Flux for sparse data

I need to learn how to compute the flux for sparse data.

I looked at the volumetric GPM data for precipitation and it does not look like a continuous function, rather it is sparse partitioned and distanced spread of regions of moisture.

So my previous work to compute the Laplacian might or might not apply.

I was wondering how one goes about mathematically computing the flux of sparse data.

Dara

Comments

  • 1.

    Kriging? http://en.wikipedia.org/wiki/Kriging

    AKA Gaussian process regression and a form of geo-spatial interpolation

    Comment Source:Kriging? <http://en.wikipedia.org/wiki/Kriging> AKA Gaussian process regression and a form of geo-spatial interpolation
  • 2.

    thanx paul, I study this

    Comment Source:thanx paul, I study this
  • 3.

    Paul check out the NonlinearModelFit [ ] in Mathematica I think it is using the Gaussian process.

    Comment Source:Paul check out the NonlinearModelFit [ ] in Mathematica I think it is using the Gaussian process.
  • 4.

    Dara, I am using these options for NonLinearModelFit :

    Method -> "NMinimize"

    Method -> "Differential Evolution"

    Comment Source:Dara, I am using these options for NonLinearModelFit : Method -> "NMinimize" Method -> "Differential Evolution"
  • 5.

    You are a genius Paul.

    Paul Mathematica's Differential Evolution is not parallelized, I want to write a version in C that is parallelized in GPU servers.

    Dara

    Comment Source:You are a genius Paul. Paul Mathematica's Differential Evolution is not parallelized, I want to write a version in C that is parallelized in GPU servers. Dara
  • 6.

    Dara, I still do not know if the Eureqa tool I occasionally use incorporates Differential Evolution, but it certainly does use all 8 of the CPU's when it is running.

    Paul

    Comment Source:Dara, I still do not know if the Eureqa tool I occasionally use incorporates Differential Evolution, but it certainly does use all 8 of the CPU's when it is running. Paul
Sign In or Register to comment.