replications = posint -- This option specifies the number of replications of the sample path. By default, only one replication of the sample path is generated.

timesteps = posint -- This option specifies the number of time steps. This option is ignored if explicit time grid is specified. By default, only one time step is used.

The SamplePath command generates sample paths for a given stochastic process. This command can handle one- as well as multi-dimensional processes.

In the one-dimensional case, the value returned by SamplePath is a matrix of size $m\times n$, where $m$ is the requested number of replications of the sample path (see the replications option) and $n$ is the number of points in the time grid (the number of time steps plus one). Each row of this matrix is a single realization of the sample path of the stochastic process of interest. In the multi-dimensional case, the value returned by SamplePath is a matrix of size $m\times d\times n$, where $m$ and $n$ are the same as in the one-dimensional case, and $d$ is the dimension of the stochastic process to be simulated, which is either given explicitly or deduced from the dimension of the underlying stochastic process.

The SamplePath(expression, t = timeinterval, opts) calling sequence attempts to extract all the stochastic variables involved in expression and generate the corresponding path generator and path function using the time grid or the specified number of time steps. In this case, only a one-dimensional path can be generated. The parameter timeinterval must be of type range $T_0..T_1$, where $T_0$ and $T_1$ are non-negative constants such that $T_0<T_1$.

Note that if $0<T_0$, the value at $T_0$ will be simulated using a single step of the default discretization method and hence can suffer from a significant discretization bias. Increasing the number of time steps will refine the grid between $T_0$ and $T_1$, but will have no effect on the value at $T_0$. To reduce the bias, use a time interval of the form $0..T_1$. All stochastic variables involved in expression should be of the form $X\left(t\right)$. If $X$ is multi-dimensional stochastic, then the individual components of $X$ can be accessed using the notation ${X\left(t\right)}_i$.

The SamplePath(pathgenerator, opts) calling sequence generates the specified number of replications of the sample path using the path generator pathgenerator. For more details, see Finance[PathGenerator].

The SamplePath(transform, pathgenerator, opts) calling sequence generates sample paths for a certain one-dimensional stochastic process given a path generator pathgenerator and a procedure transform that converts paths generated by pathgenerator to paths of the stochastic process of interest. Note that pathgenerator can be multi-dimensional.

To generate a single realization of the sample path of interest, SamplePath generates a sample path for the base stochastic process using the path generator pathgenerator and stores these values as an Array, for example $A$. If pathgenerator is one-dimensional, $A$ is a one-dimensional array of size $n$. If pathgenerator is multi-dimensional, $A$ is a two-dimensional array of size $k\times n$, where $k$ is the dimension of pathgenerator. Then SamplePath evaluates $f\left(B\&comma;A\right)$, where $B$ is a preallocated one-dimensional floating-point array of the same size as $A$. The values stored in $B$ are used as the new replication of the sample path and stored as a row in the resulting matrix. This process will be repeated for the specified number of replications.

The SamplePath(d, transform, pathgenerator, opts) calling sequence uses the same method, but with a multi-dimensional process. Therefore, $A$ is now of size $d\times n$. Note the difference between the SamplePath(transform, pathgenerator, opts) and SamplePath(1, transform, pathgenerator, opts) calling sequences. The first returns an array of size $m\times n$, while the other returns an array of size $m\times 1\times n$.

The SamplePath command uses the Maple random number generator (see RandomTools[MersenneTwister]) to generate the intermediate random numbers required by the corresponding algorithm. As a consequence, all values generated by SamplePath are deterministic, meaning the same values are generated every Maple session. Use randomize or RandomTools[MersenneTwister][SetState] to randomize the state of the internal random number generator.

$\mathrm{with}\left(\mathrm{Finance}\right)\&colon;$

$X≔\mathrm{WienerProcess}\left(\right)\&colon;$

$W≔\mathrm{WienerProcess}\left(⟨⟨1.0\&comma;0.5⟩|⟨0.5\&comma;1.0⟩⟩\right)\&colon;$

$A≔\mathrm{SamplePath}\left(X\left(t\right)\&comma;t=0..1\&comma;\mathrm{timesteps}=100\&comma;\mathrm{replications}=20\right)$

$A\left[1\right]$

$B≔\mathrm{SamplePath}\left(W\left(t\right)\left[1\right]+W\left(t\right)\left[2\right]\&comma;t=0..1\&comma;\mathrm{timesteps}=100\&comma;\mathrm{replications}=20\right)$

$B\left[1\right]$

$P≔\mathrm{PathGenerator}\left(X\&comma;\mathrm{TimeGrid}\left(5\&comma;100\right)\right)\&colon;$

$A≔\mathrm{SamplePath}\left(P\&comma;\mathrm{replications}=20\right)$

$M≔\mathrm{PathGenerator}\left(W\&comma;\mathrm{TimeGrid}\left(5\&comma;100\right)\right)\&colon;$

$B≔\mathrm{SamplePath}\left(M\&comma;\mathrm{replications}=20\right)$

$A\left[1\right]$

$B\left[1\right]$

f := proc(B, A) local d, i; d := rtable_dims(A); for i from lhs(d) to rhs(d) do B[i] := exp(A[i]) end do; end proc;

$f≔\mathbf{proc}\left(B\&comma;A\right)\mathbf{local}d\&comma;i\&semi;d≔\mathrm{rtable\_dims}\left(A\right)\&semi;\mathbf{for}i\mathbf{from}\mathrm{lhs}\left(d\right)\mathbf{to}\mathrm{rhs}\left(d\right)\mathbf{do}B\&lsqb;i\&rsqb;≔\exp \left(A\&lsqb;i\&rsqb;\right)\mathbf{end\; do}\mathbf{end\; proc}$

$\mathrm{SamplePath}\left(f\&comma;P\&comma;\mathrm{replications}=20\right)$

f2 := proc(B, A) local d, i; d := rtable_dims(A)[2]; for i from lhs(d) to rhs(d) do B[i] := A[1, i]+A[2, i] end do; end proc:

$\mathrm{SamplePath}\left(\mathrm{f2}\&comma;M\&comma;\mathrm{replications}=20\right)$

F2 := proc(B, A) local d, i; d := rtable_dims(A)[2]; for i from lhs(d) to rhs(d) do B[1, i] := A[1, i] + A[2, i]; end do; end proc:

$\mathrm{SamplePath}\left(1\&comma;\mathrm{F2}\&comma;M\&comma;\mathrm{replications}=20\right)$

F := proc(B, A) local d, i; d := rtable_dims(A)[2]; for i from lhs(d) to rhs(d) do B[1, i] := A[1, i]; B[2, i] := A[2, i]; B[3, i] := A[1, i] + A[2, i]; end do; end proc:

$\mathrm{SamplePath}\left(3\&comma;F\&comma;M\&comma;\mathrm{replications}=20\right)$

$S≔\mathrm{RandomTools}:-\mathrm{GetState}\left(\right)\&colon;$

$\mathrm{RandomTools}:-\mathrm{SetState}\left(\mathrm{state}=S\right)\&colon;$

$\mathrm{A1}≔\mathrm{SamplePath}\left({X\left(t\right)}^2\&comma;t=0..10\&comma;\mathrm{timesteps}=10\right)$

$\mathrm{RandomTools}:-\mathrm{SetState}\left(\mathrm{state}=S\right)\&colon;$

f := proc(B, A) local i; for i from 1 to 11 do B[i] := A[i]^2; end do; end proc:

$\mathrm{A2}≔\mathrm{SamplePath}\left(f\&comma;X\&comma;0..10\&comma;\mathrm{timesteps}=10\right)$

$T≔\mathrm{TimeGrid}\left(\left[0\&comma;1\&comma;2\&comma;3\&comma;4\&comma;5\&comma;6\&comma;7\&comma;8\&comma;9\&comma;10\right]\right)$

$T≔\mathbf{module}\left(\right)\mathbf{end\; module}$

$\mathrm{SamplePath}\left(⟨-X\left(t\right)\&comma;X\left(t\right)⟩\&comma;t=0..10\&comma;\mathrm{timesteps}=10\right)$

$\mathrm{\Sigma }≔⟨⟨1.0\&comma;0.5⟩|⟨0.5\&comma;1.0⟩⟩$

$\mathrm{\Sigma }≔\left[\begin{array}{cc}1.0& 0.5\\ 0.5& 1.0\end{array}\right]$

$W≔\mathrm{WienerProcess}\left(\mathrm{\Sigma }\right)$

$W≔\mathrm{\_W1}$

$\mathrm{SamplePath}\left(W\left(t\right)\&comma;t=0..3\&comma;\mathrm{timesteps}=10\&comma;\mathrm{replications}={10}^4\right)$

$S≔\mathrm{SamplePath}\left(⟨W\left(t\right)\left[1\right]\&comma;W\left(t\right)\left[2\right]\&comma;W\left(t\right)\left[1\right]+W\left(t\right)\left[2\right]⟩\&comma;t=0..3\&comma;\mathrm{timesteps}=9\&comma;\mathrm{replications}={10}^4\right)$

$\mathrm{map}\left(\mathrm{evalf}\left[5\right]\&comma;S\left[1\right]\right)$

$\left[\begin{array}{cccccccccc}0.& 0.31717& \mathrm{-0.16490}& \mathrm{-1.1906}& \mathrm{-1.4647}& \mathrm{-1.5699}& \mathrm{-1.6450}& \mathrm{-1.9399}& \mathrm{-1.5251}& \mathrm{-1.3382}\\ 0.& 0.060433& \mathrm{-0.35577}& \mathrm{-1.0683}& \mathrm{-0.69408}& \mathrm{-0.12185}& \mathrm{-0.22109}& \mathrm{-0.98585}& \mathrm{-1.2004}& \mathrm{-1.1880}\\ 0.& 0.37760& \mathrm{-0.52067}& \mathrm{-2.2589}& \mathrm{-2.1587}& \mathrm{-1.6918}& \mathrm{-1.8661}& \mathrm{-2.9258}& \mathrm{-2.7255}& \mathrm{-2.5262}\end{array}\right]$

$J≔\mathrm{PoissonProcess}\left(0.3\right)$

$J≔\mathrm{\_P}$

$S≔\mathrm{SamplePath}\left(5\left(J\left(t\right)-t\right)+W\left(t\right)\left[1\right]\&comma;t=0..3\&comma;\mathrm{timesteps}=100\&comma;\mathrm{replications}=10\right)$

$\mathrm{PathPlot}\left(S\&comma;\mathrm{timegrid}=0..3\&comma;\mathrm{color}=\mathrm{red}..\mathrm{blue}\&comma;\mathrm{thickness}=3\right)$

$Y≔\mathrm{WienerProcess}\left(\right)$

$Y≔\mathrm{\_W2}$

$Z≔t\mapsto \exp \left(0.05\cdot t+0.3\cdot Y\left(t\right)\right)$

$Z≔t\mapsto {\&ExponentialE;}^{0.05\cdot t+0.3\cdot Y\left(t\right)}$

$U≔t\mapsto \mathrm{maximize}\left(Z\left(u\right)\&comma;u=0..t\right)$

$U≔t\mapsto \mathrm{maximize}\left(Z\left(u\right)\&comma;u=0..t\right)$

$S≔\mathrm{SamplePath}\left(⟨Z\left(t\right)\&comma;U\left(t\right)⟩\&comma;t=0..3\&comma;\mathrm{timesteps}=100\&comma;\mathrm{replications}=5\right)$

$V≔t\mapsto \frac{\mathrm{int}\left(Z\left(u\right)\&comma;u=0..t\right)}{t}$

$V≔t\mapsto \frac{{\int }_0^{t}Z\left(u\right)\&DifferentialD;u}{t}$

$S≔\mathrm{SamplePath}\left(⟨Z\left(t\right)\&comma;V\left(t\right)⟩\&comma;t=0..3\&comma;\mathrm{timesteps}=100\&comma;\mathrm{replications}=5\right)$

testproc := proc(Y, X) local i; for i from 1 to 11 do Y[i] := exp(X[i]); end do; end proc:

$S≔\mathrm{SamplePath}\left(\mathrm{testproc}\&comma;Y\&comma;0..3\&comma;\mathrm{timesteps}=10\&comma;\mathrm{replications}=10\right)$

$\mathrm{PathPlot}\left(S\&comma;\mathrm{timegrid}=0..3\&comma;\mathrm{thickness}=3\&comma;\mathrm{color}=\mathrm{red}..\mathrm{blue}\right)$

Glasserman, P., Monte Carlo Methods in Financial Engineering. New York: Springer-Verlag, 2004.

Hull, J., Options, Futures, and Other Derivatives, 5th. edition. Upper Saddle River, New Jersey: Prentice Hall, 2003.

Kloeden, P., and Platen, E., Numerical Solution of Stochastic Differential Equations, New York: Springer-Verlag, 1999.

The Finance[SamplePath] command was introduced in Maple 15.

For more information on Maple 15 changes, see Updates in Maple 15.