Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-L] how research is done : exploring a maze using only local information




On 2015, Sep 18, , at 11:35, Richard Tarara <rtarara@saintmarys.edu> wrote:

But is a 'wrong' result recognized as such or is the data/analysis bent to agree with the hypothesis? We here might not call that 'science' but there seem to be plenty of others who would, and thus we end up with climate change nay-sayers, anti-evolution fanatics, even young earth advocates.

rwt



This reminds me of an incorrect conclusion regarding the linearity of a beam oscillator. The result was, within the statistic the author used, no change in frequency w/ amplitude. Ignoring what I knew about the PDE of beams and experimental evidence,(1) I immediately questioned the result, because the change (four amplitudes) in frequency (as a function of amplitude) was monotonic increasing.

I suppose there is a test for the probability of such a monotonic change tho there isn’t. (No change shown by a more “accurate” measurement and theoretically.) At this point in writing I searched this computer for the article and found the method was to video the motion, plot, and fit a simple cosine.(2) The author didn’t report the fitting program, but the result is in the same form as Kaleidagraph’s, which gives the vales +/- values. The author concluded the “… variation in angular frequency is statistically insignificant.” The author used the mean and standard deviation of the four trials (amplitudes —min/max ~ 5.8/11.0 cm 50 cm length) to conclude this, because, for a doubling of the amplitude the percent deviation was 0.003%. However, not only is this method invalid, but also the correct deviation is 0.3%. A better simple method is to note the fit result over five cycles is 3.025 +/- 0.0016 rad/s (the authors example), therefore, assuming similar accuracy for the other three trials, the change in frequency is well above any error. The conclusion that the cantilever beam oscillator is non-linear is inescapable. However, the authors assumption that the amplitude must not be very constant to measure the frequency as a function of beam length is correct. (with in the desired accuracy and used amplitude)


(1) A driven hack saw oscillator exhibits jumps characteristic of a non-linear oscillator. http://senate.universityofcalifornia.edu/inmemoriam/RonaldH.Ruby.htm (This was ca. 1970.)

(2)Tho the data graphed for the one example given shows decay. A better fit is A*[exp(-E*t)]*cos(B*t+C)+D


bc