Does anyone know if there is a function (in any language or package) for simulating of random sequence with given correlation function? I didn't see such a thing either in MATLAB or in R. maybe anywhere else?
Hi
Here is how it is done for structural analysis with random correlated fields, e.g. wind loads that blow across structures.
http://www.egr.msu.edu/software/abaqus/Documentation/docs/v6.7/books/stm/default.htm?startat=ch02s05ath31.html
http://www.egr.msu.edu/software/abaqus/Documentation/docs/v6.7/books/usb/default.htm?startat=pt03ch06s03at15.html
Sincerely
Claes
Thank you,
I saw this mathematics. But is there a ready-made function anywhere? Suppose it should be because how can one simulate any natural process without such a function?
And if this function exists yet I'd like to use it instead of programming it myself as this mathematics is not very simple.
Hi
I guess a fundamental problem to your question is that there may be ready-made software but this software differs from one case to the next as these have different problems.
To exemplify, low altitude wind correlation differs from that at high height, i.e. you would use different equations for wind loads on a building and that of an aeroplane.
The mathematics you refer to, at least for wind, usually is a bunch of equations that have been gathered and verified for particular situations. Their implementation on a FE model may look overwhelmning but tends to boil down to a limited set of equations that express loads as a function of space, time, frequency etc.
Sincerely
Claes
There are different problems with different levels of abstraction. Sometimes the maximum we can hope to know about the random process is its mean and maybe variance, both very roughly because of the great spatial and temporal variability of the environment e.g. of the ocean one.
There are two ways in such situations. Either to assume that we know a lot of data (that we don't know in reality) and then obtain accurate results (which don't mean anything) or to use the poor available information and get rough results. I prefer the second way. And the function I'm looking for is necessary to check some conclusions I had obtained on this way, by means of simulation.
Nevertheless thank you for your answers.
Maria
Hi
Aha, you want to identify, not apply, random processes.
Perhaps this is more what you are looking for?
WAFO is a free Matlab toolbox. (I list what is written there for the sake of other readers and searchability)
Fatigue Analysis
- Fatigue life prediction for random loads
- Theoretical density of rainflow cycles
Sea modelling
- Simulation of linear and non-linear Gaussian waves
- Estimation of seamodels (spectrums)
- Joint wave height, wave steepness, wave period distributions
Statistics
- Extreme value analysis
- Kernel density estimation
- Hidden markov models
http://www.maths.lth.se/matstat/wafo/
You might like also the HOSA- Higher Order Spectral Analysis Toolbox
http://www.mathworks.com/matlabcentral/fileexchange/3013-hosa-higher-order-spectral-analysis-toolbox
Hope this helps
Claes
Thank you Claes,
especially for WAFO, never heard about it.
Sincerely
Maria
Hi Maria
WAFO is worth looking at as it contains many manyears collected research effort across several countries.
Enjoy
Claes
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics