On Mon, Feb 20, 2012 at 9:01 AM, Harold E. <[email protected]> wrote: > I'm not sure to be able to fix issue 1941, and even in this case I > don't wish to work the whole summer on a technical point (and I was > hoping that with time there would be something new, like the new core, > to avoid this problem). The examples of the current Unit class show > that it can combine together, so I was thinking to try to use the same > mechanism if possible (because now I had ideas to simplify the > operations between units). In the worst case if it's not possible, I > was thinking to collect explicitly unit objects in the operations, but > it's not very proper.
OK, that's fine too. > > Concerning my approach to the units: during my previous module, I have > implemented a way of defining unit systems (by giving some base units, > like meter, kilogram and second in MKS), and to tell to sympy which > one should be used to do computations. And after some thoughts, I > found an improvement of this, because it was not easy to change > slightly a system without redefining a whole object, e.g. if you want > to use meter/gram/second instead of MKS, or the natural system. > The main idea is the same, that a system can be defined with units > which form the basis, but also with constants (like the speed of > light). Then, you can change a system by simply replacing some units. > But here, the main simplification is to use matrix representations for > all the computations, because a unit system can be defined as the > identity matrix, and the basis units as vectors, and then another > system is just another matrix (different from identity so). And > changing the system is just changing the basis of the space. Then we > can easily convert general units, which are general vectors in this > representation. Moreover, this representation permits us to check if > an unit system is well defined (by computing the rank of the matrix, > etc.). > Concerning the use of constants as units, it's still not very clear in > my mind, but I think it would be possible to define an option to hide > it or not when printing the unit, and also to try to determine which > are the constants to add in an expression to recover the right one > (for example E = m to E = mc²). > All this let us to simplify the big mess I did concerning the various > unit classes, since I defined BaseUnit, DerivedUnit, PrefixedUnit, > etc. Now, since these notions depend on the system, it's not useful > anymore to mare a distinction (but I'm still not sure how to define > the prefix). How does this work if you have units that are not related to each other linearly, like Celsius and Fahrenheit? Also, how easy would it be to define your own base units (not depending on already existing physical quantities)? As I remember, one of the strengths of your dictionary idea was that doing this would be easy. > > Concerning the other parts of physics, I don't have yet very specific > ideas of what I want to do, but my strong courses was all the kind of > field theories (including GR, QM...), so for example the idea named > "All this let us to simplify the big mess I did concerning the various > unit classes" could be very fine. > I was also thinking to write some modules of statistical mechanics/ > thermodynamics (gases...), but I don't know if it interest you (in my > case, I need this for a project of modelling planetary systems and > atmospheres). > > Harold Others will have to comment as far as this goes. Aaron Meurer > > On 19 fév, 19:39, Aaron Meurer <[email protected]> wrote: >> Cool. I'd like to hear more about what you think should be done for the >> units. >> >> Do you think it should involve issue 1941? That might be enough to >> fill a project alone, depending on how you do it (this could be pretty >> tricky, as it involves changing the core). >> >> Or if you want to focus the rest on other parts of physics, what >> specifically would you want to do? >> >> Aaron Meurer >> >> >> >> >> >> >> >> On Sun, Feb 19, 2012 at 2:04 PM, Harold E. <[email protected]> wrote: >> > Hi, >> >> > I'm thinking to apply in GSoC to work on sympy, but yet I'm still >> > hesitating, because at first I wanted to not work this summer to have >> > a break before beginning the PhD. >> > Ondrej advised me to post anyway in order to define an outline for a >> > project: the only thing I wished to do is to continue the improvements >> > of the units modules on which I began to work several months ago (see >> > #1940 on Google bugtracker); I was thinking a lot of time on this, so >> > I can explain how I see this. In addition to this point, I can work on >> > the physics module, e.g. QM. >> > In order to present me a little (I will also fill the application >> > template on the wiki later), I have studied theoretical physics in the >> > University Paris 7, in the École Normale Supérieure (Paris) and in the >> > Perimeter Institute for Theoretical Physics. My PhD subject will >> > certainly be on the adS/CFT holography applied to QCD. I'm using >> > Python since four years and I began to write a book on it. >> >> > Best regards, >> >> > Harold >> >> > -- >> > You received this message because you are subscribed to the Google Groups >> > "sympy" group. >> > To post to this group, send email to [email protected]. >> > To unsubscribe from this group, send email to >> > [email protected]. >> > For more options, visit this group >> > athttp://groups.google.com/group/sympy?hl=en. > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/sympy?hl=en. > -- You received this message because you are subscribed to the Google Groups "sympy" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/sympy?hl=en.
