mmcdara

847 Reputation

12 Badges

3 years, 183 days

MaplePrimes Activity


These are questions asked by mmcdara

Hi, 

This thread is more or less related to a previous one about the Statistics:-Sample procedure.
(see https://www.mapleprimes.com/questions/228421-A-Serious-Problem-With-StatisticsSample )
I've just implemented two variant of the Box-Muller procedure to sample normal rvs.
The source is "The art of computer programming", Donald E. Knuth, 2nd edition, p117 (aka "algorithm P").

The first implementation (BoxMuller_1) is basically what D.E. Knuth writes, except that I "vectorize" some operations in order to avoid using an if.. then..else structure (as a minor consequence I generate a little bit more numbers than required).
This procedure uses the build-in Maple's procedure select (please see the link above and acer's and carl love's replies).
It appears to be relatively slow.
More of this CodeTools:-Usage(BoxMuller_1(10^6)) generates a "conenction to kernel lost" for some unknown reason.

The variant named BoxMuller_2 uses sort and ListTools:-BinaryPlace instead of select.
Ir appears to be around 3 times faster than BoxMuller_1, but remains 10 time slower than Statistics:-Sample(..., method=envelope) (here again, see the link above to understand why method=envelope is needed).

I wonder how it could be possible to speed up this procedure. In particular acer showed in one of his reply (again see the link above) how using hfloats can improve the efficiency of a procedure, but I'm very incompetent on this point.
Does anyone have any idea?

Thanks in advance.

PS: this file has been written with Maple 2015.2


 

restart:

BoxMuller_1 := proc(N)
  local V, S, T, L, X1, X2:
  V  := Statistics:-Sample(Uniform(-1, 1), [ceil(2*N/3), 2]):
  S  := V[..,1]^~2 +~ V[..,2]^~2;
  T  := < subs(NULL=infinity, select(`<`, S, 1)) | V >;
  T[.., 1]  := (-2 *~ log~(T[.., 1]) /~ T[.., 1])^~(0.5);
  X1 := select[flatten](type, T[.., 1] *~  T[..,2], 'float');
  X2 := select[flatten](type, T[.., 1] *~  T[..,3], 'float');
  return <X1 , X2>:
end proc:


BoxMuller_2 := proc(N)
  local V1, V2, S, W, u, r, T, X1, X2:
  V1 := Statistics:-Sample(Uniform(-1, 1), ceil(2*N/3)):
  V2 := Statistics:-Sample(Uniform(-1, 1), ceil(2*N/3)):
  S  := V1^~2 +~ V2^~2;
  W  := sort(S, output = [sorted, permutation]):
  u  := ListTools:-BinaryPlace(W[1], 1);
  r  := [$1..u]:
  T  := (-2 *~ log~(W[1][r]) /~ W[1][r])^~(0.5);
  X1 := T *~ V1[W[2][r]];
  X2 := T *~ V2[W[2][r]];
  return <X1 | X2>^+:
end proc:  

CodeTools:-Usage(Statistics:-Sample(Uniform(-1, 1), 10^5)):
CodeTools:-Usage(Statistics:-Sample(Uniform(-1, 1), 10^5, method=envelope)):
# CodeTools:-Usage(BoxMuller_1(10^6)): # generates a "connection to kernel lost" error msg
CodeTools:-Usage(BoxMuller_2(10^6)):

print():
CodeTools:-Usage(BoxMuller_1(10^5)):
CodeTools:-Usage(BoxMuller_2(10^5)):
print():

S1 := BoxMuller_1(10^5):
S2 := BoxMuller_2(10^5):

memory used=1.17MiB, alloc change=0 bytes, cpu time=12.00ms, real time=12.00ms, gc time=0ns
memory used=3.32MiB, alloc change=32.00MiB, cpu time=64.00ms, real time=64.00ms, gc time=0ns

memory used=148.17MiB, alloc change=119.93MiB, cpu time=706.00ms, real time=637.00ms, gc time=14.63ms

 

 

memory used=67.43MiB, alloc change=186.80MiB, cpu time=946.00ms, real time=762.00ms, gc time=285.14ms
memory used=14.81MiB, alloc change=0 bytes, cpu time=67.00ms, real time=62.00ms, gc time=0ns

 

(1)

if false then
DocumentTools:-Tabulate(
  [
    plots:-display(
      Statistics:-Histogram(S1),
      plot(Statistics:-PDF(Normal(0, 1), x), x=-4..4, color=red,thickness=3)
    ),
    plots:-display(
      Statistics:-Histogram(S2),
      plot(Statistics:-PDF(Normal(0, 1), x), x=-4..4, color=red,thickness=3)
    )
  ], width=60
)
end if:
  

# carl love's procedure with slight modifications t

SampleCheck := proc(X, f, N::posint)
uses St= Statistics;
  local S, M, O, E:
  S:= f(N):
  M:= numelems(S):
  O:= <St:-TallyInto(S, <[$(floor@min-1..ceil@max)(S)]>)>:
  E:= St:-Probability~(X <~ (rhs@lhs)~(O), 'numeric') * M:
  <rhs~(O)[2..] | E[2..] - E[..-2]>
end proc:
 

interface(rtablesize=20):
SampleCheck(Statistics:-RandomVariable(Normal(0,1)), 'BoxMuller_1', 10^5),
SampleCheck(Statistics:-RandomVariable(Normal(0,1)), 'BoxMuller_2', 10^5);

Matrix(11, 2, {(1, 1) = 1, (1, 2) = HFloat(0.029943974977392658), (2, 1) = 5, (2, 2) = HFloat(3.28979552036377), (3, 1) = 121, (3, 2) = HFloat(138.1791685600983), (4, 1) = 2276, (4, 2) = HFloat(2243.2153196005097), (5, 1) = 14488, (5, 2) = HFloat(14245.846696531142), (6, 1) = 35630, (6, 2) = HFloat(35780.43897239682), (7, 1) = 35886, (7, 2) = HFloat(35780.43897239682), (8, 1) = 14071, (8, 2) = HFloat(14245.846696531138), (9, 1) = 2212, (9, 2) = HFloat(2243.2153196005092), (10, 1) = 130, (10, 2) = HFloat(138.1791685601056), (11, 1) = 2, (11, 2) = HFloat(3.2897955203516176)}), Matrix(10, 2, {(1, 1) = 1, (1, 2) = HFloat(3.293310594473029), (2, 1) = 127, (2, 2) = HFloat(138.32680996055558), (3, 1) = 2241, (3, 2) = HFloat(2245.6121457991635), (4, 1) = 14226, (4, 2) = HFloat(14261.068070193269), (5, 1) = 35632, (5, 2) = HFloat(35818.66958395649), (6, 1) = 35864, (6, 2) = HFloat(35818.66958395649), (7, 1) = 14451, (7, 2) = HFloat(14261.068070193272), (8, 1) = 2266, (8, 2) = HFloat(2245.6121457991685), (9, 1) = 125, (9, 2) = HFloat(138.32680996054842), (10, 1) = 1, (10, 2) = HFloat(3.293310594468494)})

(2)

 


 

Download Box-Muller.mw

Hi, 
The choice method=default (the implicit choice) gives dramatically false results as demonstrated here for the sampling of a normal r.v.
It's likely to be the case for any continuous distribution.
In a few words the tails of the distribution are not correctly sampled (far from the mean by 4 standard deviations for a normal r.v., not an extremely rare event in practical applications).

I think the attached file is sufficiently documented for you to understand the problem.


 

restart:


The problem presented here also exists with Maple 2018 and Maple 2019

interface(version)

`Standard Worksheet Interface, Maple 2015.2, Mac OS X, December 21 2015 Build ID 1097895`

(1)

with(Statistics):

X := RandomVariable(Normal(0, 1)):


Generate a sample of X of size 10^6

Here is the R code for those who would like to verify the results and the performances

 

N <- 10^6

R <- 20

Q <- c(0:5)

 

M <- matrix(nrow=R, ncol=length(Q))

 

for (r in c(1:20)) {

  S <- rnorm(N, mean=0, sd=1)

  for (q in Q) {

    M[r, q+1] <-length(S[S>q])

  }

}

 

Expected.Number.of.Outliers <- floor(N - pnorm(Q, mean=0, sd=1) * N)

Observed.Number.of.Outliers <-  round(colMeans(M))

 

Expected.Number.of.Outliers

Observed.Number.of.Outliers

 

Absolute.Differences <- M - kronecker(t(Expected.Number.of.Outliers), rep(1, R))

boxplot(differences)

Remark the loop below takes about 30 seconds to run (to be compared to the 20 minutes
it would take am I used the build-in procedure Select, and to  the 3 seconds R demands).

N := 10^6:
R := 20:

Q  := [$0..5]:
nQ := numelems(Q):
M  := Matrix(R, nQ, 0):

for r from 1 to R do
  S := Sample(X, N):
  for q in Q do
    M[r, q+1] := add(Heaviside~(S -~ q)):
  end do:
end do:

Expected_Number_Of_Outliers := Vector[row](nQ, i -> Probability(X > Q[i], numeric)*N);

Expected_Number_Of_Outliers := Vector[row](6, {(1) = HFloat(500000.0), (2) = HFloat(158655.25393145697), (3) = HFloat(22750.13194817921), (4) = HFloat(1349.8980316301036), (5) = HFloat(31.67124183311998), (6) = HFloat(0.2866515719235352)})

(2)

Observed_Number_Of_Outliers := Mean(M);

Observed_Number_Of_Outliers := Vector[row](6, {(1) = 500291.4, (2) = 158782.3, (3) = 22761.35, (4) = 1349.5, (5) = 51.6, (6) = 1.9}, datatype = float[8])

(3)


While the values of  the Observed_Number_of_Outliers seem to be reasonably correct for q =0, 1, 2 and 3
there are highly suspicious for q = 4 and likely for q = 5 (but N is too small to be categorical)

So let's examine more closely rhe results for q = 4

`q=4`:= convert(M[..,5], list);
 min(`q=4`)

[HFloat(49.0), HFloat(63.0), HFloat(42.0), HFloat(47.0), HFloat(44.0), HFloat(68.0), HFloat(46.0), HFloat(51.0), HFloat(58.0), HFloat(55.0), HFloat(50.0), HFloat(51.0), HFloat(68.0), HFloat(45.0), HFloat(58.0), HFloat(44.0), HFloat(59.0), HFloat(49.0), HFloat(47.0), HFloat(38.0)]

 

HFloat(38.0)

(4)


This means we obtained 20 estimations out of 20 of the probability that X exceeds q=4.
There are only about on chance out of 1 million to get such a result (roughly (1/2)^20 ).

Here is the result R returned for the same quantile q=4

 M[,5]

 [1] 24 37 23 38 30 43 23 30 25 39 25 29 29 36 39 41 35 37 36 34

(9 values out of 20 are less than the expected value of 31.67)

A very simple test based on the Binomial distribution will prove, without any doubt, that the
distribution of the 20 numbers of outliers for q = 4 implies the sampling algorithm is of
is of very poor quality.



Is it possible to obtain correct results with Maple ?


(simulations done below for the case q=4 only)

N := 10^6:
R := 20:

Menv := Vector[column](R, 0):

for r from 1 to R do
  S := Sample(X, N, method=envelope):
  Menv[r] := add(Heaviside~(S -~ 4)):
end do:

`q=4`:= convert(Menv, list);
Mean(Menv);
Variance(Menv);

[HFloat(26.0), HFloat(26.0), HFloat(35.0), HFloat(34.0), HFloat(30.0), HFloat(30.0), HFloat(28.0), HFloat(24.0), HFloat(27.0), HFloat(31.0), HFloat(28.0), HFloat(21.0), HFloat(34.0), HFloat(33.0), HFloat(29.0), HFloat(38.0), HFloat(38.0), HFloat(33.0), HFloat(39.0), HFloat(25.0)]

 

HFloat(30.45)

 

HFloat(24.892105263157895)

(5)



Conclusion, the "default" sampling strategy suffers strong flaw.

Remark: it's well known that one of the best method to sample normal random variables
is Box-Mueller's: why it has not been programmed by default ?

 

 


 

Download Bug_Normal_Sampler.mw


 

Hi,

Sorry to ask such a stupid question but I can't find out where my error is. Probably it's so huge it blinds me!

The double loop and the matrix product F^+ . F should give the same result, no? (it seems that F^+ . F has its rows reordered ?)


 

restart:

N   := 3:
P   := 2:
niv := [seq(Z[i], i=1..N)];
f   := Matrix(N^P, P, (i,j) -> `if`(j=P, niv[(i mod 3)+1], niv[iquo(i-1,3)+1]));

niv := [Z[1], Z[2], Z[3]]

 

f := Matrix(9, 2, {(1, 1) = Z[1], (1, 2) = Z[2], (2, 1) = Z[1], (2, 2) = Z[3], (3, 1) = Z[1], (3, 2) = Z[1], (4, 1) = Z[2], (4, 2) = Z[2], (5, 1) = Z[2], (5, 2) = Z[3], (6, 1) = Z[2], (6, 2) = Z[1], (7, 1) = Z[3], (7, 2) = Z[2], (8, 1) = Z[3], (8, 2) = Z[3], (9, 1) = Z[3], (9, 2) = Z[1]})

(1)

ds := subs(niv =~ [$0..N-1], f);

ds := Matrix(9, 2, {(1, 1) = 0, (1, 2) = 1, (2, 1) = 0, (2, 2) = 2, (3, 1) = 0, (3, 2) = 0, (4, 1) = 1, (4, 2) = 1, (5, 1) = 1, (5, 2) = 2, (6, 1) = 1, (6, 2) = 0, (7, 1) = 2, (7, 2) = 1, (8, 1) = 2, (8, 2) = 2, (9, 1) = 2, (9, 2) = 0})

(2)

vs := [ seq(V__||i, i=1..P)]:
es := unapply( sort( [ seq( mul(vs ^~ [entries(ds[i,..], nolist)]), i=1..N^P) ] ), vs);
 

proc (V__1, V__2) options operator, arrow; [1, V__1, V__2, V__1^2, V__2^2, V__1*V__2, V__1*V__2^2, V__1^2*V__2, V__1^2*V__2^2] end proc

(3)

ff := convert([ seq(es(entries(ffd[i,..], nolist)), i=1..N^P) ], Matrix);


UnityRoots := [solve(z^3=1, z)]:
F := simplify(subs(niv =~ UnityRoots, ff)) /~ sqrt(N^P):

ff := Matrix(9, 9, {(1, 1) = 1, (1, 2) = Z[1], (1, 3) = Z[2], (1, 4) = Z[1]^2, (1, 5) = Z[2]^2, (1, 6) = Z[1]*Z[2], (1, 7) = Z[1]*Z[2]^2, (1, 8) = Z[1]^2*Z[2], (1, 9) = Z[1]^2*Z[2]^2, (2, 1) = 1, (2, 2) = Z[1], (2, 3) = Z[3], (2, 4) = Z[1]^2, (2, 5) = Z[3]^2, (2, 6) = Z[1]*Z[3], (2, 7) = Z[1]*Z[3]^2, (2, 8) = Z[1]^2*Z[3], (2, 9) = Z[1]^2*Z[3]^2, (3, 1) = 1, (3, 2) = Z[1], (3, 3) = Z[1], (3, 4) = Z[1]^2, (3, 5) = Z[1]^2, (3, 6) = Z[1]^2, (3, 7) = Z[1]^3, (3, 8) = Z[1]^3, (3, 9) = Z[1]^4, (4, 1) = 1, (4, 2) = Z[2], (4, 3) = Z[2], (4, 4) = Z[2]^2, (4, 5) = Z[2]^2, (4, 6) = Z[2]^2, (4, 7) = Z[2]^3, (4, 8) = Z[2]^3, (4, 9) = Z[2]^4, (5, 1) = 1, (5, 2) = Z[2], (5, 3) = Z[3], (5, 4) = Z[2]^2, (5, 5) = Z[3]^2, (5, 6) = Z[2]*Z[3], (5, 7) = Z[2]*Z[3]^2, (5, 8) = Z[2]^2*Z[3], (5, 9) = Z[2]^2*Z[3]^2, (6, 1) = 1, (6, 2) = Z[2], (6, 3) = Z[1], (6, 4) = Z[2]^2, (6, 5) = Z[1]^2, (6, 6) = Z[1]*Z[2], (6, 7) = Z[1]^2*Z[2], (6, 8) = Z[1]*Z[2]^2, (6, 9) = Z[1]^2*Z[2]^2, (7, 1) = 1, (7, 2) = Z[3], (7, 3) = Z[2], (7, 4) = Z[3]^2, (7, 5) = Z[2]^2, (7, 6) = Z[2]*Z[3], (7, 7) = Z[2]^2*Z[3], (7, 8) = Z[2]*Z[3]^2, (7, 9) = Z[2]^2*Z[3]^2, (8, 1) = 1, (8, 2) = Z[3], (8, 3) = Z[3], (8, 4) = Z[3]^2, (8, 5) = Z[3]^2, (8, 6) = Z[3]^2, (8, 7) = Z[3]^3, (8, 8) = Z[3]^3, (8, 9) = Z[3]^4, (9, 1) = 1, (9, 2) = Z[3], (9, 3) = Z[1], (9, 4) = Z[3]^2, (9, 5) = Z[1]^2, (9, 6) = Z[1]*Z[3], (9, 7) = Z[1]^2*Z[3], (9, 8) = Z[1]*Z[3]^2, (9, 9) = Z[1]^2*Z[3]^2})

(4)

Scalar products of pairs of comumn vectors

F must be an orthogonal array

for i1 from 1 to N^P do
  for i2 from 1 to N^P do
    printf("%a ", simplify(add(F[..,i1] . F[.., i2])))
  end do:
  printf("\n"):
end do:
printf("\n");

1 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0
0 0 0 0 1 0 0 0 0
0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
 

 

or more simply:

simplify(F^+ . F)

Matrix([[1, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 1, 0, 0, 0, 0, 0], [0, 0, 0, 0, 1, 0, 0, 0, 0], [0, 1, 0, 0, 0, 0, 0, 0, 0], [0, 0, 1, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 1], [0, 0, 0, 0, 0, 0, 0, 1, 0], [0, 0, 0, 0, 0, 0, 1, 0, 0], [0, 0, 0, 0, 0, 1, 0, 0, 0]])

(5)

 


 

Download Too_Blind_To_Find_My_Mistake.mw

Hi, 

I need to plot some correlation matrices C1, C2, ... and I use matrixplot for this.
I would like to use the same absolute scale (-1..+1) for all of them.
For instance is I decide to uses colorscheme=["blue", "white", "red"] I would like blue to correspond to value -1, white to value 0 and red to value 1.
Unfortunately colorscheme set to blue the cell with the mininum value (not necessarily -1) and to red the maximum one (not necessarily +1).
Here is an example

restart:
with(plots):
with(Statistics):
randomize():
N := 10:
P := 3:
A := Sample(Uniform(0, 1), [N, P]):
C := CorrelationMatrix(A):
matrixplot(
  C,

 heights=histogram,
 axes=frame,
​​​​​​​  gap=0.25,
​​​​​​​  color=((x,y)->(C[x,y]+1)/2),
​​​​​​​  orientation=[0, 0, 0],
​​​​​​​  lightmodel=none,
​​​​​​​  tickmarks=[[seq(i+1/2=A||i, i=1..P)], [seq(i+1/2=A||i, i=1..P)], default],
​​​​​​​  labels=[("")$3]​​​​​​​
​​​​​​​  );


​​​​​​​I also tried to use color=((x,y) -> (C[x, y]+1)/2) instead of colorscheme but here again matrixplot uses a local scale defined by the reange of the correlation matrix to plot.

I fixed this by using something like seq(seq(PLOT(POLYGONS(...), i=1..P), j=1..P) instead of matrixplot, but I think it is a shame to do so.

So my question: is it possible to force matrixplot not to use a scale defined by the matrix to plot, but a "user" scale?

PS: I'm using Maple 2015 


Thanks in advance

Hi, 

I solve numerically an ode for different values of its parameters ( dsolve(..., numeric, parameters=[...] ) and I would like to "stack" the different solutions in a container (the container (list, vector, table) is of no matter).

Here is a notional example where I try to construct a sequence where the first element should be the solution when the parameter is equal to 1 and second one when the parameter is equal to 2.
It happens that some "premature evaluation" seems to occur which makes the two elements identical.

Please do not pay attention to the obvious simplicity of the problem: the true one is more complicated but can be illustrated by the on below.

Thanks in advance
 

restart:

f := dsolve({diff(x(t),t)=A*t, x(0)=0}, numeric, parameters=[A]);

proc (x_rkf45) local _res, _dat, _vars, _solnproc, _xout, _ndsol, _pars, _n, _i; option `Copyright (c) 2000 by Waterloo Maple Inc. All rights reserved.`; if 1 < nargs then error "invalid input: too many arguments" end if; _EnvDSNumericSaveDigits := Digits; Digits := 15; if _EnvInFsolve = true then _xout := evalf[_EnvDSNumericSaveDigits](x_rkf45) else _xout := evalf(x_rkf45) end if; _dat := Array(1..4, {(1) = proc (_xin) local _xout, _dtbl, _dat, _vmap, _x0, _y0, _val, _dig, _n, _ne, _nd, _nv, _pars, _ini, _par, _i, _j, _k, _src; option `Copyright (c) 2002 by Waterloo Maple Inc. All rights reserved.`; table( [( "complex" ) = false ] ) _xout := _xin; _pars := [A = A]; _dtbl := array( 1 .. 4, [( 1 ) = (array( 1 .. 24, [( 1 ) = (datatype = float[8], order = C_order, storage = rectangular), ( 2 ) = (datatype = float[8], order = C_order, storage = rectangular), ( 3 ) = ([0, 0, 0, Array(1..0, 1..2, {}, datatype = float[8], order = C_order)]), ( 4 ) = (Array(1..54, {(1) = 1, (2) = 1, (3) = 0, (4) = 0, (5) = 1, (6) = 0, (7) = 0, (8) = 0, (9) = 0, (10) = 0, (11) = 0, (12) = 0, (13) = 0, (14) = 0, (15) = 0, (16) = 0, (17) = 0, (18) = 0, (19) = 30000, (20) = 0, (21) = 0, (22) = 1, (23) = 4, (24) = 0, (25) = 1, (26) = 15, (27) = 1, (28) = 0, (29) = 1, (30) = 3, (31) = 3, (32) = 0, (33) = 1, (34) = 0, (35) = 0, (36) = 0, (37) = 0, (38) = 0, (39) = 0, (40) = 0, (41) = 0, (42) = 0, (43) = 1, (44) = 0, (45) = 0, (46) = 0, (47) = 0, (48) = 0, (49) = 0, (50) = 50, (51) = 1, (52) = 0, (53) = 0, (54) = 0}, datatype = integer[8])), ( 5 ) = (Array(1..28, {(1) = .0, (2) = 0.10e-5, (3) = .0, (4) = 0.500001e-14, (5) = .0, (6) = .0, (7) = .0, (8) = 0.10e-5, (9) = .0, (10) = .0, (11) = .0, (12) = .0, (13) = 1.0, (14) = .0, (15) = .49999999999999, (16) = .0, (17) = 1.0, (18) = 1.0, (19) = .0, (20) = .0, (21) = 1.0, (22) = 1.0, (23) = .0, (24) = .0, (25) = 0.10e-14, (26) = .0, (27) = .0, (28) = .0}, datatype = float[8], order = C_order)), ( 6 ) = (Array(1..2, {(1) = 0., (2) = Float(undefined)})), ( 7 ) = ([Array(1..4, 1..7, {(1, 1) = .0, (1, 2) = .203125, (1, 3) = .3046875, (1, 4) = .75, (1, 5) = .8125, (1, 6) = .40625, (1, 7) = .8125, (2, 1) = 0.6378173828125e-1, (2, 2) = .0, (2, 3) = .279296875, (2, 4) = .27237892150878906, (2, 5) = -0.9686851501464844e-1, (2, 6) = 0.1956939697265625e-1, (2, 7) = .5381584167480469, (3, 1) = 0.31890869140625e-1, (3, 2) = .0, (3, 3) = -.34375, (3, 4) = -.335235595703125, (3, 5) = .2296142578125, (3, 6) = .41748046875, (3, 7) = 11.480712890625, (4, 1) = 0.9710520505905151e-1, (4, 2) = .0, (4, 3) = .40350341796875, (4, 4) = 0.20297467708587646e-1, (4, 5) = -0.6054282188415527e-2, (4, 6) = -0.4770040512084961e-1, (4, 7) = .77858567237854}, datatype = float[8], order = C_order), Array(1..6, 1..6, {(1, 1) = .0, (1, 2) = .0, (1, 3) = .0, (1, 4) = .0, (1, 5) = .0, (1, 6) = 1.0, (2, 1) = .25, (2, 2) = .0, (2, 3) = .0, (2, 4) = .0, (2, 5) = .0, (2, 6) = 1.0, (3, 1) = .1875, (3, 2) = .5625, (3, 3) = .0, (3, 4) = .0, (3, 5) = .0, (3, 6) = 2.0, (4, 1) = .23583984375, (4, 2) = -.87890625, (4, 3) = .890625, (4, 4) = .0, (4, 5) = .0, (4, 6) = .2681884765625, (5, 1) = .1272735595703125, (5, 2) = -.5009765625, (5, 3) = .44921875, (5, 4) = -0.128936767578125e-1, (5, 5) = .0, (5, 6) = 0.626220703125e-1, (6, 1) = -0.927734375e-1, (6, 2) = .626220703125, (6, 3) = -.4326171875, (6, 4) = .1418304443359375, (6, 5) = -0.861053466796875e-1, (6, 6) = .3131103515625}, datatype = float[8], order = C_order), Array(1..6, {(1) = .0, (2) = .386, (3) = .21, (4) = .63, (5) = 1.0, (6) = 1.0}, datatype = float[8], order = C_order), Array(1..6, {(1) = .25, (2) = -.1043, (3) = .1035, (4) = -0.362e-1, (5) = .0, (6) = .0}, datatype = float[8], order = C_order), Array(1..6, 1..5, {(1, 1) = .0, (1, 2) = .0, (1, 3) = .0, (1, 4) = .0, (1, 5) = .0, (2, 1) = 1.544, (2, 2) = .0, (2, 3) = .0, (2, 4) = .0, (2, 5) = .0, (3, 1) = .9466785280815533, (3, 2) = .25570116989825814, (3, 3) = .0, (3, 4) = .0, (3, 5) = .0, (4, 1) = 3.3148251870684886, (4, 2) = 2.896124015972123, (4, 3) = .9986419139977808, (4, 4) = .0, (4, 5) = .0, (5, 1) = 1.2212245092262748, (5, 2) = 6.019134481287752, (5, 3) = 12.537083329320874, (5, 4) = -.687886036105895, (5, 5) = .0, (6, 1) = 1.2212245092262748, (6, 2) = 6.019134481287752, (6, 3) = 12.537083329320874, (6, 4) = -.687886036105895, (6, 5) = 1.0}, datatype = float[8], order = C_order), Array(1..6, 1..5, {(1, 1) = .0, (1, 2) = .0, (1, 3) = .0, (1, 4) = .0, (1, 5) = .0, (2, 1) = -5.6688, (2, 2) = .0, (2, 3) = .0, (2, 4) = .0, (2, 5) = .0, (3, 1) = -2.4300933568337584, (3, 2) = -.20635991570891224, (3, 3) = .0, (3, 4) = .0, (3, 5) = .0, (4, 1) = -.10735290581452621, (4, 2) = -9.594562251021896, (4, 3) = -20.470286148096154, (4, 4) = .0, (4, 5) = .0, (5, 1) = 7.496443313968615, (5, 2) = -10.246804314641219, (5, 3) = -33.99990352819906, (5, 4) = 11.708908932061595, (5, 5) = .0, (6, 1) = 8.083246795922411, (6, 2) = -7.981132988062785, (6, 3) = -31.52159432874373, (6, 4) = 16.319305431231363, (6, 5) = -6.0588182388340535}, datatype = float[8], order = C_order), Array(1..3, 1..5, {(1, 1) = .0, (1, 2) = .0, (1, 3) = .0, (1, 4) = .0, (1, 5) = .0, (2, 1) = 10.126235083446911, (2, 2) = -7.487995877607633, (2, 3) = -34.800918615557414, (2, 4) = -7.9927717075687275, (2, 5) = 1.0251377232956207, (3, 1) = -.6762803392806898, (3, 2) = 6.087714651678606, (3, 3) = 16.43084320892463, (3, 4) = 24.767225114183653, (3, 5) = -6.5943891257167815}, datatype = float[8], order = C_order)]), ( 9 ) = ([Array(1..1, {(1) = .1}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order), Array(1..1, 1..1, {(1, 1) = .0}, datatype = float[8], order = C_order), Array(1..1, 1..1, {(1, 1) = .0}, datatype = float[8], order = C_order), Array(1..1, 1..6, {(1, 1) = .0, (1, 2) = .0, (1, 3) = .0, (1, 4) = .0, (1, 5) = .0, (1, 6) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = 0}, datatype = integer[8]), Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order)]), ( 8 ) = ([Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..2, {(1) = .0, (2) = .0}, datatype = float[8], order = C_order), Array(1..1, {(1) = .0}, datatype = float[8], order = C_order), 0, 0]), ( 11 ) = (Array(1..6, 0..1, {(1, 1) = .0, (2, 0) = .0, (2, 1) = .0, (3, 0) = .0, (3, 1) = .0, (4, 0) = .0, (4, 1) = .0, (5, 0) = .0, (5, 1) = .0, (6, 0) = .0, (6, 1) = .0}, datatype = float[8], order = C_order)), ( 10 ) = ([proc (N, X, Y, YP) option `[Y[1] = x(t)]`; YP[1] := Y[2]*X; 0 end proc, -1, 0, 0, 0, 0, 0, 0]), ( 13 ) = (), ( 12 ) = (), ( 15 ) = ("rkf45"), ( 14 ) = ([0, 0]), ( 18 ) = ([]), ( 19 ) = (0), ( 16 ) = ([0, 0, 0, []]), ( 17 ) = ([proc (N, X, Y, YP) option `[Y[1] = x(t)]`; YP[1] := Y[2]*X; 0 end proc, -1, 0, 0, 0, 0, 0, 0]), ( 22 ) = (0), ( 23 ) = (0), ( 20 ) = ([]), ( 21 ) = (0), ( 24 ) = (0)  ] ))  ] ); _y0 := Array(0..2, {(1) = 0., (2) = 0.}); _vmap := array( 1 .. 1, [( 1 ) = (1)  ] ); _x0 := _dtbl[1][5][5]; _n := _dtbl[1][4][1]; _ne := _dtbl[1][4][3]; _nd := _dtbl[1][4][4]; _nv := _dtbl[1][4][16]; if not type(_xout, 'numeric') then if member(_xout, ["start", "left", "right"]) then if _Env_smart_dsolve_numeric = true or _dtbl[1][4][10] = 1 then if _xout = "left" then if type(_dtbl[2], 'table') then return _dtbl[2][5][1] end if elif _xout = "right" then if type(_dtbl[3], 'table') then return _dtbl[3][5][1] end if end if end if; return _dtbl[1][5][5] elif _xout = "method" then return _dtbl[1][15] elif _xout = "storage" then return evalb(_dtbl[1][4][10] = 1) elif _xout = "leftdata" then if not type(_dtbl[2], 'array') then return NULL else return eval(_dtbl[2]) end if elif _xout = "rightdata" then if not type(_dtbl[3], 'array') then return NULL else return eval(_dtbl[3]) end if elif _xout = "enginedata" then return eval(_dtbl[1]) elif _xout = "enginereset" then _dtbl[2] := evaln(_dtbl[2]); _dtbl[3] := evaln(_dtbl[3]); return NULL elif _xout = "initial" then return procname(_y0[0]) elif _xout = "laxtol" then return _dtbl[`if`(member(_dtbl[4], {2, 3}), _dtbl[4], 1)][5][18] elif _xout = "numfun" then return `if`(member(_dtbl[4], {2, 3}), _dtbl[_dtbl[4]][4][18], 0) elif _xout = "parameters" then return [seq(_y0[_n+_i], _i = 1 .. nops(_pars))] elif _xout = "initial_and_parameters" then return procname(_y0[0]), [seq(_y0[_n+_i], _i = 1 .. nops(_pars))] elif _xout = "last" then if _dtbl[4] <> 2 and _dtbl[4] <> 3 or _x0-_dtbl[_dtbl[4]][5][1] = 0. then error "no information is available on last computed point" else _xout := _dtbl[_dtbl[4]][5][1] end if elif _xout = "function" then if _dtbl[1][4][33]-2. = 0 then return eval(_dtbl[1][10], 1) else return eval(_dtbl[1][10][1], 1) end if elif _xout = "map" then return copy(_vmap) elif type(_xin, `=`) and type(rhs(_xin), 'list') and member(lhs(_xin), {"initial", "parameters", "initial_and_parameters"}) then _ini, _par := [], []; if lhs(_xin) = "initial" then _ini := rhs(_xin) elif lhs(_xin) = "parameters" then _par := rhs(_xin) elif select(type, rhs(_xin), `=`) <> [] then _par, _ini := selectremove(type, rhs(_xin), `=`) elif nops(rhs(_xin)) < nops(_pars)+1 then error "insufficient data for specification of initial and parameters" else _par := rhs(_xin)[-nops(_pars) .. -1]; _ini := rhs(_xin)[1 .. -nops(_pars)-1] end if; _xout := lhs(_xout); if _par <> [] then `dsolve/numeric/process_parameters`(_n, _pars, _par, _y0) end if; if _ini <> [] then `dsolve/numeric/process_initial`(_n-_ne, _ini, _y0, _pars, _vmap) end if; `dsolve/numeric/SC/reinitialize`(_dtbl, _y0, _n, procname, _pars); if _Env_smart_dsolve_numeric = true and type(_y0[0], 'numeric') and _dtbl[1][4][10] <> 1 then procname("right") := _y0[0]; procname("left") := _y0[0] end if; if _xout = "initial" then return [_y0[0], seq(_y0[_vmap[_i]], _i = 1 .. _n-_ne)] elif _xout = "parameters" then return [seq(_y0[_n+_i], _i = 1 .. nops(_pars))] else return [_y0[0], seq(_y0[_vmap[_i]], _i = 1 .. _n-_ne)], [seq(_y0[_n+_i], _i = 1 .. nops(_pars))] end if elif _xin = "eventstop" then if _nv = 0 then error "this solution has no events" end if; _i := _dtbl[4]; if _i <> 2 and _i <> 3 then return 0 end if; if _dtbl[_i][4][10] = 1 and assigned(_dtbl[5-_i]) and _dtbl[_i][4][9] < 100 and 100 <= _dtbl[5-_i][4][9] then _i := 5-_i; _dtbl[4] := _i; _j := round(_dtbl[_i][4][17]); return round(_dtbl[_i][3][1][_j, 1]) elif 100 <= _dtbl[_i][4][9] then _j := round(_dtbl[_i][4][17]); return round(_dtbl[_i][3][1][_j, 1]) else return 0 end if elif _xin = "eventstatus" then if _nv = 0 then error "this solution has no events" end if; _i := [selectremove(proc (a) options operator, arrow; _dtbl[1][3][1][a, 7] = 1 end proc, {seq(_j, _j = 1 .. round(_dtbl[1][3][1][_nv+1, 1]))})]; return ':-enabled' = _i[1], ':-disabled' = _i[2] elif _xin = "eventclear" then if _nv = 0 then error "this solution has no events" end if; _i := _dtbl[4]; if _i <> 2 and _i <> 3 then error "no events to clear" end if; if _dtbl[_i][4][10] = 1 and assigned(_dtbl[5-_i]) and _dtbl[_i][4][9] < 100 and 100 < _dtbl[5-_i][4][9] then _dtbl[4] := 5-_i; _i := 5-_i end if; if _dtbl[_i][4][9] < 100 then error "no events to clear" elif _nv < _dtbl[_i][4][9]-100 then error "event error condition cannot be cleared" else _j := _dtbl[_i][4][9]-100; if irem(round(_dtbl[_i][3][1][_j, 4]), 2) = 1 then error "retriggerable events cannot be cleared" end if; _j := round(_dtbl[_i][3][1][_j, 1]); for _k to _nv do if _dtbl[_i][3][1][_k, 1] = _j then if _dtbl[_i][3][1][_k, 2] = 3 then error "range events cannot be cleared" end if; _dtbl[_i][3][1][_k, 8] := _dtbl[_i][3][1][_nv+1, 8] end if end do; _dtbl[_i][4][17] := 0; _dtbl[_i][4][9] := 0; if _dtbl[1][4][10] = 1 then if _i = 2 then try procname(procname("left")) catch:  end try else try procname(procname("right")) catch:  end try end if end if end if; return  elif type(_xin, `=`) and member(lhs(_xin), {"eventdisable", "eventenable"}) then if _nv = 0 then error "this solution has no events" end if; if type(rhs(_xin), {('list')('posint'), ('set')('posint')}) then _i := {op(rhs(_xin))} elif type(rhs(_xin), 'posint') then _i := {rhs(_xin)} else error "event identifiers must be integers in the range 1..%1", round(_dtbl[1][3][1][_nv+1, 1]) end if; if select(proc (a) options operator, arrow; _nv < a end proc, _i) <> {} then error "event identifiers must be integers in the range 1..%1", round(_dtbl[1][3][1][_nv+1, 1]) end if; _k := {}; for _j to _nv do if member(round(_dtbl[1][3][1][_j, 1]), _i) then _k := `union`(_k, {_j}) end if end do; _i := _k; if lhs(_xin) = "eventdisable" then _dtbl[4] := 0; _j := [evalb(assigned(_dtbl[2]) and member(_dtbl[2][4][17], _i)), evalb(assigned(_dtbl[3]) and member(_dtbl[3][4][17], _i))]; for _k in _i do _dtbl[1][3][1][_k, 7] := 0; if assigned(_dtbl[2]) then _dtbl[2][3][1][_k, 7] := 0 end if; if assigned(_dtbl[3]) then _dtbl[3][3][1][_k, 7] := 0 end if end do; if _j[1] then for _k to _nv+1 do if _k <= _nv and not type(_dtbl[2][3][4][_k, 1], 'undefined') then userinfo(3, {'events', 'eventreset'}, `reinit #2, event code `, _k, ` to defined init `, _dtbl[2][3][4][_k, 1]); _dtbl[2][3][1][_k, 8] := _dtbl[2][3][4][_k, 1] elif _dtbl[2][3][1][_k, 2] = 0 and irem(iquo(round(_dtbl[2][3][1][_k, 4]), 32), 2) = 1 then userinfo(3, {'events', 'eventreset'}, `reinit #2, event code `, _k, ` to rate hysteresis init `, _dtbl[2][5][24]); _dtbl[2][3][1][_k, 8] := _dtbl[2][5][24] elif _dtbl[2][3][1][_k, 2] = 0 and irem(iquo(round(_dtbl[2][3][1][_k, 4]), 2), 2) = 0 then userinfo(3, {'events', 'eventreset'}, `reinit #2, event code `, _k, ` to initial init `, _x0); _dtbl[2][3][1][_k, 8] := _x0 else userinfo(3, {'events', 'eventreset'}, `reinit #2, event code `, _k, ` to fireinitial init `, _x0-1); _dtbl[2][3][1][_k, 8] := _x0-1 end if end do; _dtbl[2][4][17] := 0; _dtbl[2][4][9] := 0; if _dtbl[1][4][10] = 1 then procname(procname("left")) end if end if; if _j[2] then for _k to _nv+1 do if _k <= _nv and not type(_dtbl[3][3][4][_k, 2], 'undefined') then userinfo(3, {'events', 'eventreset'}, `reinit #3, event code `, _k, ` to defined init `, _dtbl[3][3][4][_k, 2]); _dtbl[3][3][1][_k, 8] := _dtbl[3][3][4][_k, 2] elif _dtbl[3][3][1][_k, 2] = 0 and irem(iquo(round(_dtbl[3][3][1][_k, 4]), 32), 2) = 1 then userinfo(3, {'events', 'eventreset'}, `reinit #3, event code `, _k, ` to rate hysteresis init `, _dtbl[3][5][24]); _dtbl[3][3][1][_k, 8] := _dtbl[3][5][24] elif _dtbl[3][3][1][_k, 2] = 0 and irem(iquo(round(_dtbl[3][3][1][_k, 4]), 2), 2) = 0 then userinfo(3, {'events', 'eventreset'}, `reinit #3, event code `, _k, ` to initial init `, _x0); _dtbl[3][3][1][_k, 8] := _x0 else userinfo(3, {'events', 'eventreset'}, `reinit #3, event code `, _k, ` to fireinitial init `, _x0+1); _dtbl[3][3][1][_k, 8] := _x0+1 end if end do; _dtbl[3][4][17] := 0; _dtbl[3][4][9] := 0; if _dtbl[1][4][10] = 1 then procname(procname("right")) end if end if else for _k in _i do _dtbl[1][3][1][_k, 7] := 1 end do; _dtbl[2] := evaln(_dtbl[2]); _dtbl[3] := evaln(_dtbl[3]); _dtbl[4] := 0; if _dtbl[1][4][10] = 1 then if _x0 <= procname("right") then try procname(procname("right")) catch:  end try end if; if procname("left") <= _x0 then try procname(procname("left")) catch:  end try end if end if end if; return  elif type(_xin, `=`) and lhs(_xin) = "eventfired" then if not type(rhs(_xin), 'list') then error "'eventfired' must be specified as a list" end if; if _nv = 0 then error "this solution has no events" end if; if _dtbl[4] <> 2 and _dtbl[4] <> 3 then error "'direction' must be set prior to calling/setting 'eventfired'" end if; _i := _dtbl[4]; _val := NULL; if not assigned(_EnvEventRetriggerWarned) then _EnvEventRetriggerWarned := false end if; for _k in rhs(_xin) do if type(_k, 'integer') then _src := _k elif type(_k, 'integer' = 'anything') and type(evalf(rhs(_k)), 'numeric') then _k := lhs(_k) = evalf[max(Digits, 18)](rhs(_k)); _src := lhs(_k) else error "'eventfired' entry is not valid: %1", _k end if; if _src < 1 or round(_dtbl[1][3][1][_nv+1, 1]) < _src then error "event identifiers must be integers in the range 1..%1", round(_dtbl[1][3][1][_nv+1, 1]) end if; _src := {seq(`if`(_dtbl[1][3][1][_j, 1]-_src = 0., _j, NULL), _j = 1 .. _nv)}; if nops(_src) <> 1 then error "'eventfired' can only be set/queried for root-finding events and time/interval events" end if; _src := _src[1]; if _dtbl[1][3][1][_src, 2] <> 0. and _dtbl[1][3][1][_src, 2]-2. <> 0. then error "'eventfired' can only be set/queried for root-finding events and time/interval events" elif irem(round(_dtbl[1][3][1][_src, 4]), 2) = 1 then if _EnvEventRetriggerWarned = false then WARNING(`'eventfired' has no effect on events that retrigger`) end if; _EnvEventRetriggerWarned := true end if; if _dtbl[_i][3][1][_src, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_src, 4]), 32), 2) = 1 then _val := _val, undefined elif type(_dtbl[_i][3][4][_src, _i-1], 'undefined') or _i = 2 and _dtbl[2][3][1][_src, 8] < _dtbl[2][3][4][_src, 1] or _i = 3 and _dtbl[3][3][4][_src, 2] < _dtbl[3][3][1][_src, 8] then _val := _val, _dtbl[_i][3][1][_src, 8] else _val := _val, _dtbl[_i][3][4][_src, _i-1] end if; if type(_k, `=`) then if _dtbl[_i][3][1][_src, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_src, 4]), 32), 2) = 1 then error "cannot set event code for a rate hysteresis event" end if; userinfo(3, {'events', 'eventreset'}, `manual set event code `, _src, ` to value `, rhs(_k)); _dtbl[_i][3][1][_src, 8] := rhs(_k); _dtbl[_i][3][4][_src, _i-1] := rhs(_k) end if end do; return [_val] elif type(_xin, `=`) and lhs(_xin) = "direction" then if not member(rhs(_xin), {-1, 1, ':-left', ':-right'}) then error "'direction' must be specified as either '1' or 'right' (positive) or '-1' or 'left' (negative)" end if; _src := `if`(_dtbl[4] = 2, -1, `if`(_dtbl[4] = 3, 1, undefined)); _i := `if`(member(rhs(_xin), {1, ':-right'}), 3, 2); _dtbl[4] := _i; _dtbl[_i] := `dsolve/numeric/SC/IVPdcopy`(_dtbl[1], `if`(assigned(_dtbl[_i]), _dtbl[_i], NULL)); if 0 < _nv then for _j to _nv+1 do if _j <= _nv and not type(_dtbl[_i][3][4][_j, _i-1], 'undefined') then userinfo(3, {'events', 'eventreset'}, `reinit #4, event code `, _j, ` to defined init `, _dtbl[_i][3][4][_j, _i-1]); _dtbl[_i][3][1][_j, 8] := _dtbl[_i][3][4][_j, _i-1] elif _dtbl[_i][3][1][_j, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_j, 4]), 32), 2) = 1 then userinfo(3, {'events', 'eventreset'}, `reinit #4, event code `, _j, ` to rate hysteresis init `, _dtbl[_i][5][24]); _dtbl[_i][3][1][_j, 8] := _dtbl[_i][5][24] elif _dtbl[_i][3][1][_j, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_j, 4]), 2), 2) = 0 then userinfo(3, {'events', 'eventreset'}, `reinit #4, event code `, _j, ` to initial init `, _x0); _dtbl[_i][3][1][_j, 8] := _x0 else userinfo(3, {'events', 'eventreset'}, `reinit #4, event code `, _j, ` to fireinitial init `, _x0-2*_i+5.0); _dtbl[_i][3][1][_j, 8] := _x0-2*_i+5.0 end if end do end if; return _src elif _xin = "eventcount" then if _dtbl[1][3][1] = 0 or _dtbl[4] <> 2 and _dtbl[4] <> 3 then return 0 else return round(_dtbl[_dtbl[4]][3][1][_nv+1, 12]) end if else return "procname" end if end if; if _xout = _x0 then return [_x0, seq(evalf(_dtbl[1][6][_vmap[_i]]), _i = 1 .. _n-_ne)] end if; _i := `if`(_x0 <= _xout, 3, 2); if _xin = "last" and 0 < _dtbl[_i][4][9] and _dtbl[_i][4][9] < 100 then _dat := eval(_dtbl[_i], 2); _j := _dat[4][20]; return [_dat[11][_j, 0], seq(_dat[11][_j, _vmap[_i]], _i = 1 .. _n-_ne-_nd), seq(_dat[8][1][_vmap[_i]], _i = _n-_ne-_nd+1 .. _n-_ne)] end if; if not type(_dtbl[_i], 'array') then _dtbl[_i] := `dsolve/numeric/SC/IVPdcopy`(_dtbl[1], `if`(assigned(_dtbl[_i]), _dtbl[_i], NULL)); if 0 < _nv then for _j to _nv+1 do if _j <= _nv and not type(_dtbl[_i][3][4][_j, _i-1], 'undefined') then userinfo(3, {'events', 'eventreset'}, `reinit #5, event code `, _j, ` to defined init `, _dtbl[_i][3][4][_j, _i-1]); _dtbl[_i][3][1][_j, 8] := _dtbl[_i][3][4][_j, _i-1] elif _dtbl[_i][3][1][_j, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_j, 4]), 32), 2) = 1 then userinfo(3, {'events', 'eventreset'}, `reinit #5, event code `, _j, ` to rate hysteresis init `, _dtbl[_i][5][24]); _dtbl[_i][3][1][_j, 8] := _dtbl[_i][5][24] elif _dtbl[_i][3][1][_j, 2] = 0 and irem(iquo(round(_dtbl[_i][3][1][_j, 4]), 2), 2) = 0 then userinfo(3, {'events', 'eventreset'}, `reinit #5, event code `, _j, ` to initial init `, _x0); _dtbl[_i][3][1][_j, 8] := _x0 else userinfo(3, {'events', 'eventreset'}, `reinit #5, event code `, _j, ` to fireinitial init `, _x0-2*_i+5.0); _dtbl[_i][3][1][_j, 8] := _x0-2*_i+5.0 end if end do end if end if; if _xin <> "last" then if 0 < 0 then if `dsolve/numeric/checkglobals`(op(_dtbl[1][14]), _pars, _n, _y0) then `dsolve/numeric/SC/reinitialize`(_dtbl, _y0, _n, procname, _pars, _i) end if end if; if _dtbl[1][4][7] = 0 then error "parameters must be initialized before solution can be computed" end if end if; _dat := eval(_dtbl[_i], 2); _dtbl[4] := _i; try _src := `dsolve/numeric/SC/IVPrun`(_dat, _xout) catch: userinfo(2, `dsolve/debug`, print(`Exception in solnproc:`, [lastexception][2 .. -1])); error  end try; if _src = 0 and 100 < _dat[4][9] then _val := _dat[3][1][_nv+1, 8] else _val := _dat[11][_dat[4][20], 0] end if; if _src <> 0 or _dat[4][9] <= 0 then _dtbl[1][5][1] := _xout else _dtbl[1][5][1] := _val end if; if _i = 3 and _val < _xout then Rounding := -infinity; if _dat[4][9] = 1 then error "cannot evaluate the solution further right of %1, probably a singularity", evalf[8](_val) elif _dat[4][9] = 2 then error "cannot evaluate the solution further right of %1, maxfun limit exceeded (see <a href='http://www.maplesoft.com/support/help/search.aspx?term=dsolve,maxfun' target='_new'>?dsolve,maxfun</a> for details)", evalf[8](_val) elif _dat[4][9] = 3 then if _dat[4][25] = 3 then error "cannot evaluate the solution past the initial point, problem may be initially singular or improperly set up" else error "cannot evaluate the solution past the initial point, problem may be complex, initially singular or improperly set up" end if elif _dat[4][9] = 4 then error "cannot evaluate the solution further right of %1, accuracy goal cannot be achieved with specified 'minstep'", evalf[8](_val) elif _dat[4][9] = 5 then error "cannot evaluate the solution further right of %1, too many step failures, tolerances may be too loose for problem", evalf[8](_val) elif _dat[4][9] = 6 then error "cannot evaluate the solution further right of %1, cannot downgrade delay storage for problems with delay derivative order > 1, try increasing delaypts", evalf[8](_val) elif _dat[4][9] = 10 then error "cannot evaluate the solution further right of %1, interrupt requested", evalf[8](_val) elif 100 < _dat[4][9] then if _dat[4][9]-100 = _nv+1 then error "constraint projection failure on event at t=%1", evalf[8](_val) elif _dat[4][9]-100 = _nv+2 then error "index-1 and derivative evaluation failure on event at t=%1", evalf[8](_val) elif _dat[4][9]-100 = _nv+3 then error "maximum number of event iterations reached (%1) at t=%2", round(_dat[3][1][_nv+1, 3]), evalf[8](_val) else if _Env_dsolve_nowarnstop <> true then `dsolve/numeric/warning`(StringTools:-FormatMessage("cannot evaluate the solution further right of %1, event #%2 triggered a halt", evalf[8](_val), round(_dat[3][1][_dat[4][9]-100, 1]))) end if; Rounding := 'nearest'; _xout := _val end if else error "cannot evaluate the solution further right of %1", evalf[8](_val) end if elif _i = 2 and _xout < _val then Rounding := infinity; if _dat[4][9] = 1 then error "cannot evaluate the solution further left of %1, probably a singularity", evalf[8](_val) elif _dat[4][9] = 2 then error "cannot evaluate the solution further left of %1, maxfun limit exceeded (see <a href='http://www.maplesoft.com/support/help/search.aspx?term=dsolve,maxfun' target='_new'>?dsolve,maxfun</a> for details)", evalf[8](_val) elif _dat[4][9] = 3 then if _dat[4][25] = 3 then error "cannot evaluate the solution past the initial point, problem may be initially singular or improperly set up" else error "cannot evaluate the solution past the initial point, problem may be complex, initially singular or improperly set up" end if elif _dat[4][9] = 4 then error "cannot evaluate the solution further left of %1, accuracy goal cannot be achieved with specified 'minstep'", evalf[8](_val) elif _dat[4][9] = 5 then error "cannot evaluate the solution further left of %1, too many step failures, tolerances may be too loose for problem", evalf[8](_val) elif _dat[4][9] = 6 then error "cannot evaluate the solution further left of %1, cannot downgrade delay storage for problems with delay derivative order > 1, try increasing delaypts", evalf[8](_val) elif _dat[4][9] = 10 then error "cannot evaluate the solution further right of %1, interrupt requested", evalf[8](_val) elif 100 < _dat[4][9] then if _dat[4][9]-100 = _nv+1 then error "constraint projection failure on event at t=%1", evalf[8](_val) elif _dat[4][9]-100 = _nv+2 then error "index-1 and derivative evaluation failure on event at t=%1", evalf[8](_val) elif _dat[4][9]-100 = _nv+3 then error "maximum number of event iterations reached (%1) at t=%2", round(_dat[3][1][_nv+1, 3]), evalf[8](_val) else if _Env_dsolve_nowarnstop <> true then `dsolve/numeric/warning`(StringTools:-FormatMessage("cannot evaluate the solution further left of %1, event #%2 triggered a halt", evalf[8](_val), round(_dat[3][1][_dat[4][9]-100, 1]))) end if; Rounding := 'nearest'; _xout := _val end if else error "cannot evaluate the solution further left of %1", evalf[8](_val) end if end if; if _EnvInFsolve = true then _dig := _dat[4][26]; _dat[4][26] := _EnvDSNumericSaveDigits; _Env_dsolve_SC_native := true; if _dat[4][25] = 1 then _i := 1; _dat[4][25] := 2 else _i := _dat[4][25] end if; _val := `dsolve/numeric/SC/IVPval`(_dat, _xout, _src); _dat[4][25] := _i; _dat[4][26] := _dig; [_xout, seq(_val[_vmap[_i]], _i = 1 .. _n-_ne)] else Digits := _dat[4][26]; _val := `dsolve/numeric/SC/IVPval`(eval(_dat, 2), _xout, _src); [_xout, seq(_val[_vmap[_i]], _i = 1 .. _n-_ne)] end if end proc, (2) = Array(0..0, {}), (3) = [t, x(t)], (4) = [A = A]}); _vars := _dat[3]; _pars := map(rhs, _dat[4]); _n := nops(_vars)-1; _solnproc := _dat[1]; if not type(_xout, 'numeric') then if member(x_rkf45, ["start", 'start', "method", 'method', "left", 'left', "right", 'right', "leftdata", "rightdata", "enginedata", "eventstop", 'eventstop', "eventclear", 'eventclear', "eventstatus", 'eventstatus', "eventcount", 'eventcount', "laxtol", 'laxtol', "numfun", 'numfun', NULL]) then _res := _solnproc(convert(x_rkf45, 'string')); if 1 < nops([_res]) then return _res elif type(_res, 'array') then return eval(_res, 1) elif _res <> "procname" then return _res end if elif member(x_rkf45, ["last", 'last', "initial", 'initial', "parameters", 'parameters', "initial_and_parameters", 'initial_and_parameters', NULL]) then _xout := convert(x_rkf45, 'string'); _res := _solnproc(_xout); if _xout = "parameters" then return [seq(_pars[_i] = _res[_i], _i = 1 .. nops(_pars))] elif _xout = "initial_and_parameters" then return [seq(_vars[_i+1] = [_res][1][_i+1], _i = 0 .. _n), seq(_pars[_i] = [_res][2][_i], _i = 1 .. nops(_pars))] else return [seq(_vars[_i+1] = _res[_i+1], _i = 0 .. _n)] end if elif type(_xout, `=`) and member(lhs(_xout), ["initial", 'initial', "parameters", 'parameters', "initial_and_parameters", 'initial_and_parameters', NULL]) then _xout := convert(lhs(x_rkf45), 'string') = rhs(x_rkf45); if type(rhs(_xout), 'list') then _res := _solnproc(_xout) else error "initial and/or parameter values must be specified in a list" end if; if lhs(_xout) = "initial" then return [seq(_vars[_i+1] = _res[_i+1], _i = 0 .. _n)] elif lhs(_xout) = "parameters" then return [seq(_pars[_i] = _res[_i], _i = 1 .. nops(_pars))] else return [seq(_vars[_i+1] = [_res][1][_i+1], _i = 0 .. _n), seq(_pars[_i] = [_res][2][_i], _i = 1 .. nops(_pars))] end if elif type(_xout, `=`) and member(lhs(_xout), ["eventdisable", 'eventdisable', "eventenable", 'eventenable', "eventfired", 'eventfired', "direction", 'direction', NULL]) then return _solnproc(convert(lhs(x_rkf45), 'string') = rhs(x_rkf45)) elif _xout = "solnprocedure" then return eval(_solnproc) elif _xout = "sysvars" then return _vars end if; if procname <> unknown then return ('procname')(x_rkf45) else _ndsol; _ndsol := pointto(_dat[2][0]); return ('_ndsol')(x_rkf45) end if end if; try _res := _solnproc(_xout); [seq(_vars[_i+1] = _res[_i+1], _i = 0 .. _n)] catch: error  end try end proc

(1)

k := 1:
for a in [1, 2] do
  f(parameters=[a]):
  printf(" f(1) = %a\n", f(1)):
  g||k := unapply(f(t), t):
  for kk from 1 to k do
    printf("g%d(1) = %a\n", kk, g||kk(1)):
  end do:
  k := k+1:
  print():
end do:

 f(1) = [t = 1., x(t) = .500000000000001]
g1(1) = [t = 1., x(t) = .500000000000001]

 

 

 f(1) = [t = 1., x(t) = .999999999999999]
g1(1) = [t = 1., x(t) = .999999999999999]
g2(1) = [t = 1., x(t) = .999999999999999]

 

(2)

# how must I correct this in order to prevent the
# "over writting" of g1 when g2 is instanciated
# and get
#
#  f(1) = [t = 1., x(t) = .999999999999999]
# g1(1) = [t = 1., x(t) = 0.5]
# g1(2) = [t = 1., x(t) = .999999999999999]
#

 


 

Download Stack_procedures.mw

1 2 3 4 5 6 7 Last Page 1 of 12