Loading ...
Sorry, an error occurred while loading the content.

[ai-geostats] "Constrained" estimation / simulation

Expand Messages
  • Steinar Løve Ellefmo
    Hi All, I m estimating and simulating an iron ore deposit. The decisive ore parameters are total iron, FeTot and iron in magnetite, FeMagn. Clearly FeMagn can
    Message 1 of 3 , Jul 20, 2004
    • 0 Attachment
      Hi All,

      I'm estimating and simulating an iron ore deposit. The decisive ore
      parameters are total iron, FeTot and iron in magnetite, FeMagn. Clearly
      FeMagn can not be larger than FeTot.

      FeMagn and FeTot is not correlated (coefficient of variation (CV) of
      FeMagn is much larger than the CV of FeTot). Domaining based on ore
      type is not practically possible.

      Is it possible to somehow "constrain" FeMagn realisations or estimations
      so that FeTot are the largest of the two?

      Any publications on the subject?

      I'm using Isatis.


      Best regards,

      Steinar Ellefmo
      Norwegian University of Science and Technology
      Norway
    • Jürgen Kropp
      Dear Steinar, you can use so-called qualitative or semi-quantitative modelling approaches. In such environments it is possible to define functions (varibles)
      Message 2 of 3 , Jul 20, 2004
      • 0 Attachment
        Dear Steinar,
        you can use so-called qualitative or semi-quantitative modelling
        approaches. In such environments it is possible to define functions
        (varibles) by means of monotonocity and thresholds. In addition,
        variables are defined by means of qualitative landmarks. Regarding
        the quantity spaces different variables can be constrained, as e.g.
        one function may be always larger than the other. But note in such
        simulations also the time is only defined in a qualitative sense.
        The relevant publication is:
        B. Kuipers (1994): Qualitative reasoning: Modeling and Simulation with
        incomplete knowledge. MIT Press, Cambridge.
        Best wishes
        Juergen


        Steinar Løve Ellefmo wrote:
        > Hi All,
        >
        > I'm estimating and simulating an iron ore deposit. The decisive ore
        > parameters are total iron, FeTot and iron in magnetite, FeMagn. Clearly
        > FeMagn can not be larger than FeTot.
        >
        > FeMagn and FeTot is not correlated (coefficient of variation (CV) of
        > FeMagn is much larger than the CV of FeTot). Domaining based on ore
        > type is not practically possible.
        >
        > Is it possible to somehow "constrain" FeMagn realisations or estimations
        > so that FeTot are the largest of the two?
        >
        > Any publications on the subject?
        >
        > I'm using Isatis.
        >
        >
        > Best regards,
        >
        > Steinar Ellefmo
        > Norwegian University of Science and Technology
        > Norway
        >
        >
        >
        > ------------------------------------------------------------------------
        >
        > * By using the ai-geostats mailing list you agree to follow its rules
        > ( see http://www.ai-geostats.org/help_ai-geostats.htm )
        >
        > * To unsubscribe to ai-geostats, send the following in the subject or in the body (plain text format) of an email message to sympa@...
        >
        > Signoff ai-geostats


        --
        _\\|//_
        (' O^O ')
        +---------------------o00-(_)-00o------------------------+
        | Dr. Juergen Kropp |
        | Potsdam Institute for Climate Impact Research |
        | Dept. Integrated Systems Analysis |
        | P.O.Box 60 12 03, 14412 Potsdam, Germany |
        | Office: Phone: +49 (0) 331 288 2526 |
        | Fax: +49 (0) 331 288 2640 |
        | mail: kropp@... |
        |--------------------------------------------------------|
        | Homepage: http://www.pik-potsdam.de/~kropp/ |
        | Projects: http://www.pik-potsdam.de/skalenanalyse/ |
        | http://www.pik-potsdam.de/~kropp/compromise/ |
        +--------------------------------------------------------+
      • Julián Ortiz
        Hi Steinar, The exact same problem you have is addressed in: Stepwise Conditional Transformation for Simulation of Multiple Variables by Oy Leuangthong and
        Message 3 of 3 , Jul 20, 2004
        • 0 Attachment
          Hi Steinar,

          The exact same problem you have is addressed in:

          "Stepwise Conditional Transformation for Simulation of Multiple Variables"
          by
          Oy Leuangthong and Clayton V. Deutsch, Math Geology 35(2):155-173, Feb.
          2003.

          What they propose is to normally transform the secondary variable
          conditionally to
          classes of the normally transformed primary variable. This transformation
          removes
          the cross-correlation at distance 0 (collocated) and often minimizes
          cross-correlation
          at lags different than 0, between the variables. Gaussian simulation can
          then be performed
          independently for each conditionally transformed variable. The simulated
          values must be
          back-transformed using the same "rule" created in the forward
          transformation, which will
          reinject any order relation they may have (such as FeTot>=FeMagn).

          Hope this helps.

          Julián Ortiz C., Ph. D.
          Assistant Professor
          Department of Mining Engineering
          University of Chile

          Phone: +56 2 678 4585
          Fax : +56 2 672 3504
          Web : www.ualberta.ca/~jmo1


          -----Mensaje original-----
          De: Steinar Løve Ellefmo [mailto:steinar.ellefmo@...]
          Enviado el: martes, 20 de julio de 2004 5:49
          Para: ai-geostats@...
          Asunto: [ai-geostats] "Constrained" estimation / simulation


          Hi All,

          I'm estimating and simulating an iron ore deposit. The decisive ore
          parameters are total iron, FeTot and iron in magnetite, FeMagn. Clearly
          FeMagn can not be larger than FeTot.

          FeMagn and FeTot is not correlated (coefficient of variation (CV) of
          FeMagn is much larger than the CV of FeTot). Domaining based on ore
          type is not practically possible.

          Is it possible to somehow "constrain" FeMagn realisations or estimations
          so that FeTot are the largest of the two?

          Any publications on the subject?

          I'm using Isatis.


          Best regards,

          Steinar Ellefmo
          Norwegian University of Science and Technology
          Norway
        Your message has been successfully submitted and would be delivered to recipients shortly.