For quick calculations I can do this on the linux command line: ...but when it comes to floating point I get C style integer operations and results. Is there a way to specify to use float or double?

I have seen that I can use bc (which is a powerful and very easy to use) :but still I wonder if bash provides a way to cast or print the result in floating point.

No.

Regards,

Dave

Footnote: I'm thinking that bc is probably the "most popular" way, but you could also use GNU octave from bash (or other shell):

Ahh yes octave.
wow
So I see that 22/7 is not that close as you said back in December

Now I wonder what a(1) is... and 4* ???

man bc:
MATH LIBRARY
a (x) The arctangent of x, arctangent returns radians.

boy I'm gonna have to get out a math book...

Thanks dave

I threw that in just for the heck of it because you expressed an interest in octave some time ago. GNU octave uses 64-bit floating point double precision variables, good for about 16 significant decimal digits. The bc program is a terrific "arbitrary precision" machine that can give as much precision as you specify with the "scale=..." statement.

Refresh me: What the heck did I say back in December? (I mean, last Tuesday is ancient history to me; last December? Oh, man...)

Here's what I say now:

22/7 is an infinitely repeating decimal fraction. If you round off 22/7 to four significant decimal digits you get

3.143


If you round off pi to four significant decimal digits you get

3.142


If you round off 22/7 to three significant digits you get

3.14


If you round off pi to three significant digits you get

3.14


My conclusion is that 22/7 is an approximation to pi that is correct to three significant decimal digits.


Here's the reasoning behind that little jewel:

The arctangent function returns an angle whose tangent is given by the argument. In fact there are an infinite number of angles whose tangent is a given value. Most program implementations of the atan function return a value between -pi/2 and pi/2.
See, for example Inverse Tangent

Anyhow...

The tangent of pi/4 radians is 1

Therefore the arctangent of 1 is pi/4

Therefore, pi is equal to 4 times the arctangent of 1.

Of course, the arctangent function calculates an approximate value using whatever precision is available, so an approximation of pi can be obtained by multiplying arctangent(1) by four. Note that there is slight difference (two units in the 21st decimal digit) between pi and the approximation calculated by (4 times arctangent(1)). This is due to roundoff error/truncation error in the arctangent function evaluation being propagated to the product.


Regards,

Dave

I hear that! , It was just this thread to nowhere...Ring a bell?

re: octave
Yes, Thanks for the example. I was getting into it and then it got cold! ("what the heck?" you say)
Short story , This is an old farm house, we are using a wood stove exclusively. My fedora box is in the "Outback".
I shut the door a few months ago and used the my old laptop with linuxfromscratch in a warm room.
I didn't want to fool with building all the stuff for octave....
Now I'm back in here, but so many things to do...

I did build tlc and tk and learned to make a graphical plotting program though!!!: )
Very interesting to use tlc and see what the pro's do with object oriented programming.

Now that issss interesting.
This would be that "bank" quality numeric handling you have talked about in the past.

re: a (x)
What you explain makes sense and I almost understand.
What I really need to do in that department is start back at algebra and trig 1 and learn math!
I find I understand it better now than in high school ,, but still , a long way to go.

Thanks, Howard();

In that thread I said that 22/7 was correct to 3 significant decimal digits. What makes you say that your results are "not as close" as what I said?

What a terrific project! Regardless of the value of learning to build the kernel and tools and file system (and so on) from scratch, learning what actually happens when you boot up into Linux is simply priceless. It is also mind-boggling. (I'm thinking that this kind of boggling is a Good Thing, since it doesn't involve pharmaceuticals. Or as Archie McNally says from time to time, "It left me with my flabber totally gasted.") That's one time that I bought the book (a year or so ago) even though Open Source documentation is available.

Well, to me "bank" quality arithmetic must use fixed point decimal arithmetic. You simply can't guarantee results down to the penny with binary/floating point arithmetic. Or, putting it another way, even if someone "guarantees" it, I'm not going for it.

Decimal arithmetic is available natively in COBOL, and I know that there are BCD class libraries for C++. Or---you can roll your own arbitrary precision decimal arithmetic package. How tough could that be for people who build Linux from scratch?


Regards,

Dave