuple|seminar|vdh>>

<\body>
  <with|paragraph mode|center|<with|color|red|<strong|<\with|font size|1.41>
    A differential

    intermediate value theorem
  </with>>>>

  \;

  \;

  \;

  <\with|paragraph mode|center>
    <with|color|blue|<with|mode|math|\<infty\><space|-0.4spc>\<infty\><space|-0.4spc>\<infty\>>>

    \;

    <with|font shape|small-caps|by Joris van der Hoeven><vspace|0.5fn>

    \;

    C.N.R.S., Université Paris-Sud<vspace|0.5fn>

    \;

    <with|font|chancellary|Groningen, 29/5/2001>

    \;

    <with|color|blue|<with|mode|math|\<infty\><space|-0.4spc>\<infty\><space|-0.4spc>\<infty\>>>
  </with>

  \;

  \;

  \;

  <with|paragraph mode|center|<tabular|<tformat|<cwith|1|-1|1|-1|cell
  mode|c>|<cwith|1|2|1|2|cell valign|c>|<table|<row|<cell|<postscript|gleeson_head.gif|/2|/2||||>>|<cell|<with|font
  size|0.84| Made with <with|color|dark green|GNU
  <with|color|brown|T<space|-0.2spc>><rsub|<with|index level|0|font
  shape|small-caps|E>><with|color|brown|X<space|-0.2spc>><rsub|<with|index
  level|0|font shape|small-caps|m<space|-0.2spc>a<space|-0.4spc>c<space|-0.2spc>s>>>
  (<tt|www.texmacs.org>)>>>>>>><new_page>

  <with|paragraph mode|center|<section|PART I>>

  <section|Transseries>

  Field <with|mode|math|\<bbb-T\>> of grid-based transseries in
  <with|mode|math|x\<succ\>1>:

  <\itemize>
    <item>Serial structure and total relation
    <with|mode|math|\<preccurlyeq\>>.

    <item>Stable under exp and log.

    <item>Real algebraically closed.

    <item>Stable under differentiation, integration.

    <item>Stable under composition, inversion.
  </itemize>

  <section|Origins>

  <\description>
    <expand|item*|Model theory>

    <\itemize>
      <item>Dahn & Göring, 1984, 1986

      <item>Van den Dries, Ressayre, etc. 199*
    </itemize>

    <expand|item*|Analysis>

    <itemize|<item>Écalle, 1989, 1992>

    <expand|item*|Computer algebra>

    <\itemize>
      <item>Shackell, Salvy, etc. 1990-*

      <item>vdH, 1994-*
    </itemize>
  </description>

  <new_page>

  <section|Examples>

  <\itemize>
    <item><with|mode|math|1+x<rsup|-1>+x<rsup|-2>+x<rsup|-e>+x<rsup|-3>+x<rsup|-e-1>+\<cdots\>>

    <item><with|mode|math|1+<frac|1|x>+<frac|1|x<rsup|2>>+\<cdots\>+e<rsup|-x>+<frac|e<rsup|-x>|x>+\<cdots\>+e<rsup|-2*x>+\<cdots\>>

    <item><with|mode|math|><with|mode|math|1+2<rsup|-x>+3<rsup|-x>+4<rsup|-x>+\<cdots\>>

    <item><with|mode|math|x<rsup|-1>+x<rsup|-\<pi\>>+x<rsup|-\<pi\><rsup|2>>+x<rsup|-\<pi\><rsup|3>>+\<cdots\>>

    <item><with|mode|math|x+<sqrt|x>+<sqrt|<sqrt|x>>+<sqrt|<sqrt|<sqrt|x>>>+\<cdots\>>

    <item><with|mode|math|e<rsup|e<rsup|x><space|0.2spc>+<space|0.2spc><frac|e<rsup|x>|x><space|0.2spc>+<space|0.2spc><frac|e<rsup|x>|x<rsup|2>><space|0.2spc>+\<cdots\>>+x<rsup|-1>*e<rsup|e<rsup|x><space|0.2spc>+<space|0.2spc><frac|e<rsup|x>|x><space|0.2spc>+<space|0.2spc><frac|e<rsup|x>|x<rsup|2>><space|0.2spc>+\<cdots\>>+\<cdots\>>

    <item><with|mode|math|\<Gamma\>(x-\<pi\>)+log
    \<Gamma\>(e<rsup|\<Gamma\>(x<rsup|2>)>)*x<rsup|x<rsup|x<rsup|x>>>>

    <item><with|mode|math|<frac|1|x>+<frac|1|e<rsup|x>>+<frac|1|e<rsup|e<rsup|x>>>+<frac|1|e<rsup|e<rsup|e<rsup|x>>>>+\<cdots\>>

    <item><with|mode|math|f(x)=<frac|1|x>+f(x<rsup|2>)+f(e<rsup|log<rsup|2>
    x>)>

    <item><with|mode|math|e<rsup|<sqrt|x>+e<rsup|<sqrt|log
    x>+e<rsup|<sqrt|log log x>+\<udots\>>>>><new_page>
  </itemize>

  <section|Well-ordered power series>

  <\itemize>
    <item>Totally ordered constant field <with|mode|math|C>.

    <item>Monomial group <with|mode|math|\<frak-M\>>, with total ordering
    <with|mode|math|\<succcurlyeq\>>.

    <item>[Hahn 1907] Set of <with|color|blue|well-ordered series>

    <expand|equation*|C[[\<frak-M\>]]={f:\<frak-M\>\<rightarrow\>C\|supp
    f<with|mode|text| is well-ordered>}>

    forms a totally ordered field.

    <item><with|mode|math|f=c<rsub|f>*\<frak-d\><rsub|f>*(1+\<delta\><rsub|f>)>

    <item><with|mode|math|f\<preccurlyeq\>g\<Leftrightarrow\>\<frak-d\><rsub|f>\<preccurlyeq\>\<frak-d\><rsub|g>>

    <item>Canonical decomposition:

    <expand|eqnarray*|<tformat|<table|<row|<cell|f=f<rsup|\<uparrow\>>>|<cell|+f<rsup|=>+>|<cell|f<rsup|\<downarrow\>>>>|<row|<cell|\<\|\|\>>|<cell|\<\|\|\>>|<cell|\<\|\|\>>>|<row|<cell|<big|sum><rsub|\<frak-m\>\<succ\>1>f<rsub|\<frak-m\>>*\<frak-m\>>|<cell|f<rsub|1>>|<cell|<big|sum><rsub|\<frak-m\>\<prec\>1>f<rsub|\<frak-m\>>*\<frak-m\>>>>>>
  </itemize>

  <section|Grid-based series>

  <with|mode|math|f> grid-based <with|mode|math|\<Longleftrightarrow\>><with|mode|math|\<exists\>>
  <with|mode|math|\<frak-m\><rsub|1>,\<ldots\>,\<frak-m\><rsub|k>\<prec\>1>
  and <with|mode|math|\<frak-n\><rsub|1>,\<ldots\>,\<frak-n\><rsub|l>> :

  <expand|equation*|supp f\<subseteq\>{\<frak-m\><rsub|1>,\<ldots\>,\<frak-m\><rsub|k>}<rsup|\<ast\>>*{\<frak-n\><rsub|1>,\<ldots\>,\<frak-n\><rsub|l>}.>

  <with|mode|math|C<gb|\<frak-M\>>\<subseteq\>C[[\<frak-M\>]]>: field of
  <with|color|blue|grid-based series>.<new_page>

  <section|Construction of <with|mode|math|\<bbb-T\>>>

  <subsection|Logarithmic transseries>

  Start with monomial group

  <expand|equation*|\<frak-L\>=\<frak-E\><rsub|0>=x<rsup|\<bbb-R\>>*(log
  x)<rsup|\<bbb-R\>>*(log log x)<rsup|\<bbb-R\>>*\<cdots\>>

  and logarithm on <with|mode|math|\<bbb-R\><gb|\<frak-L\>><rsup|+><rsub|\<ast\>>>:

  <expand|equation*|<expand|tabular*|<tformat|<table|<row|<cell|log
  (c*x<rsup|\<alpha\><rsub|0>>*\<cdots\>log<rsub|l><rsup|\<alpha\><rsub|l>>
  x(1+\<delta\>))=>>|<row|<cell|log c+\<alpha\><rsub|0>*log
  x+\<cdots\>+\<alpha\><rsub|l>*log<rsub|l+1> x+log (1+\<delta\>).>>>>>>

  <subsection|Inductive step>

  Assume <with|mode|math|\<frak-E\><rsub|n>> given, with logarithm on
  <with|mode|math|\<bbb-R\><gb|\<frak-E\><rsub|n>><rsup|+><rsub|\<ast\>>>.

  <expand|equation*|\<frak-E\><rsub|n+1>=exp
  \<bbb-R\><gb|\<frak-E\><rsub|n>><rsup|\<uparrow\>>,>

  with

  <expand|equation*|exp f<rsup|<space|0.2spc>\<uparrow\>>\<succcurlyeq\>exp
  g<rsup|\<uparrow\>>\<Leftrightarrow\>f\<geqslant\>g.>

  Take

  <expand|equation*|log (c*e<rsup|f<rsup|<space|0.2spc>\<uparrow\>>>*(1+\<delta\>))=log
  c+f<rsup|<space|0.2spc>\<uparrow\>>+log (1+\<delta\>).>

  Inductive limit: <with|mode|math|\<bbb-T\>=C<gb|\<frak-E\><rsub|0>\<cup\>\<frak-E\><rsub|1>\<cup\>\<cdots\>>>.<vspace|1fn>

  <with|color|red|<with|font series|bold|Example.>>
  <with|mode|math|e<rsup|e<rsup|x>(1+<space|0.2spc><frac|1|x><space|0.2spc>+<space|0.2spc><frac|1|x<rsup|2>><space|0.2spc>+\<cdots\>)>\<in\>\<frak-E\><rsub|2>>.<new_page>

  <section|Transbases>

  A <with|color|blue|transbasis> is a tuple
  <with|mode|math|\<frak-B\>=(\<frak-b\><rsub|1>,\<ldots\>,\<frak-b\><rsub|n>)>
  with

  <\description>
    <expand|item*|TB0.><with|mode|math|1\<prec\>\<frak-b\><rsub|1>\<prec\>\<cdots\>\<prec\>\<frak-b\><rsub|n>>
    transmonomials.

    <expand|item*|TB1.><with|mode|math|\<frak-b\><rsub|1>=exp<rsub|l> x>,
    with <with|mode|math|l\<in\>\<bbb-Z\>>.

    <expand|item*|TB2.><with|mode|math|\<frak-b\><rsub|i>\<in\>exp
    \<bbb-R\><gb|\<frak-b\><rsub|1>;\<ldots\>;\<frak-b\><rsub|i-1>>> for
    <with|mode|math|i\<gtr\>1>.
  </description>

  Recursive expansion of <with|mode|math|f\<in\>C<gb|\<frak-b\><rsub|1>;\<ldots\>;\<frak-b\><rsub|n>>>:

  <expand|eqnarray*|<tformat|<table|<row|<cell|f>|<cell|=>|<cell|<big|sum><rsub|\<alpha\><rsub|n>><rsup|>f<rsub|\<alpha\><rsub|n>>*\<frak-b\><rsub|n><rsup|\<alpha\><rsub|n>>;>>|<row|<cell|f<rsub|\<alpha\><rsub|n>>>|<cell|=>|<cell|<big|sum><rsub|\<alpha\><rsub|n-1>><rsup|>f<rsub|\<alpha\><rsub|n>,\<alpha\><rsub|n-1>>*\<frak-b\><rsub|n-1><rsup|\<alpha\><rsub|n-1>>;>>|<row|<cell|>|<cell|\<vdots\>>|<cell|>>|<row|<cell|f<rsub|\<alpha\><rsub|n>,\<ldots\>,\<alpha\><rsub|2>>>|<cell|=>|<cell|<big|sum><rsub|\<alpha\><rsub|1>><rsup|>f<rsub|\<alpha\><rsub|n>,\<ldots\>,\<alpha\><rsub|1>>*\<frak-b\><rsub|1><rsup|\<alpha\><rsub|1>>.>>>>>

  <subsection|Incomplete transbasis theorem>

  <theorem|Let <with|mode|math|\<frak-B\><rsub|0>> be a transbasis and
  <with|mode|math|f\<in\>\<bbb-T\>> a transseries. Then there exists a
  supertransbasis <with|mode|math|\<frak-B\>\<supseteq\>\<frak-B\><rsub|0>>,
  such that <with|mode|math|f\<in\>\<bbb-R\><gb|\<frak-B\><rsup|C>>.>>

  <with|color|red|<with|font series|bold|Example.>> <with|mode|math|log
  (x+e<rsup|<frac|-x<rsup|2>|1-x<rsup|-1>>>)\<in\>\<bbb-R\><gb|log
  x;x;e<rsup|x<rsup|2>+x>>><space|-0.4spc>.<new_page>

  <section|Differentiation>

  <subsection|Differentiation on <with|mode|math|\<bbb-L\>>>

  For <with|mode|math|\<frak-m\>=x<rsup|\<alpha\><rsub|0>>*\<cdots\>(log<rsub|l>
  x)<rsup|\<alpha\><rsub|l>><rsup|>\<in\>\<frak-L\>>:

  <expand|equation*|\<frak-m\><rprime|'>=\<frak-m\>*<left|(><frac|\<alpha\><rsub|0>|x>+\<cdots\>+<frac|\<alpha\><rsub|l>|x*\<cdots\>*log<rsub|l>
  x><right|)>.>

  ``Extension by strong linearity'' to <with|mode|math|\<bbb-L\>=\<bbb-R\><gb|\<frak-L\>>>.

  <subsection|Differentiation on <with|mode|math|\<bbb-T\>>>

  For <with|mode|math|\<frak-m\>=e<rsup|f<rsup|<space|0.2spc>\<uparrow\>>>\<in\>exp
  \<bbb-R\><gb|\<frak-E\><rsub|n>><rsup|\<uparrow\>>>:

  <expand|equation*|\<frak-m\><rprime|'>=(f<rsup|<space|0.2spc>\<uparrow\>>)<rprime|'>*\<frak-m\>>

  ``Extension by strong linearity'' to <with|mode|math|\<bbb-R\><gb|\<frak-E\><rsub|n+1>>>.

  <section|Upward & downward shifting>

  <\itemize>
    <item><with|mode|math|f\<uparrow\>=f\<circ\>exp>.

    <item><with|mode|math|f\<downarrow\>=f\<circ\>log.>
  </itemize>

  <new_page>

  <with|paragraph mode|center|<section|Part II>>

  <subsection|Algebraic differential polynomials>

  <expand|eqnarray*|<tformat|<table|<row|<cell|P>|<cell|=>|<cell|<big|sum><rsub|d>P<rsub|d>>>|<row|<cell|P<rsub|d>>|<cell|=>|<cell|<big|sum><rsub|i<rsub|0>+\<cdots\>+i<rsub|r>=d>P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>*f<rsup|<space|0.2spc>i<rsub|0>>*\<cdots\>*(f<rsup|(r)>)<rsup|i<rsub|r>>>>|<row|<cell|deg
  P>|<cell|=>|<cell|max {i<rsub|0>+\<cdots\>+i<rsub|r>\|P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>\<neq\>0}>>|<row|<cell|\<\|\|\>P\<\|\|\>>|<cell|=>|<cell|max
  {i<rsub|1>+\<cdots\>+r*i<rsub|r>\|P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>\<neq\>0}>>|<row|<cell|P<rsub|+h>(f)>|<cell|=>|<cell|P(f+h)>>|<row|<cell|P<rsub|\<times\>h>(f)>|<cell|=>|<cell|P(f*h)>>|<row|<cell|P<space|0.2spc>\<uparrow\>(f\<uparrow\>)>|<cell|=>|<cell|P(f)\<uparrow\>>>>>>

  <new_page>

  <section|Dominant parts>

  <expand|eqnarray*|<tformat|<table|<row|<cell|\<frak-d\><rsub|P>>|<cell|=>|<cell|max<rsub|\<preccurlyeq\>>
  {\<frak-d\><rsub|P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>>\|P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>\<neq\>0}>>|<row|<cell|D<rsub|P>>|<cell|=>|<cell|<big|sum><rsub|i<rsub|0>,\<ldots\>,i<rsub|r>>P<rsub|i<rsub|0>,\<ldots\>,i<rsub|r>,\<frak-d\><rsub|P>>*<space|0.2spc>c<rsup|i<rsub|0>>*\<cdots\>*(c<rsup|(r)>)<rsup|i<rsub|r>>>>>>>

  <\theorem>
    For some <with|mode|math|k\<leqslant\>\<\|\|\>P\<\|\|\>>, there exists a
    <with|mode|math|N<rsub|P>=N\<in\>\<bbb-R\>[c]*(c<rprime|'>)<rsup|\<bbb-N\>>>,
    such that for every <with|mode|math|l\<geqslant\>k>:

    <expand|equation*|D<rsub|P\<uparrow\><rsub|l>>=N<rsub|P>.>
  </theorem>

  <with|color|red|<with|font series|bold|Example.>>

  <expand|eqnarray*|<tformat|<table|<row|<cell|P>|<cell|=>|<cell|f*f<rprime|''>-(f<rprime|'>)<rsup|2>>>|<row|<cell|P<space|0.2spc>\<uparrow\>>|<cell|=>|<cell|e<rsup|-x>*(f*f<rprime|''>-f*f<rprime|'>-(f<rprime|'>)<rsup|2>)>>|<row|<cell|P<space|0.2spc>\<uparrow\>\<uparrow\>>|<cell|=>|<cell|-e<rsup|-e<rsup|x>-x>*f*f<rprime|'>+e<rsup|-e<rsup|x>-2*x>*(f*f<rprime|''>-(f<rprime|'>)<rsup|2>)>>|<row|<cell|N<rsub|P>>|<cell|=>|<cell|-f*f<rprime|'>>>>>>

  <with|mode|math|N<rsub|P>> : Newton polynomial associated to
  <with|mode|math|1>.<new_page>

  <section|Differential Newton polygons>

  <equation|<label|ade>P(f)=0<space|5spc>(f\<prec\>\<frak-v\>)>

  <subsection|Newton degree>

  <\itemize>
    <item><with|mode|math|\<frak-m\>\<prec\>\<frak-v\>>
    <with|color|blue|pot.<inactive|<group|>> dom.<inactive|<group|>>
    monomial> if <with|mode|math|N<rsub|P<rsub|\<times\>\<frak-m\>>>\<nin\>C>.

    <item><with|mode|math|\<tau\>=c*\<frak-m\>\<prec\>\<frak-v\>>
    <with|color|blue|pot.<inactive|<group|>> dom.<inactive|<group|>> term> if
    <with|mode|math|N<rsub|P<rsub|\<times\>\<frak-m\>>>(c)=0.>

    <item><with|color|blue|Multiplicity> of <with|mode|math|\<tau\>>: mult.
    of <with|mode|math|c> as root of <with|mode|math|N<rsub|<space|-0.4spc>P<rsub|\<times\>\<frak-m\>>>>.

    <item><with|color|blue|Newton degree>: max.<inactive|<group|>>
    deg.<inactive|<group|>> of <with|mode|math|N<rsub|<space|-0.4spc>P<rsub|\<times\>\<frak-m\>>>>
    (<with|mode|math|\<frak-m\>\<prec\>\<frak-v\>>).
  </itemize>

  <subsection|Potential dominant monomial types>

  <\description>
    <expand|item*|Algebraic.><with|mode|math|N<rsub|P<rsub|\<times\>\<frak-m\>>>\<in\>\<bbb-R\>[c]\\C>.

    <expand|item*|Differential.><with|mode|math|N<rsub|P<rsub|\<times\>\<frak-m\>>>\<in\>\<bbb-R\>*(c<rprime|'>)<rsup|\<bbb-N\><rsup|\<ast\>>>>.

    <expand|item*|Mixed.><with|mode|math|N<rsub|P<rsub|\<times\>\<frak-m\>>>\<in\>(\<bbb-R\>[c]\\\<bbb-R\>)*(c<rprime|'>)<rsup|\<bbb-N\><rsup|\<ast\>>>.>
  </description>

  <subsection|Algebraic and mixed p.d.m.s>

  <proposition|Assume that <with|mode|math|P<rsub|i>\<neq\>0> and
  <with|mode|math|P<rsub|j>\<neq\>0>. Then there exists a unique equalizer
  <with|mode|math|\<frak-m\>=\<frak-e\><rsub|i,j>>, such that
  <with|mode|math|N<rsub|(P<rsub|i>+P<rsub|j>)<rsub|\<times\>\<frak-m\>>>> is
  non homogeneous.>

  <new_page>

  <subsection|Differential p.d.m.s>

  Ricatti polynomials:

  <expand|equation*|P<rsub|i>(f)=R<rsub|P,i>(f<rprime|\<dag\>>)*f<rsup|<space|0.2spc>i>>

  <\proposition>
    The monomial <with|mode|math|\<frak-m\>\<prec\>\<frak-v\>> is a potential
    dominant monomial w.r.t.

    <expand|equation*|P<rsub|i>(f)=0>

    if and only if

    <expand|equation*|R<rsub|P,i,\<frak-m\><rprime|\<dag\>>>(f<rprime|\<dag\>>)=0<space|5spc>(f<rprime|\<dag\>>\<prec\><with|formula
    style|false|<frac|1|x*log x*log log x*\<cdots\>>)>>

    has strictly positive Newton degree.
  </proposition>

  <subsection|Refinements>

  Change of variables + asymptotic constraint

  <equation|<label|ref>f=\<varphi\>+<wide|f|~><space|5spc>(<wide|f|~>\<prec\><wide|\<frak-v\>|~>)>

  transforms (<reference|ade>) into

  <equation|<label|ade-ref>P<rsub|+\<varphi\>>(<wide|f|~>)=0<space|5spc>(<wide|f|~>\<prec\><wide|\<frak-v\>|~>)>

  <theorem|Let <with|mode|math|\<tau\>> be the dominant term of
  <with|mode|math|\<varphi\>>. Then the Newton degree of <with|font
  shape|right|(<reference|ade-ref>)> equals the multiplicity of
  <with|mode|math|\<tau\>> as a potential dominant term for <with|font
  shape|right|(<reference|ade>)>.>

  <new_page>

  <subsection|Quasi-linear equations>

  <with|color|blue|Quasi-linear>: of Newton degree 1.

  <theorem|Any quasi-linear equation admits a ``distinguished solution''.>

  <subsection|Unravellings>

  Newton degree <with|mode|math|d> does not decrease for refinement

  <expand|equation*|f=\<tau\>+<wide|f|~><space|5spc>(<wide|f|~>\<prec\><wide|\<frak-v\>|~>)>

  <with|mode|math|\<longrightarrow\>> consider refinements like

  <expand|equation*|f=\<psi\>+<wide|f|~><space|5spc>(<wide|f|~>\<prec\><wide|\<frak-v\>|~>),>

  with

  <expand|equation*|<frac|\<partial\>P|\<partial\>f>
  (\<psi\>)=0<space|5spc>(\<psi\>\<prec\>\<frak-v\>).>

  <with|mode|math|\<longrightarrow\>> unravelling: refinement
  (<reference|ref>) with

  <\description>
    <expand|item*|U1.>The Newton degree of (<reference|ade-ref>) is
    <with|mode|math|d>.

    <expand|item*|U2.>For any <with|mode|math|<wide|\<varphi\>|~>\<prec\><wide|\<frak-v\>|~>>,
    the Newton degree of

    <expand|equation*|P<rsub|\<varphi\>+<wide|\<varphi\>|~>>(<wide|<wide|f|~>|~>)=0<space|5spc>(<wide|<wide|f|~>|~>\<prec\>\<frak-d\>(<wide|\<varphi\>|~>))>

    is <with|mode|math|\<less\>d>.
  </description>

  <new_page>

  <subsection|Structure of solutions>

  <theorem|If the coefficients of <with|mode|math|P> are all purely
  exponential, then there exists an integer <with|mode|math|U<rsub|d,r,w>>,
  such that the logarithmic depth of each root of <with|mode|math|P> is
  bounded by <with|mode|math|U<rsub|d,r,w>>.>

  <subsection|Example>

  <expand|equation*|P=f+f*f<rprime|''>-(f<rprime|'>)<rsup|2>.>

  Algebraic potential dominant term:

  <expand|equation*|\<tau\>=<with|formula style|false|<frac|1|2>>*x<rsup|2>.>

  Differential potential dominant terms:

  <expand|equation*|\<mu\>*e<rsup|\<lambda\>*x>>

  with <with|mode|math|\<lambda\>\<gtr\>0>.

  <subsection|Exercise>

  Further terms of expansion...<new_page>

  <with|paragraph mode|center|<section|Part III>>

  <subsection|Intermediate value theorem>

  <theorem|Let <with|mode|math|P> be a differential polynomial with
  coefficients in <with|mode|math|\<bbb-T\>>. If
  <with|mode|math|\<varphi\>\<less\>\<psi\>> are such that
  <with|mode|math|P(\<varphi\>)*P(\<psi\>)\<less\>0>, then there exists an
  <with|mode|math|f\<in\>(\<varphi\>,\<psi\>)>, with
  <with|mode|math|P(f)=0>.>

  Example:

  <expand|equation*|P=f<rsup|7>+e<rsup|e<rsup|x>>*f<rsup|<space|0.2spc>3>*f<rprime|'''>+\<Gamma\>(log
  \<Gamma\>(x)+1)=0.>

  <subsection|Proof strategy>

  Generalize the theorem to include <with|mode|math|\<varphi\>> and
  <with|mode|math|\<psi\>> like

  <\itemize>
    <item><with|mode|math|\<varphi\>=\<xi\>\<pm\>\<backepsilon\><rsub|\<bbb-T\>>>.

    <item><with|mode|math|\<varphi\>=\<xi\>\<pm\>\<mho\><rsub|\<bbb-T\>>>.

    <item><with|mode|math|\<varphi\>=\<xi\>\<pm\>\<backepsilon\>*\<frak-m\>>

    <item><with|mode|math|\<varphi\>=\<xi\>\<pm\>\<mho\>*\<frak-m\>.>

    <item><with|mode|math|\<varphi\>=\<xi\>\<pm\>\<gamma\>>, where
    <with|mode|math|\<gamma\>=<frac|1|x*log x*log log x*\<cdots\>>>.
  </itemize>

  <new_page>

  <subsection|Behaviour near zero and infinity>

  <lemma|<with|mode|math|P(<inactive|<group|\<pm\>>>f)> has constant sign for
  sufficiently large <with|mode|math|f\<in\>\<bbb-T\>>.>

  <proof|Rewrite <with|mode|math|P> in <with|mode|math|f,f<rprime|\<dag\>>,f<rprime|\<dag\>\<dag\>>,\<ldots\>>>

  <lemma|<with|mode|math|P(<inactive|<group|\<pm\>>>\<varepsilon\>)> has
  constant sign for sufficiently small <with|mode|math|f\<in\>\<bbb-T\>>.>

  <subsection|Behaviour near constants>

  <\itemize>
    <item><with|mode|math|\<sigma\><rsub|P>(f)=sign P(f)>.

    <item><with|mode|math|f\<precprec\>g\<Leftrightarrow\>log
    \|f\|\<prec\>log \|g\|>.

    <item><with|mode|math|N<rsub|P>=Q(c)*(c<rprime|'>)<rsup|\<nu\>>> and
    <with|mode|math|P> purely exponential.

    <item><with|mode|math|\<mu\>> multiplicity of <with|mode|math|c> as a
    root of <with|mode|math|Q>.
  </itemize>

  <\lemma>
    For all <with|mode|math|0\<less\>\<varepsilon\>\<prec\>1> with
    <with|mode|math|\<varepsilon\>\<precprec\>e<rsup|x>>, the signs of
    <with|mode|math|P(c-\<varepsilon\>)> and
    <with|mode|math|P(c+\<varepsilon\>)> are independent of
    <with|mode|math|\<varepsilon\>> and given by

    <expand|equation*|(<inactive|<group|-1>>)<rsup|\<mu\>>*\<sigma\><rsub|P>(c-\<backepsilon\>)=(<inactive|<group|-1>>)<rsup|\<nu\>>*\<sigma\><rsub|P>(c+\<backepsilon\>)=\<sigma\><rsub|Q<rsup|(\<mu\>)>>(c).>
  </lemma>

  <\lemma>
    For <with|mode|math|P> homogeneous of degree <with|mode|math|i>:

    <expand|equation*|\<sigma\><rsub|P>(\<backepsilon\>)=\<sigma\><rsub|P>(\<varepsilon\>)=\<sigma\><rsub|R<rsub|P,i>(\<varepsilon\><rprime|\<dag\>>)>=\<sigma\><rsub|R<rsub|P,i>(-\<gamma\>)>>

    for all <with|mode|math|0\<less\>\<varepsilon\>\<prec\>1> with
    <with|mode|math|\<varepsilon\>\<precprec\>e<rsup|x>>.<new_page>
  </lemma>

  <subsection|Behaviour before & after constants>

  <\lemma>
    For all <with|mode|math|0\<less\>f\<succ\>1> with
    <with|mode|math|f\<precprec\>e<rsup|x>>, the signs of
    <with|mode|math|P(<inactive|<group|-f>>)> and <with|mode|math|P(f)> are
    independent of <with|mode|math|f> and given by

    <expand|equation*|(<inactive|<group|-1>>)<rsup|deg
    Q+\<nu\>>*\<sigma\><rsub|P>(-\<mho\>)=\<sigma\><rsub|P>(\<mho\>)=sign
    Q<rsub|deg Q>.>
  </lemma>

  <\lemma>
    For <with|mode|math|P> homogeneous of degree <with|mode|math|i>:

    <expand|equation*|\<sigma\><rsub|P>(\<mho\>)=\<sigma\><rsub|P>(f)=\<sigma\><rsub|R<rsub|P,i>>(f<rprime|\<dag\>>)=\<sigma\><rsub|R<rsub|P,i>(\<gamma\>)>,>

    for all <with|mode|math|0\<less\>f\<succ\>1> with
    <with|mode|math|f\<precprec\>e<rsup|x>>.
  </lemma>

  <subsection|Final proof>

  <\itemize>
    <item>Reduction to intervals of the form
    <with|mode|math|(\<xi\>,\<xi\>+\<backepsilon\>*\<frak-v\>)> or
    <with|mode|math|(\<xi\>+\<gamma\>,\<xi\>+\<backepsilon\>*\<frak-v\>)>.

    <item>Triple induction over order of <with|mode|math|P>, Newton degree of
    (<reference|ade>) and maximal length of chain of ``privileged
    refinements''.

    <item>Reduce interval using lemmas and by looking where the sign change
    occurs.
  </itemize>

  <new_page>

  <with|paragraph mode|center|<section|What next?>>

  <\itemize>
    <item>Complex transseries

    <\expand|itemize-minus>
      <item>At least <with|mode|math|d> complex transseries solutions.

      <item>Linear differential operators factorize.

      <item>Case separations & corresponding regions.

      <item>Analysable functions.
    </expand>

    <item>Differential-difference equations

    <\expand|itemize-minus>
      <item>OK for postcompositions with exponentiality <with|mode|math|0>
      (like <with|mode|math|z+1,q*z,z<rsup|N>)>.

      <item>Transseries of positive strength (PhD. Michael Schmeling).

      <item>Nested transseries.
    </expand>

    <item>Model theory.

    <item>Algorithms.

    <item>And much and much more...
  </itemize>

  \;
</body>

<\initial>
  <\collection>
    <associate|odd page margin|30mm>
    <associate|paragraph width|150mm>
    <associate|page medium|paper>
    <associate|page right margin|30mm>
    <associate|page top margin|30mm>
    <associate|reduction page right margin|25mm>
    <associate|reduction page bottom margin|15mm>
    <associate|even page margin|30mm>
    <associate|reduction page left margin|25mm>
    <associate|page bottom margin|30mm>
    <associate|reduction page top margin|15mm>
    <associate|show header and footer|true>
  </collection>
</initial>

<\references>
  <\collection>
    <associate|toc-30|<tuple|7|14>>
    <associate|toc-10|<tuple|<uninit>|6>>
    <associate|toc-20|<tuple|1|10>>
    <associate|toc-31|<tuple|7|14>>
    <associate|toc-11|<tuple|<uninit>|6>>
    <associate|toc-21|<tuple|1|10>>
    <associate|toc-32|<tuple|8|14>>
    <associate|toc-12|<tuple|1|7>>
    <associate|ade|<tuple|1|10>>
    <associate|toc-22|<tuple|1|10>>
    <associate|toc-33|<tuple|8|15>>
    <associate|toc-13|<tuple|1|7>>
    <associate|ref|<tuple|2|11>>
    <associate|toc-23|<tuple|3|11>>
    <associate|toc-34|<tuple|10|15>>
    <associate|toc-14|<tuple|1|7>>
    <associate|toc-24|<tuple|4|11>>
    <associate|toc-35|<tuple|12|16>>
    <associate|toc-15|<tuple|1|7>>
    <associate|toc-25|<tuple|5|12>>
    <associate|toc-36|<tuple|14|16>>
    <associate|toc-16|<tuple|1|7>>
    <associate|toc-26|<tuple|6|12>>
    <associate|toc-37|<tuple|14|17>>
    <associate|toc-17|<tuple|1|8>>
    <associate|toc-27|<tuple|6|13>>
    <associate|toc-18|<tuple|1|9>>
    <associate|toc-28|<tuple|7|13>>
    <associate|toc-19|<tuple|2|10>>
    <associate|toc-29|<tuple|7|13>>
    <associate|toc-1|<tuple|<uninit>|2>>
    <associate|toc-2|<tuple|<uninit>|2>>
    <associate|toc-3|<tuple|<uninit>|2>>
    <associate|ade-ref|<tuple|3|11>>
    <associate|toc-4|<tuple|<uninit>|3>>
    <associate|toc-5|<tuple|<uninit>|4>>
    <associate|toc-6|<tuple|<uninit>|4>>
    <associate|toc-7|<tuple|<uninit>|5>>
    <associate|toc-8|<tuple|<uninit>|5>>
    <associate|toc-9|<tuple|<uninit>|5>>
  </collection>
</references>

<\auxiliary>
  <\collection>
    <\associate|toc>
      PART I<value|toc-dots><pageref|toc-1>

      Transseries<value|toc-dots><pageref|toc-2>

      Origins<value|toc-dots><pageref|toc-3>

      Examples<value|toc-dots><pageref|toc-4>

      Well-ordered power series<value|toc-dots><pageref|toc-5>

      Grid-based series<value|toc-dots><pageref|toc-6>

      Construction of <with|mode|<quote|math>|\<bbb-T\>><value|toc-dots><pageref|toc-7>

      <with|left margin|<quote|1.5fn>|Logarithmic
      transseries<value|toc-dots><pageref|toc-8>>

      <with|left margin|<quote|1.5fn>|Inductive
      step<value|toc-dots><pageref|toc-9>>

      Transbases<value|toc-dots><pageref|toc-10>

      <with|left margin|<quote|1.5fn>|Incomplete transbasis
      theorem<value|toc-dots><pageref|toc-11>>

      Differentiation<value|toc-dots><pageref|toc-12>

      <with|left margin|<quote|1.5fn>|Differentiation on
      <with|mode|<quote|math>|\<bbb-L\>><value|toc-dots><pageref|toc-13>>

      <with|left margin|<quote|1.5fn>|Differentiation on
      <with|mode|<quote|math>|\<bbb-T\>><value|toc-dots><pageref|toc-14>>

      Upward & downward shifting<value|toc-dots><pageref|toc-15>

      Part II<value|toc-dots><pageref|toc-16>

      <with|left margin|<quote|1.5fn>|Algebraic differential
      polynomials<value|toc-dots><pageref|toc-17>>

      Dominant parts<value|toc-dots><pageref|toc-18>

      Differential Newton polygons<value|toc-dots><pageref|toc-19>

      <with|left margin|<quote|1.5fn>|Newton
      degree<value|toc-dots><pageref|toc-20>>

      <with|left margin|<quote|1.5fn>|Potential dominant monomial
      types<value|toc-dots><pageref|toc-21>>

      <with|left margin|<quote|1.5fn>|Algebraic and mixed
      p.d.m.s<value|toc-dots><pageref|toc-22>>

      <with|left margin|<quote|1.5fn>|Differential
      p.d.m.s<value|toc-dots><pageref|toc-23>>

      <with|left margin|<quote|1.5fn>|Refinements<value|toc-dots><pageref|toc-24>>

      <with|left margin|<quote|1.5fn>|Quasi-linear
      equations<value|toc-dots><pageref|toc-25>>

      <with|left margin|<quote|1.5fn>|Unravellings<value|toc-dots><pageref|toc-26>>

      <with|left margin|<quote|1.5fn>|Structure of
      solutions<value|toc-dots><pageref|toc-27>>

      <with|left margin|<quote|1.5fn>|Example<value|toc-dots><pageref|toc-28>>

      <with|left margin|<quote|1.5fn>|Exercise<value|toc-dots><pageref|toc-29>>

      Part III<value|toc-dots><pageref|toc-30>

      <with|left margin|<quote|1.5fn>|Intermediate value
      theorem<value|toc-dots><pageref|toc-31>>

      <with|left margin|<quote|1.5fn>|Proof
      strategy<value|toc-dots><pageref|toc-32>>

      <with|left margin|<quote|1.5fn>|Behaviour near zero and
      infinity<value|toc-dots><pageref|toc-33>>

      <with|left margin|<quote|1.5fn>|Behaviour near
      constants<value|toc-dots><pageref|toc-34>>

      <with|left margin|<quote|1.5fn>|Behaviour before & after
      constants<value|toc-dots><pageref|toc-35>>

      <with|left margin|<quote|1.5fn>|Final
      proof<value|toc-dots><pageref|toc-36>>

      What next?<value|toc-dots><pageref|toc-37>
    </associate>
  </collection>
</auxiliary>

