## Boot Pi and Embrace Tau?

July 1st, 2011 by Math Tricks | 4 Comments | Filed in Bad Math

``` <!-- google_ad_client = "ca-pub-2176115693811858"; /* Jelly Babies */ google_ad_slot = "2358050469"; google_ad_width = 468; google_ad_height = 60; //--> ```

There has been a movement over the last decade to replace pi with it’s 2X equivalent – tau.  Bob Palais of the University of Utah and his followers insist that this would be a more meaningful ‘constant’, and would lead to less mistakes in the calculations of innumerable individuals.

Nice try, Palais et al.  Would you also have us trash Euler’s identity or the countless text books in circulation?  Sounds like you are just trying to cement your place in mathematical history.  Well, IMHO, this is not a good idea, and I am 100% against it.  I really don’t think much will come of this, but, just in case, maybe I should take in some of the textbook publishers out there!

## Excel Rounding

September 13th, 2009 by Math Tricks | 3 Comments | Filed in Bad Math

``` <!-- google_ad_client = "ca-pub-2176115693811858"; /* Excel Rounding */ google_ad_slot = "7409143208"; google_ad_width = 468; google_ad_height = 60; //--> ```

## Excel Rounding

In my line of work, we generate polynomial curves very often.  It is important for us to have software that will generate polynomial equations for the data that we generate – equations for the standard curves that we generate so that we can determine values for our unknown samples.

This is both an example of real life math and bad math.  To start, I’ll just say that we used to generate linear curves for our assays.  The data for these assays were analyzed in Microsoft Excel, and we were very satisfied with the linear curves generated by Excel.

Last year, we changed our assay such that the standard curve that was generated was no longer linear.  The new type of curve was a second-order polynomial (of the form AX2 + BX + C).  So, OK, Excel can handle polynomial data, right?  Well, we discovered a problem with Excel – Excel rounding of the polynomial coefficients!  The problem seems to exist with only large coefficients, but we do deal with polynomials with large coefficients every day.

To illustrate the problem, I generated some data for the equation:

Y = 8386111X2 + 24421000X + 0.058

Here is a table of data generated using this equation:

 X Y 0.00003 732.6955475 0.00033 8059.901247 0.00333 81414.98075 0.03333 823268.0256 0.33333 9072023.463 3.33333 174582076.7

So far so good.  OK, so using Excel (Microsoft Office 2003), I generated this polynomial curve:

As you can see, the equation of the curve generated by Excel is not the same as the equation that we used to generate the data used in our example.  There was some Excel rounding of the coefficients A and B:

Y = 8386111X2 + 24421000X + 0.058     equation used to generate data

Y = 8E+06X2 + 2E+07X + 0.058    equation generated by Excel

Close, but no cigar!  In our line of work, we require precise and accurate data, ant this Excel rounding problem is unacceptable.  We just got in Office 2007, but the problem still exists there too.  We are now using GraphPad, which does not round.

Well, as it is getting close to Halloween, I thought it appropriate to write about this example of Microsoft math tricks – not treats.  Has anybody else discovered any limitations with Excel?  Please feel free to post!