Skip to first unread message
Emmanuel Charpentier
Jun 17, 2013, 3:29:52 PM6/17/13
to sage-s...@googlegroups.com
Dear list,
I am trying to render a directed graph with LaTeX and tikz. I'ts a Bayesian network based on some causal reasoning (hence the necessity for a directed graph), but I'll reduce my problem to a simple test case.
The following code :
var("a", latex_name="\mathrm{a}")
var("b", latex_name="\mathrm{b}")
var("c", latex_name="\mathrm{c}")
var("d", latex_name="\mathrm{d}")
var("e", latex_name="\mathrm{e}")
H=DiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$'}, c:{e:'$\\beta_{c,e}$'}})
H.set_pos(H.layout_acyclic())
H.set_latex_options(tkz_style="Custom", graphic_size=(4,4), vertex_size=0.5, vertex_labels=True, edge_labels=True, edge_label_sloped=False)
view(H)
gives me almost what I want (drag-and-drop from a notebook] :
The graph has the desired layout, the markup (vertices and edges) is correct. But, for reasons I haven't be able to fathom, edges are rendered as circle arcs (about pi/6). It is not especially good in my test case, and makes my full graph positively awful...
Trying to use another style of graph makes me lose [a) edge markup and/or b) vertex markup], and c) arrow heads. Looking in the documentation about arrow heads gave no clue. Trying to specify latex parameters for edges and vertices has no point, since these setups are forgotten by the new style (and yes, I tried to use two set_latex_options instructions : same results in both orders...).
Any hint ?
Emmanuel Charpentier
I am trying to render a directed graph with LaTeX and tikz. I'ts a Bayesian network based on some causal reasoning (hence the necessity for a directed graph), but I'll reduce my problem to a simple test case.
The following code :
var("a", latex_name="\mathrm{a}")
var("b", latex_name="\mathrm{b}")
var("c", latex_name="\mathrm{c}")
var("d", latex_name="\mathrm{d}")
var("e", latex_name="\mathrm{e}")
H=DiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$'}, c:{e:'$\\beta_{c,e}$'}})
H.set_pos(H.layout_acyclic())
H.set_latex_options(tkz_style="Custom", graphic_size=(4,4), vertex_size=0.5, vertex_labels=True, edge_labels=True, edge_label_sloped=False)
view(H)
gives me almost what I want (drag-and-drop from a notebook] :
The graph has the desired layout, the markup (vertices and edges) is correct. But, for reasons I haven't be able to fathom, edges are rendered as circle arcs (about pi/6). It is not especially good in my test case, and makes my full graph positively awful...
Trying to use another style of graph makes me lose [a) edge markup and/or b) vertex markup], and c) arrow heads. Looking in the documentation about arrow heads gave no clue. Trying to specify latex parameters for edges and vertices has no point, since these setups are forgotten by the new style (and yes, I tried to use two set_latex_options instructions : same results in both orders...).
Any hint ?
Emmanuel Charpentier
Dima Pasechnik
Jun 17, 2013, 4:03:32 PM6/17/13
to sage-s...@googlegroups.com
On 2013-06-17, Emmanuel Charpentier <emanuel.c...@gmail.com> wrote:
> ------=_Part_260_31328394.1371497392901
> Content-Type: text/plain; charset=ISO-8859-1
>
> Dear list,
>
> I am trying to render a *directed* graph with LaTeX and tikz. I'ts a
so, try a forum frequented by tikz wizards,
e.g. http://tex.stackexchange.com/
Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX file along
the way. Maybe then it would be more clear how to tweak the drawing.
HTH,
Dmitrii
>
> Emmanuel Charpentier
>
> ------=_Part_260_31328394.1371497392901
> Content-Type: text/plain; charset=ISO-8859-1
>
> Dear list,
>
> I am trying to render a *directed* graph with LaTeX and tikz. I'ts a
> Bayesian network based on some causal reasoning (hence the necessity for a
> directed graph), but I'll reduce my problem to a simple test case.
>
> The following code :
>
> var("a", latex_name="\mathrm{a}")
> var("b", latex_name="\mathrm{b}")
> var("c", latex_name="\mathrm{c}")
> var("d", latex_name="\mathrm{d}")
> var("e", latex_name="\mathrm{e}")
> H=DiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$'},
> c:{e:'$\\beta_{c,e}$'}})
> H.set_pos(H.layout_acyclic())
> H.set_latex_options(tkz_style="Custom", graphic_size=(4,4),
> vertex_size=0.5, vertex_labels=True, edge_labels=True,
> edge_label_sloped=False)
> view(H)
>
> gives me *almost* what I want (drag-and-drop from a notebook] :
> directed graph), but I'll reduce my problem to a simple test case.
>
> The following code :
>
> var("a", latex_name="\mathrm{a}")
> var("b", latex_name="\mathrm{b}")
> var("c", latex_name="\mathrm{c}")
> var("d", latex_name="\mathrm{d}")
> var("e", latex_name="\mathrm{e}")
> H=DiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$'},
> c:{e:'$\\beta_{c,e}$'}})
> H.set_pos(H.layout_acyclic())
> H.set_latex_options(tkz_style="Custom", graphic_size=(4,4),
> vertex_size=0.5, vertex_labels=True, edge_labels=True,
> edge_label_sloped=False)
> view(H)
>
>
>
> The graph has the desired layout, the markup (vertices and edges) is
> correct. But, for reasons I haven't be able to fathom, edges are rendered
> as circle arcs (about pi/6). It is not especially good in my test case, and
> makes my full graph positively awful...
>
> Trying to use another style of graph makes me lose [a) edge markup and/or
> b) vertex markup], and c) arrow heads. Looking in the documentation about
> arrow heads gave no clue. Trying to specify latex parameters for edges and
> vertices has no point, since these setups are forgotten by the new style
> (and yes, I tried to use two set_latex_options instructions : same results
> in both orders...).
>
> Any hint ?
Your subject says that you narrowed down the problem to LaTeX/tikz. If
>
> The graph has the desired layout, the markup (vertices and edges) is
> correct. But, for reasons I haven't be able to fathom, edges are rendered
> as circle arcs (about pi/6). It is not especially good in my test case, and
> makes my full graph positively awful...
>
> Trying to use another style of graph makes me lose [a) edge markup and/or
> b) vertex markup], and c) arrow heads. Looking in the documentation about
> arrow heads gave no clue. Trying to specify latex parameters for edges and
> vertices has no point, since these setups are forgotten by the new style
> (and yes, I tried to use two set_latex_options instructions : same results
> in both orders...).
>
> Any hint ?
so, try a forum frequented by tikz wizards,
e.g. http://tex.stackexchange.com/
Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX file along
the way. Maybe then it would be more clear how to tweak the drawing.
HTH,
Dmitrii
>
> Emmanuel Charpentier
>
Emmanuel Charpentier
Jun 19, 2013, 4:44:23 PM6/19/13
to sage-s...@googlegroups.com
Dear Dmitrii, dear list,
Following Dmitrii hints, I tried latex(H). It turns out that the curved edges are a feature, not a bug (a better look at the GraphLatex page showed me explicit documentation about this). This design choice seems embedded in the source code for latex generation of directed graphs, and there exist (currently) no option for changing it (no sage-settable parameter for edge_style and arrowheads). Too bad the result is ugly as sin...
I could hack a new style for DiGraphs. I could also hack the code and make these parameters tunable. Both of these options are lengthy, and my RealLife(TM) constraints make them a bit difficult. It seems faster to either a) hack convenient parameters to ....DiGraph.plot and generate a PDF or b) directly generate tikz code from the graph specifications.
This bit of ad-hockery is probably the only way I can cope with my current problem. Bit I think a better solution would be to rewrite (part of) the GraphLatex specification to give them more degree of freedom.
Looking at the Trac tickets for this subject hnted that such a task is probably not easy. So I won't, for now, create new ticket. I'll have to look first at the source code for GraphLatex (wich I haven't yet located, shame on me...).
Emmanuel Charpentier
BTW : tex.stackexchange.com gave me few hints about this problem : people here tend to hack their solutions directly in \LaTeX, or even in plain \TeX (God forbid...).
Dima Pasechnik
Jun 20, 2013, 4:37:33 AM6/20/13
to sage-s...@googlegroups.com
On 2013-06-19, Emmanuel Charpentier <emanuel.c...@gmail.com> wrote:
> ------=_Part_3454_23024622.1371674663187
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable
>
> Dear Dmitrii, dear list,
>
> Following Dmitrii hints, I tried latex(H). It turns out that the curved=20
> edges are a feature, not a bug (a better look at the GraphLatex page showed=
>=20
> me explicit documentation about this). This design choice seems embedded in=
>=20
> the source code for latex generation of directed graphs, and there exist=20
> (currently) no option for changing it (no sage-settable parameter for=20
>=20
> these parameters tunable. Both of these options are lengthy, and my=20
> RealLife(TM) constraints make them a bit difficult. It seems faster to=20
> either a) hack convenient parameters to ....DiGraph.plot and generate a PDF=
>=20
>=20
> problem. Bit I think a better solution would be to rewrite (part of) the=20
> probably not easy. So I won't, for now, create new ticket. I'll have to=20
> look first at the source code for GraphLatex (wich I haven't yet located,=
>=20
> shame on me...).
it is in sage/graphs/graph_latex.py
>
> Emmanuel=
>=20
> Charpentier
>
> BTW : tex.stackexchange.com gave me few hints about this problem : people=
>=20
> here tend to hack their solutions directly in \LaTeX, or even in plain \TeX=
>=20
> (God forbid...).
>
> Le lundi 17 juin 2013 22:03:32 UTC+2, Dima Pasechnik a =E9crit :
>>
>> On 2013-06-17, Emmanuel Charpentier <emanuel.c...@gmail.com <javascript:>=
>>=20
>> wrote:=20
>> > ------=3D_Part_260_31328394.1371497392901=20
>> > Content-Type: text/plain; charset=3DISO-8859-1=20
>> >=20
>> > Dear list,=20
>> >=20
>> > I am trying to render a *directed* graph with LaTeX and tikz. I'ts a=20
>> > Bayesian network based on some causal reasoning (hence the necessity fo=
> r=20
>> a=20
>> > directed graph), but I'll reduce my problem to a simple test case.=20
>> >=20
>> > The following code :=20
>> >=20
>> > var("a", latex_name=3D"\mathrm{a}")=20
>> > var("b", latex_name=3D"\mathrm{b}")=20
>> > var("c", latex_name=3D"\mathrm{c}")=20
>> > var("d", latex_name=3D"\mathrm{d}")=20
>> > var("e", latex_name=3D"\mathrm{e}")=20
>> > H=3DDiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$=
> '},=20
>> > c:{e:'$\\beta_{c,e}$'}})=20
>> > H.set_pos(H.layout_acyclic())=20
>> > H.set_latex_options(tkz_style=3D"Custom", graphic_size=3D(4,4),=20
>> > vertex_size=3D0.5, vertex_labels=3DTrue, edge_labels=3DTrue,=20
>> > edge_label_sloped=3DFalse)=20
>> > view(H)=20
>> >=20
>> > gives me *almost* what I want (drag-and-drop from a notebook] :=20
>> >=20
>> >=20
>> > The graph has the desired layout, the markup (vertices and edges) is=20
>> > correct. But, for reasons I haven't be able to fathom, edges are=20
>> rendered=20
>> > as circle arcs (about pi/6). It is not especially good in my test case,=
>=20
>> and=20
>> > makes my full graph positively awful...=20
>> >=20
>> > Trying to use another style of graph makes me lose [a) edge markup=20
>> and/or=20
>> > b) vertex markup], and c) arrow heads. Looking in the documentation=20
>> about=20
>> > arrow heads gave no clue. Trying to specify latex parameters for edges=
>=20
>> and=20
>> > vertices has no point, since these setups are forgotten by the new styl=
> e=20
>> > (and yes, I tried to use two set_latex_options instructions : same=20
>> results=20
>> > in both orders...).=20
>> >=20
>> > Any hint ?=20
>> Your subject says that you narrowed down the problem to LaTeX/tikz. If=20
>> so, try a forum frequented by tikz wizards,=20
>> e.g. http://tex.stackexchange.com/=20
>>
>> Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX file=
>=20
>> along=20
>> the way. Maybe then it would be more clear how to tweak the drawing.=20
>>
>> HTH,=20
>> Dmitrii=20
>>
>>
>>
>> >=20
>> > Emmanuel Charpentier=20
>> >=20
>>
>>
>
> --=20
> You received this message because you are subscribed to the Google Groups "=
> sage-support" group.
> To unsubscribe from this group and stop receiving emails from it, send an e=
> mail to sage-support...@googlegroups.com.
> To post to this group, send email to sage-s...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-support.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>
> ------=_Part_3454_23024622.1371674663187
> Content-Type: text/html; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable
>
> Dear Dmitrii, dear list,<div><br></div><div>Following Dmitrii hints, I trie=
> d latex(H). It turns out that the curved edges are a feature, not a bug (a =
> better look at the GraphLatex page showed me explicit documentation about t=
> his). This design choice seems embedded in the source code for latex genera=
> tion of directed graphs, and there exist (currently) no option for changing=
> it (no sage-settable parameter for edge_style and arrowheads). Too bad the=
> result is ugly as sin...</div><div><br></div><div>I could hack a new style=
> for DiGraphs. I could also hack the code and make these parameters tunable=
> . Both of these options are lengthy, and my RealLife(TM) constraints make t=
> hem a bit difficult. It seems faster to either a) hack convenient parameter=
> s to ....DiGraph.plot and generate a PDF or b) directly generate tikz code =
> from the graph specifications.</div><div><br></div><div>This bit of ad-hock=
> ery is probably the only way I can cope with my current problem. Bit I thin=
> k a better solution would be to rewrite (part of) the GraphLatex specificat=
> ion to give them more degree of freedom.</div><div><br></div><div>Looking a=
> t the Trac tickets for this subject hnted that such a task is probably not =
> easy. So I won't, for now, create new ticket. I'll have to look first=
> at the source code for GraphLatex (wich I haven't yet located, shame on me=
> ...).</div><div><br></div><div> &n=
> bsp; =
>  =
> ; Emmanuel Charpentier</div><div><=
> br></div><div>BTW : tex.stackexchange.com gave me few hints about this prob=
> lem : people here tend to hack their solutions directly in \LaTeX, or even =
> in plain \TeX (God forbid...).<br><br>Le lundi 17 juin 2013 22:03:32 UTC+2,=
> Dima Pasechnik a =E9crit :<blockquote class=3D"gmail_quote" style=3D"=
> margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;=
> ">On 2013-06-17, Emmanuel Charpentier <<a href=3D"javascript:" target=3D=
> "_blank" gdf-obfuscated-mailto=3D"gdW3cqa1SREJ">emanuel.c...@gmail.com</a><=
> wbr>> wrote:
><br>> ------=3D_Part_260_31328394.<wbr>1371497392901
><br>> Content-Type: text/plain; charset=3DISO-8859-1
><br>>
><br>> Dear list,
><br>>
><br>> I am trying to render a *directed* graph with LaTeX and tikz. I'ts=
> a=20
><br>> Bayesian network based on some causal reasoning (hence the necessi=
> ty for a=20
><br>> directed graph), but I'll reduce my problem to a simple test case.
><br>>
><br>> The following code :
><br>>
><br>> var("a", latex_name=3D"\mathrm{a}")
><br>> var("b", latex_name=3D"\mathrm{b}")
><br>> var("c", latex_name=3D"\mathrm{c}")
><br>> var("d", latex_name=3D"\mathrm{d}")
><br>> var("e", latex_name=3D"\mathrm{e}")
><br>> H=3DDiGraph({a:{b:'$\\beta_{a,b}<wbr>$',c:'$\\beta_{a,c}$',d:'$\\<=
> wbr>beta_{a,d}$'},=20
><br>> c:{e:'$\\beta_{c,e}$'}})
><br>> H.set_pos(H.layout_acyclic())
><br>> H.set_latex_options(tkz_style=3D<wbr>"Custom", graphic_size=3D(4,4=
> ),=20
><br>> vertex_size=3D0.5, vertex_labels=3DTrue, edge_labels=3DTrue,=20
><br>> edge_label_sloped=3DFalse)
><br>> view(H)
><br>>
><br>> gives me *almost* what I want (drag-and-drop from a notebook] :
><br>>
><br>>
><br>> The graph has the desired layout, the markup (vertices and edges) =
> is=20
><br>> correct. But, for reasons I haven't be able to fathom, edges=
> are rendered=20
><br>> as circle arcs (about pi/6). It is not especially good in my test =
> case, and=20
><br>> makes my full graph positively awful...
><br>>
><br>> Trying to use another style of graph makes me lose [a) edge markup=
> and/or=20
><br>> b) vertex markup], and c) arrow heads. Looking in the documentatio=
> n about=20
><br>> arrow heads gave no clue. Trying to specify latex parameters for e=
> dges and=20
><br>> vertices has no point, since these setups are forgotten by the new=
> style=20
><br>> (and yes, I tried to use two set_latex_options instructions : same=
> results=20
><br>> in both orders...).
><br>>
><br>> Any hint ?
><br>Your subject says that you narrowed down the problem to LaTeX/tikz. If
><br>so, try a forum frequented by tikz wizards,=20
><br>e.g. <a href=3D"http://tex.stackexchange.com/" target=3D"_blank">http:/=
> /tex.stackexchange.com/</a>
><br>
><br>Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX fil=
> e along
><br>the way. Maybe then it would be more clear how to tweak the drawing.
><br>
><br>HTH,
><br>Dmitrii
><br>
><br>
><br>
><br>>
><br>> &nb=
> sp; &=
> nbsp; Emmanuel Charpentier
><br>>
><br>
><br></blockquote></div>
>
><p></p>
>
> -- <br />
> You received this message because you are subscribed to the Google Groups &=
> quot;sage-support" group.<br />
> To unsubscribe from this group and stop receiving emails from it, send an e=
> mail to sage-support...@googlegroups.com.<br />
> To post to this group, send email to sage-s...@googlegroups.com.<br />
> Visit this group at <a href=3D"http://groups.google.com/group/sage-support"=
>>http://groups.google.com/group/sage-support</a>.<br />
> For more options, visit <a href=3D"https://groups.google.com/groups/opt_out=
> ">https://groups.google.com/groups/opt_out</a>.<br />
> <br />
> <br />
>
> ------=_Part_3454_23024622.1371674663187--
>
> ------=_Part_3454_23024622.1371674663187
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable
>
> Dear Dmitrii, dear list,
>
> Following Dmitrii hints, I tried latex(H). It turns out that the curved=20
> edges are a feature, not a bug (a better look at the GraphLatex page showed=
>=20
> me explicit documentation about this). This design choice seems embedded in=
>=20
> the source code for latex generation of directed graphs, and there exist=20
> (currently) no option for changing it (no sage-settable parameter for=20
> edge_style and arrowheads). Too bad the result is ugly as sin...
>
> I could hack a new style for DiGraphs. I could also hack the code and make=
>
>=20
> these parameters tunable. Both of these options are lengthy, and my=20
> RealLife(TM) constraints make them a bit difficult. It seems faster to=20
> either a) hack convenient parameters to ....DiGraph.plot and generate a PDF=
>=20
> or b) directly generate tikz code from the graph specifications.
>
> This bit of ad-hockery is probably the only way I can cope with my current=
>
>=20
> problem. Bit I think a better solution would be to rewrite (part of) the=20
> GraphLatex specification to give them more degree of freedom.
>
> Looking at the Trac tickets for this subject hnted that such a task is=20
>
> probably not easy. So I won't, for now, create new ticket. I'll have to=20
> look first at the source code for GraphLatex (wich I haven't yet located,=
>=20
> shame on me...).
it is in sage/graphs/graph_latex.py
>
> Emmanuel=
>=20
> Charpentier
>
> BTW : tex.stackexchange.com gave me few hints about this problem : people=
>=20
> here tend to hack their solutions directly in \LaTeX, or even in plain \TeX=
>=20
> (God forbid...).
>
> Le lundi 17 juin 2013 22:03:32 UTC+2, Dima Pasechnik a =E9crit :
>>
>> On 2013-06-17, Emmanuel Charpentier <emanuel.c...@gmail.com <javascript:>=
>>=20
>> wrote:=20
>> > ------=3D_Part_260_31328394.1371497392901=20
>> > Content-Type: text/plain; charset=3DISO-8859-1=20
>> >=20
>> > Dear list,=20
>> >=20
>> > I am trying to render a *directed* graph with LaTeX and tikz. I'ts a=20
>> > Bayesian network based on some causal reasoning (hence the necessity fo=
> r=20
>> a=20
>> > directed graph), but I'll reduce my problem to a simple test case.=20
>> >=20
>> > The following code :=20
>> >=20
>> > var("a", latex_name=3D"\mathrm{a}")=20
>> > var("b", latex_name=3D"\mathrm{b}")=20
>> > var("c", latex_name=3D"\mathrm{c}")=20
>> > var("d", latex_name=3D"\mathrm{d}")=20
>> > var("e", latex_name=3D"\mathrm{e}")=20
>> > H=3DDiGraph({a:{b:'$\\beta_{a,b}$',c:'$\\beta_{a,c}$',d:'$\\beta_{a,d}$=
> '},=20
>> > c:{e:'$\\beta_{c,e}$'}})=20
>> > H.set_pos(H.layout_acyclic())=20
>> > H.set_latex_options(tkz_style=3D"Custom", graphic_size=3D(4,4),=20
>> > vertex_size=3D0.5, vertex_labels=3DTrue, edge_labels=3DTrue,=20
>> > edge_label_sloped=3DFalse)=20
>> > view(H)=20
>> >=20
>> > gives me *almost* what I want (drag-and-drop from a notebook] :=20
>> >=20
>> >=20
>> > The graph has the desired layout, the markup (vertices and edges) is=20
>> > correct. But, for reasons I haven't be able to fathom, edges are=20
>> rendered=20
>> > as circle arcs (about pi/6). It is not especially good in my test case,=
>=20
>> and=20
>> > makes my full graph positively awful...=20
>> >=20
>> > Trying to use another style of graph makes me lose [a) edge markup=20
>> and/or=20
>> > b) vertex markup], and c) arrow heads. Looking in the documentation=20
>> about=20
>> > arrow heads gave no clue. Trying to specify latex parameters for edges=
>=20
>> and=20
>> > vertices has no point, since these setups are forgotten by the new styl=
> e=20
>> > (and yes, I tried to use two set_latex_options instructions : same=20
>> results=20
>> > in both orders...).=20
>> >=20
>> > Any hint ?=20
>> Your subject says that you narrowed down the problem to LaTeX/tikz. If=20
>> so, try a forum frequented by tikz wizards,=20
>> e.g. http://tex.stackexchange.com/=20
>>
>> Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX file=
>=20
>> along=20
>> the way. Maybe then it would be more clear how to tweak the drawing.=20
>>
>> HTH,=20
>> Dmitrii=20
>>
>>
>>
>> >=20
>> > Emmanuel Charpentier=20
>> >=20
>>
>>
>
> --=20
> You received this message because you are subscribed to the Google Groups "=
> sage-support" group.
> To unsubscribe from this group and stop receiving emails from it, send an e=
> mail to sage-support...@googlegroups.com.
> To post to this group, send email to sage-s...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-support.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>
> ------=_Part_3454_23024622.1371674663187
> Content-Type: text/html; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable
>
> Dear Dmitrii, dear list,<div><br></div><div>Following Dmitrii hints, I trie=
> d latex(H). It turns out that the curved edges are a feature, not a bug (a =
> better look at the GraphLatex page showed me explicit documentation about t=
> his). This design choice seems embedded in the source code for latex genera=
> tion of directed graphs, and there exist (currently) no option for changing=
> it (no sage-settable parameter for edge_style and arrowheads). Too bad the=
> result is ugly as sin...</div><div><br></div><div>I could hack a new style=
> for DiGraphs. I could also hack the code and make these parameters tunable=
> . Both of these options are lengthy, and my RealLife(TM) constraints make t=
> hem a bit difficult. It seems faster to either a) hack convenient parameter=
> s to ....DiGraph.plot and generate a PDF or b) directly generate tikz code =
> from the graph specifications.</div><div><br></div><div>This bit of ad-hock=
> ery is probably the only way I can cope with my current problem. Bit I thin=
> k a better solution would be to rewrite (part of) the GraphLatex specificat=
> ion to give them more degree of freedom.</div><div><br></div><div>Looking a=
> t the Trac tickets for this subject hnted that such a task is probably not =
> easy. So I won't, for now, create new ticket. I'll have to look first=
> at the source code for GraphLatex (wich I haven't yet located, shame on me=
> ...).</div><div><br></div><div> &n=
> bsp; =
>  =
> ; Emmanuel Charpentier</div><div><=
> br></div><div>BTW : tex.stackexchange.com gave me few hints about this prob=
> lem : people here tend to hack their solutions directly in \LaTeX, or even =
> in plain \TeX (God forbid...).<br><br>Le lundi 17 juin 2013 22:03:32 UTC+2,=
> Dima Pasechnik a =E9crit :<blockquote class=3D"gmail_quote" style=3D"=
> margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;=
> ">On 2013-06-17, Emmanuel Charpentier <<a href=3D"javascript:" target=3D=
> "_blank" gdf-obfuscated-mailto=3D"gdW3cqa1SREJ">emanuel.c...@gmail.com</a><=
> wbr>> wrote:
><br>> ------=3D_Part_260_31328394.<wbr>1371497392901
><br>> Content-Type: text/plain; charset=3DISO-8859-1
><br>>
><br>> Dear list,
><br>>
><br>> I am trying to render a *directed* graph with LaTeX and tikz. I'ts=
> a=20
><br>> Bayesian network based on some causal reasoning (hence the necessi=
> ty for a=20
><br>> directed graph), but I'll reduce my problem to a simple test case.
><br>>
><br>> The following code :
><br>>
><br>> var("a", latex_name=3D"\mathrm{a}")
><br>> var("b", latex_name=3D"\mathrm{b}")
><br>> var("c", latex_name=3D"\mathrm{c}")
><br>> var("d", latex_name=3D"\mathrm{d}")
><br>> var("e", latex_name=3D"\mathrm{e}")
><br>> H=3DDiGraph({a:{b:'$\\beta_{a,b}<wbr>$',c:'$\\beta_{a,c}$',d:'$\\<=
> wbr>beta_{a,d}$'},=20
><br>> c:{e:'$\\beta_{c,e}$'}})
><br>> H.set_pos(H.layout_acyclic())
><br>> H.set_latex_options(tkz_style=3D<wbr>"Custom", graphic_size=3D(4,4=
> ),=20
><br>> vertex_size=3D0.5, vertex_labels=3DTrue, edge_labels=3DTrue,=20
><br>> edge_label_sloped=3DFalse)
><br>> view(H)
><br>>
><br>> gives me *almost* what I want (drag-and-drop from a notebook] :
><br>>
><br>>
><br>> The graph has the desired layout, the markup (vertices and edges) =
> is=20
><br>> correct. But, for reasons I haven't be able to fathom, edges=
> are rendered=20
><br>> as circle arcs (about pi/6). It is not especially good in my test =
> case, and=20
><br>> makes my full graph positively awful...
><br>>
><br>> Trying to use another style of graph makes me lose [a) edge markup=
> and/or=20
><br>> b) vertex markup], and c) arrow heads. Looking in the documentatio=
> n about=20
><br>> arrow heads gave no clue. Trying to specify latex parameters for e=
> dges and=20
><br>> vertices has no point, since these setups are forgotten by the new=
> style=20
><br>> (and yes, I tried to use two set_latex_options instructions : same=
> results=20
><br>> in both orders...).
><br>>
><br>> Any hint ?
><br>Your subject says that you narrowed down the problem to LaTeX/tikz. If
><br>so, try a forum frequented by tikz wizards,=20
><br>e.g. <a href=3D"http://tex.stackexchange.com/" target=3D"_blank">http:/=
> /tex.stackexchange.com/</a>
><br>
><br>Anyhow, I'd try to use sagetex to draw your graph, generating LaTeX fil=
> e along
><br>the way. Maybe then it would be more clear how to tweak the drawing.
><br>
><br>HTH,
><br>Dmitrii
><br>
><br>
><br>
><br>>
><br>> &nb=
> sp; &=
> nbsp; Emmanuel Charpentier
><br>>
><br>
><br></blockquote></div>
>
><p></p>
>
> -- <br />
> You received this message because you are subscribed to the Google Groups &=
> quot;sage-support" group.<br />
> To unsubscribe from this group and stop receiving emails from it, send an e=
> mail to sage-support...@googlegroups.com.<br />
> To post to this group, send email to sage-s...@googlegroups.com.<br />
> Visit this group at <a href=3D"http://groups.google.com/group/sage-support"=
>>http://groups.google.com/group/sage-support</a>.<br />
> For more options, visit <a href=3D"https://groups.google.com/groups/opt_out=
> ">https://groups.google.com/groups/opt_out</a>.<br />
> <br />
> <br />
>
> ------=_Part_3454_23024622.1371674663187--
>
Reply all
Reply to author
Forward
0 new messages