#Results log .txt scidavis mac os#
This behavior occurs with both Mac OS 10.14 and Windows 10. For example, to append to log.txt : Text icacls d:\ /restore c:\temp tfsperms.txt /t /c > 'log.txt' If you wanted to capture error messages also, redirect both standard output and standard error like this: Text icacls d:\ /restore c:\temp tfsperms.txt /t /c > 'log. But, this is really perverse and took quite a while to figure out! I don't know of a command-line switch to turn on icacls logging. Of course, that is something that shouldn't be done, but the program doesn't object and plots a curve that looks like a perfectly nice fit. To my way of thinking, the program would be well within its rights to tell the user that it won't do a curve fit with text data. The problem we have found arises when a simple 2-dimensional data graph is made with one of the columns set to text data, instead of numeric, and then a curve fit is attempted. If the data type is then changed to numeric, the data points stay where they wore, but the fit curve changes to reflect the wacky parameters. But, the parameters for the fit are wacky. The problem we have found arises when a simple 2-dimensional data graph is made with one of the columns set to text data, instead of numeric, and then a curve fit is attempted. PS: The behaviour above i seen in both 2.3.0 and 1.26 versions, independently of OS (tried Linux (debian), Mac OS and Windows 10).Every spring, I use SciDavis in a biochemistry lab course with about 70-80 undergraduates, and nearly every year, someone finds a new way to break it! It looks like this year is no exception. This bug seems to be only related to fits where yErr are requested to be taken into account.
However, user-defining an identical function y0+A*exp(-x/t), only the Nelder-Mead Simplex succeeds, while the Levenberg-Marquardts end up with "iteration not making progress towards solution" and report fit-parameter values different from N-M Simplex. PS: Strangely, when trying polynomial fits instead of the exponential one, the results match including the fit-parameter uncertainties.Īd 2) When playing with the fit options, we have also met another strange behaviour in SciDavis: when fitting the built-in ExpDecay1 function, all three algorithms (Scaled/Unscaled Levenberg-Marquardt and Nelder-Mead Simplex) report sucess and same results. I want to save that output because the bash-window just opens for a split second. Another window opens with '/usr/bin/bash -login -i \test.sh' in the title bar, shows me 'bash: blablubb: command not found' and then closes immediately. The fits account for the y-uncertainties yErr, however, similar discrepancy is seen when doing unweighted fits. After that I open the terminal (cmd.exe), switch to the directory and type test.sh 2> log.txt. I.e while the central values match, the uncertainties do not. However, using the SciDavis ExpDecay1 the results (both Scaled/Unscaled Levenberg-Marquardt and Nelder-Mead Simplex) are: The results from GnuPlot, python-scipy and are: The data we are trying to fit are: xĪd 1) We are trying to fit these data using an ExpDecay1 function, i.e. When using built-in function, the fit dialog does not offer possibility to fix parameters.
Using a built-in function and identical user-defined function exhibits variations in the fit behaviour. The results, namely the fit-parameter errors are not matching results from other programs (GnuPlot, python-scipy,, where the results are identical). Trying to fit data with an exponential function, we have met with several problems, in summary:
0 kommentar(er)
