Fortran and AI

[JohnNichols]

I have had my head buried in coding for the last month.  

The WWW's browsers often help with AI suggestions,  they are at best hopeless and at worst leading me to a programming cliff.  

Sample, one yesterday, I finally dragged out a textbook from 2015, I know before COVID, but it showed me in 2 minutes the mistakes in all the code I had tried.  

Here I am less than impressed.  

John

[witwald]

Out of curiosity, what were the coding mistakes in the Fortran code that were causing you grief?

[JohnNichols]

accessing drivers for data collection.

[JohnNichols]

I tried in three languages, I would have tried LISP, but it is too slow.  

The issue is the lack of samples in the current manuals, it is getting to the annoying stage.  

AI reads the manuals and spits it back, if the manual does not show a reference to a DLL then the AI is blind, whereas a human can say, there must be a dll.  As a simple sample. 

This is the loss rate for frequency for a small concrete solid block 9 kg measured on 7 frequencies over 2 weeks.  The brick came off a stack and is now hanging 1/3 over a edge.  The loss rate is 100 times higher than it should be, the question is will the new accelerometer settle down to 20 microHz per day from the current 0.02 Hz per day.  

X is the long flat axis.  Y and Z are not distinct.  The increase in the beginning is likely due to concrete gel settlement that is overcome by the load application after a few days and the gel starts to crack.  

X axis puts the gel into tension from the dynamic load, even though it is only thermal.