Carl Love

## 24830 Reputation

10 years, 115 days
Natick, Massachusetts, United States
My name was formerly Carl Devore.

## Use < p > for handy Vector-Matrix conver...

Use < p > for handy Vector-to-Matrix conversion, and use ^%T for transpose.

KroneckerProduct(< p >, < p >^%T);

## Upper bounds of loops don't match...

In the Matlab, the upper bound of the inner loop is the index of the outer loop; whereas in the Maple, the two loops have the same upper bound. The closest matching Maple code to your Matlab is

for j from 1 to N do
for i from irem(j-1,2) to j by 2 do
Dhat[i+1,j+1]:= 2*j
end do
end do;

You can also use (j-1) mod 2; I prefer irem in this case.

## Need to deal with the constants of integ...

Before you can plot it, you need to deal with the six constants of integration _C1, ..., _C6. Can you find six initial or boundary conditions?

## Use finite-field arithmetic...

You should post your matrix or the worksheet that contains it. If the entries are rational functions, then we can compute the answer numerically over different finite fields and reconstruct the final answer via Chinese remaindering. A lot of this functionality is already built into Maple if you use the correct options for the Determinant command. And I'll work on that when you post your matrix.

In the worst case, there could be 14! ~ 87 billion terms in the determinant; so you may need to be satisfied with the value of the determinant at specific numeric values of the variables.

## Extra constant "ga"...

In addition to Preben's Answer: There's an extra constant, ga, in the fourth equation, for which no value is provided. If you provide a value for this, then it only expects 9 boundary conditions.

## A module that does it...

Sorry this took a few days. It was a bit more complicated than I anticipated. The main complexity was that Graphs need to have unqiue labels for each vertex, but, of course, we may want to repeat labels in the tree.

I welcome any critiques of my module. Especially, I'd like to know if there are any modern aspects modules that I could have used.

(* Written by Carl Love, 16-Mar-2013.

A module to convert a function tree representation (such as is returned by ToInert
or various commands in XMLtools) into a tree represented in the GraphTheory package
*)
FunctionToTree:= module()
local
(*____________________________________________________________________________________
Phase 1: Build a representation of the tree in tables with each vertex represented by
by an integer and each edge by an ordered pair of integers.
``````````````````````````````````````````````````````````````````````````````````````*)
# Vertices is a table that maps the original functional representation of the
# node to an integer.  VertexLabels is simply the inverse of that table.
Vertices::table, VertexLabels::table, Vertex_index::nonnegint,

# Edges is a set (stored as a table indexed by an integer for efficiency)
# of ordered pairs of vertices in their integer representation.
Edges::table, Edge_index::nonnegint,

Edge_index:= Edge_index + 1;
Edges[Edge_index]:= [Vertices[e], Vertices[e]]
end proc,

Vertex_index:= Vertex_index + 1;
Vertices[x]:= Vertex_index;
VertexLabels[Vertex_index]:= x
end proc,

# Recursive
local x::function;
for x in f do
#Make functions unique.
if assigned(Vertices[x]) then  x:= convert(op(0,x), `local`)(op(x))  fi;
#Leaves are type function but not function(function).
if x::function(function) then thisproc(x) end if;
end do
end proc,
(*__________________________________________________________________________________
Phase 2: Shorten the function labels to something that can be shown on a plot of the
tree but which is still meaningful.
`````````````````````````````````````````````````````````````````````````````````````*)
# Prefix is the number of chars at the beginning of each function name that
# can be trimmed off.  For example, if every function name begins _Inert_, then
# Prefix should be set to 7.  This value is passed in by the user.
Prefix::nonnegint,

# Labels is a table mapping the string form of the shortened function names to
# alternative (usually abbreviated) representations.  For example, "SUM" can be
# mapped to `+`.  This is passed in by the user.
Labels::table,

#Shorten the function names
StripName:= proc(f::function)
local f0:= sprintf("%a", op(0,f))[Prefix+1 .. -1];
`if`(assigned(Labels[f0]), Labels[f0], nprintf("%s", f0))
end proc,

ShortenLabels:= proc()
local k::nonnegint, f::function, f0::symbol;
VertexMap:= table();
for k to Vertex_index do
f:= VertexLabels[k];
f0:= StripName(f);
if not f::function(function) then
#For leaves, display the function operands also.
f0:= nprintf("%Q", `if`(f0 = ``, op(f), f0(op(f))))
end if;
if assigned(VertexMap[f0]) then f0:= convert(f0, `local`) fi;
VertexMap[f0]:= f;
VertexLabels[k]:= f0
end do
end proc,
(*____________________________________________________________________________________
Main
``````````````````````````````````````````````````````````````````````````````````````*)
ModuleInits:= proc()
Edges:= table();
Edge_index:= 0;
Vertices:= table();
VertexLabels:= table();
Vertex_index:= 0
end proc,

ModuleApply:= proc(func::function(function), {Labels::table:= table(), Prefix::nonnegint:= 0})
ModuleInits();
thismodule:-Prefix:= Prefix;
thismodule:-Labels:= Labels;

ShortenLabels();  #Phase 2
#Subs the short labels into the integer labels, then build and return the graph
GraphTheory:-Graph(subs(op(eval(VertexLabels)), convert(Edges, set)))
end proc
;
# VertexMap is a table mapping the short labels to their original functions
export VertexMap::table;
end module;

A module to convert a function tree representation (such as is returned by ToInert or by various commands in XMLtools) into a tree represented in the GraphTheory package.

 >
(* Written by Carl Love, 16-Mar-2013. > Func:= ToInert(x+2*y*z^2); I came up with a scheme to abbreviate the vertex labels. In the line below, you need to map the second part of the _Inert_ function names to display labels. Note that the display label can be the empty name ``. Also note that this table must map strings (things in double quotes) to names (things in single backquotes).

 > Abbrs:= table(["SUM"= `+`, "PROD"= `*`, "INTPOS"= ``, "NAME"= ``, "POWER"= `^`]):

The Prefix= 7 argument tells it to strip off the first 7 characters of each function name.

 > G:= FunctionToTree(Func, Labels= Abbrs, Prefix= 7); For any given tree, it's hard to say which command of DrawNetwork of DrawGraph will display it better. Neither is great.

 > GraphTheory:-DrawNetwork(G); > GraphTheory:-DrawGraph(G); The module exports a table VertexMap that shows the correspondence between the vertex labels and the original functional representation.

 > eval(FunctionToTree:-VertexMap); >
 > Func:= ToInert([1,2,3,4]); > G2:= FunctionToTree(Func, Labels= Abbrs, Prefix= 7); > GraphTheory:-DrawNetwork(G2); >

## You can see code for Maple's rifsimp...

You can see the code for Maple's rifsimp:

showstat(`DEtools/Rif/rifsimp`);

## Statistics:-Distribution...

Once you do the first problem, you know that c = 3. That allows you to set the Support option of the Statistics:-Distribution command.

 > with(Statistics):
 > MyDist:= Distribution(PDF= (x-> x^2/9), Support= 0..3):
 > [Mean, Variance, StandardDeviation](MyDist); You should also learn the more-traditional ways:

 > f:= x-> x^2/9:
 > mean = Int(x*'f'(x), x= 0..3); > value(%); > mu:= eval(mean,%); > variance = Int((x-'mu')^2*'f'(x), x= 0..3); > value(%); > stdev:= sqrt(eval(variance,%)); >

## Need c such that integral = 1...

The links to your uploaded worksheets are not working. But the idea of the problem is that the integral of f(x) from 0 to c must equal 1.

eqn:= int(x^2/9, x= 0..c) = 1;
solve({eqn, c>0}, c);

See ?intsolve.

## indets...

One problem with using index numbers with op is that you don't know that the integral will be the, say, 3rd factor in each term. Indeed, the ordering of the factors within each term might change on different runs. If you just want to extract all the integrals from lhs(eq10_1), then do

indets(lhs(eq10_1), specfunc(anything, Int));

If you want to extract the integrals and maintain the order that they have in the equation (and you know that there is more than one term and at most one integral per term), then do

map(op@indets, [op](lhs(eq10_1)), specfunc(anything, Int));

## pixelation?...

I believe that the effect that you're seeing is just due to the pixel size of your screen, and has nothing to do with the number of points. Curves that are close to horizontal or close to vertical will appear jagged. We have to live with it until they start making screens at 300 dots per inch, or more. Try printing the plot on paper and seeing if the effect is still present.

## In three short lines...

In three short lines:

 > R1:= %cos ~ (< 1|1|1 >*omega*t + 2*Pi/3*<0|-1|1>):
 > T:= sqrt(2/3)*value(< R1, eval(R1, %cos= -%sin), < 1|1|1 >/sqrt(2) >):
 > (simplify@expand) ~ (T.map(diff, T, t)^%T); >