Invalid Floating Point Error and Region Settings

Some modules will return an invalid floating point error when running. In particular routines in ETM and Resample. Most likely this error is due to the Region Settings in Windows and the default Decimal Symbol. To resolve these errors it may be necessary to change the Decimal Symbol from a "," to a ".". 

You can alter this setting in Windows \ Control Panel \ Region \ Additional Settings.

 

Have more questions? Submit a request

8 Comments

  • 0
    Avatar
    Satyam Verma

    I have changed the decimal symbol in region tab/additional settings to '.'. But still, the software shows the same error when I try to run MLP in Land Change Modeler. Please let me know if there is another solution. 

  • 0
    Avatar
    Pablo Pardo Fernández

    I am experimenting the same problem running a MOLA process on the Spatial Decision Manager space with force contiguous allocations option checked. (If I don't check this option the process works fine)

    The process starts perfectly until the end of the first of the three steps. It is just at this moment when a new prompt appears with an Invalid Floating Point Error.

    I did a first try of the process using only raster files with real datatype. Then I did a second try usign only raster files with integer datatype. Both processes ends with the same message.

    My regional configuration is selected as English (United States). Decimal separator is the '.'

  • 0
    Avatar
    Pablo Pardo Fernández

    Additional information to my provious comment.

    If I try to assing an area requirement over 999999 pixels with Force contiguous allocation option checked the problem persists, but if the area requirement is under or equal to 999999 the process goes on.

    I am working with 6 raster datafiles.

    Is not possible to use values over 999999 pixels???

    Regards.

  • 0
    Avatar
    Farnoosh Aslami

    any solution to this?

  • 0
    Avatar
    Clark Labs

    Although the decimal setting is a common cause for an Invalid Floating Point error, there are other causes. 

    As for the decimal settings, we have been working on making modules more flexible with those settings, one by one as we become aware of them.  The easiest solution is to change your settings, but we also do like to keep track of modules that are problematic and to make changes to them as we have time.  If you have one of these errors, do contact technical support (submit a ticket), and let us know what version you are using (see the About box), in case it has already been fixed.

    As for the problem that Pablo Pardo Fernandez is mentioning with MOLA, I don't think it's related to regional settings.  There was a bug report for MOLA: - can get an invalid floating point error if area requested is > 1,046,529 pixels - that was fixed on 2/15/22, so the latest patch should fix it.  If not, open a tech support ticket (support request) and we'll ask you to send us your data so we can identify the problem.

     

    Satyam Verma, I would like to know what your version is and if you still have that problem.

    Farnoosh Aslami, if you are having a problem, let us know with which specific module/operation and what release/version you are using.  Also, let us know if you are using the comma or the dot for the decimal setting.

  • 0
    Avatar
    Clark Labs

    One of the common causes of an "Unexpected Error Encountered" message if the background value for your data is -INF or negative infinity (-3.4E+0038) or contains NaN values.  Some modules aren't able to handle those numbers, even though some data formats now use them as a background values.  If a module tries to do multiplication functions on –INF or NaN, it goes out of a valid range and causes an Unexpected/untrapped error.

    We have an easy fix for this -- a module called NANFIX.  NaN means Not a Number.  This module is designed to replace either NaN values or INF or -INF values with a different, user-specified value (e.g. -999).

    It will convert all of the INF/NaN values in a single RST file, or in batch mode, it will convert all of the raster files in an input folder, and create fixed files of the same names in an output folder. 

     

    Edited by Clark Labs
  • 0
    Avatar
    Dian Eva Solikha

    any solution of this problem,  i dont know how to fix it

  • 0
    Avatar
    Clark Labs

    Dian Eva Solikha,

    A couple of solutions were mentioned:

    1. Running NANFIX on your file if you have infinity (+ or - 1E38) values or NaN values in your image -- to replace them with a valid number.

    2. Setting your decimal setting to "." instead of "," if your regional settings are such that a comma (",") is the default.

    3. It doesn't mention it, but many modules were updated in the last 4 years to be able to not give an error if the decimal settings were a comma instead of a period, so making sure you are using the latest version (19.08) may solve your problem if that is what it is.

    This thread is mentioning several types of problems and your comment/question doesn't give any details about what specifically your problem is.  You might want to open a tech support ticket and give details about what operations you are running, what types of data you are using and what your decimal setting is, if it seems like that is the problem.

Please sign in to leave a comment.
Powered by Zendesk