WOLFRAM

New in Mathematica 10 › Expanded Time Series Processes ›

Broad Performance Improvements 

Mathematica 10 introduces broad performance improvements of its support for linear time series processes, including, among others, sampling, covariance function, log‐likelihood, and slices.

Timings for sampling 500 paths ensemble of a scalar ARMA(2,1) process in Mathematica 10, Mathematica 9, and in R 3.1.0 as the function of path length. The experiment was performed on an Intel Core i7 2.93 GHz 64-bit Windows system. The number at the bottom shows how much faster Mathematica 10 is than Mathematica 9.

Timings for sampling 50 paths ensemble of a two-dimensional vector ARMA(2,1) process in Mathematica 10 and Mathematica 9 as the function of path length. The experiment was performed on an Intel Xeon 3.07 GHz 64-bit Linux system. The number at the bottom shows how much faster Mathematica 10 is than Mathematica 9.

Timings for computation of an autocovariance sequence of a scalar ARMA(2,1) process in Mathematica 10, Mathematica 9, and in R 3.1.0 as the function of maximal lag. The experiment was performed on an Intel Core i7 2.93 GHz 64-bit Windows system with a time limit of 120 seconds. The number at the bottom shows how much faster Mathematica 10 is than Mathematica 9.

Timings for computation of autocovariance sequence of a two-dimensional vector ARMA(2,1) process in Mathematica 10 and Mathematica 9 as the function of maximal lag. The experiment was performed on an Intel Xeon 3.07 GHz 64-bit Linux system. The number at the bottom shows how much faster Mathematica 10 is than Mathematica 9.

Related Examples

Decide Time Series Process Family and Order from Data Automatically » Identify Conditional Heteroscedacity » Airline Passengers » Constrain the Model Selection Set »
Use Different Criteria for Model Selection » Investigate Time Series Model Residuals » Study Significance of Parameters in Fitted Model » Use Fitted Model to Forecast Time Series »
Model Multiple Exchange Rates » Vector Joint Model versus Univariate Component Models » Use Vector Models with Multiple Strongly Correlated Time Series » Vector Autoregressive Process as Discretized Vector Ornstein–Uhlenbeck Process »
Time Series Processes with Nonzero Mean » Time Series Processes with Initial Conditions » Study Non-weakly Stationary Autoregressive Process » Compute Expectation of Autoregressive Moving-Average Process from Its Definition »
Use Spectral Estimator to Find FARIMA Parameters » Volatility Clustering in a GARCHProcess » Slice Distribution of GARCH(1,1) » Moments of GARCH(1,1) »
Model the Conditional Value at Risk with an ARCHProcess » Expanded Estimation Methods » Broad Performance Improvements »
de es ja pt-br zh