Re: [boost] [units] - learning to use, more questions :)

Subject: Re: [boost] [units] - learning to use, more questions :)
From: Matthias Schabel (boost_at_[hidden])
Date: 2011-11-28 17:13:55

> And I've hit a brick wall, because I cannot get those lines to compile:
> 
>    quantity<reciprocal_area>   lambda ( 2.0*m*E / (hbar*hbar)    );
>    quantity<reciprocal_area>   alpha  ( m*omega/root<2>(hbar)    );
>    double                      lambda_per_alpha ( lambda / alpha );
> 
> Even the first one has problems, to my surprise. Maybe I really
> should change my g++ 4.4.5 version to a newer version? I will gladly try
> this if you think it makes sense.

The first line compiles correctly using Boost 1.48; I don't know when the fix was made. The second line there is a bug/missing feature; the constants don't define pow/root operations, although I don't understand exactly why the implicit conversion to the appropriate quantity type is not working. We will fix that in the next release... For the interim, you can use root<2>(hbar.value()), although I recognize that it is somewhat confusing since, in this context, value() returns a quantity, not a value_type.

However, you also have an error in the second line; alpha is not a reciprocal_area : kg * (1/s) / (kg m^2/s)^1/2 = kg^1/2 s^-1/2 m^-1. Of course, this makes the third line incorrect as well; an example of the power of Boost.Units, I guess...

In general, when debugging units, it is often helpful to output to the console. Use of auto for intermediate values is also your friend... For example, here's what I did to figure out your problem:

#include <iostream>
#include <complex>
using namespace std;
//use units
#include <boost/typeof/std/complex.hpp>
#include <boost/units/cmath.hpp>
#include <boost/units/systems/si.hpp>
#include <boost/units/systems/si/prefixes.hpp>
#include <boost/units/systems/si/codata/universal_constants.hpp>

#include <boost/units/io.hpp>

using namespace boost::units;
using namespace boost::units::si;
using namespace boost::units::si::constants::codata;
//use units end

typedef derived_dimension<length_base_dimension,-2>::type  	reciprocal_area_dimension;
typedef unit<reciprocal_area_dimension,si::system>		reciprocal_area;

int main()
{
        int					n = 5;
        quantity<mass>				m(1*kilogram);
        quantity<frequency>			omega(1*hertz);
        quantity<energy>			E(hbar*omega*(n+0.5));
        quantity<reciprocal_area>		lambda(2.0*m*E/(hbar*hbar));
        
        std::cout << m << std::endl;
        std::cout << omega << std::endl;
            std::cout << E << std::endl;        					
            std::cout << lambda << std::endl;   
//	std::cout << m*omega/root<2>(hbar) << std::endl;		// bug - this should work
        std::cout << m*omega/root<2>(hbar.value()) << std::endl;	// works but shouldn't be necessary and bad choice of syntax
        
            return 0;
}

output : 

1 kg
1 s^-1
5.80014e-34 m^2 kg s^-2
1.04308e+35 m^-2
9.73782e+16 m^-1 kg^(1/2) s^(-1/2)

Steven - since you implemented the constants.hpp header, I'm a little reluctant to rummage through it. Could you please consider

1) adding support for the pow<> and root<> operations?
2) deprecating constant.value_type and constant.value() in favor of constant.quantity_type and constant.quantity() or constant.mean_value() or constant.mean() or something?

Also, can you remind me why we have "constant" and "physical_constant" structs? They look identical to me...

Janek,

I have attached headers implementing the various prefixed short cuts for the meter and joule; these two should be sufficient to extend to the rest of the SI unit system. What remains to be done is to replicate scaled_meter.hpp/scaled_length.hpp for the remaining base units in the <boost/units/base_units/si> directory and to replicate scaled_energy.hpp for all the remaining named SI units in the <boost/units/systems/si> directory. 

Here's my test program:

#include <iostream>

#include <boost/units/io.hpp>
#include <boost/units/systems/si/length.hpp>
#include <boost/units/systems/si/energy.hpp>

#include "scaled_length.hpp"

#include "scaled_energy.hpp"

using namespace boost::units;
using namespace boost::units::si;

int main()
{
        std::cout << 1.0*nm << "\t" << 1.5*mJ << std::endl;
        
        quantity<length>	q1(1.0*nm);
        quantity<energy>	q2(1.5*mJ);
        
        std::cout << q1 << "\t" << q2 << std::endl;

        auto			q3(1.0*nm);
        auto			q4(1.5*mJ);
        
        std::cout << q3 << "\t" << q4 << std::endl;
        
        return 0;
}

and output :

1 nm	1.5 mJ
1e-09 m	0.0015 m^2 kg s^-2
1 nm	1.5 mJ

Note: if you assign these scaled quantities to unscaled quantities, the scaling information will be lost (second output line)...

Matthias