What is the default decimal precision in Matlab?

I am using polyfit to derive the coefficients. What is the default decimal precision in this case and for the function polyfit?

Thanks,

Jennifer

[JFz]

Thanks for your help. I was testing a polynomial simulation. I set the polynomial to be: y = 1+2x+3x^2+4x^3+5x^4+6x^5 and generated 1000 points. Then I used polyfit(Test1_regressor{:,1}, Test1_points{:,1}, 5); the results are: 5.993278257 5.015586479 3.987602572 3.00393415 1.999564018 1.000009082 But I expected the results should be closer to 6, 5, 4, 3, 2, 1 I wonder how to improve the precision.

Thanks. Jennifer

Answer 1

[Star Strider]

All such calculations in MATLAB are full double-precision floating point. The decimal precision displayed in the Command Window and tooltips is set by the format function.

[JFz]

Thanks for your help. I was testing a polynomial simulation. I set the polynomial to be: y = 1+2x+3x^2+4x^3+5x^4+6x^5 and generated 1000 points. Then I used polyfit(Test1_regressor{:,1}, Test1_points{:,1}, 5); the results are: 5.993278257 5.015586479 3.987602572 3.00393415 1.999564018 1.000009082 But I expected the results should be closer to 6, 5, 4, 3, 2, 1 I wonder how to improve the precision.

Thanks. Jennifer

[Star Strider]

My pleasure.

There was a recent discussion on this very topic: Failure in a simple division. You are likely dealing with floating-point approximation error that in this instance is also likely due to the propagation of such errors through many calculations. It is inherent in binary calculations, and there is no escaping it. Some calculations that involve only integers will not have floating-point conversion error, but most calculations will.