Re: [MIT-Scheme-users] Constructing a differential function with ScmUtils

[David Gray]

> On 9 Mar 2017, at 11:27, Taylor R Campbell  wrote:
> 
> pe ((D (n (up 'p_1 'p_2 'p_3 ...))) x))
Brilliant, definitely better. I forgot about using tuples. Now it's:
 (define bk7 (up 1.03961212 6.00069867e-3 0.231792344 2.00179144e-2 1.01046945 
1.03560653e2))
((D (n bk7)) 0.8) 

___
MIT-Scheme-users mailing list
MIT-Scheme-users@gnu.org
https://lists.gnu.org/mailman/listinfo/mit-scheme-users

Re: [MIT-Scheme-users] Constructing a differential function with ScmUtils

[Taylor R Campbell]

> Date: Thu, 9 Mar 2017 10:55:07 +0200
> From: David Gray 
> 
> I'm going to deal with a few materials so I thought passing a vector
> directly would be better, but I didn't manage to
> differentiate the resulting function

If you use the autocurrying notation, it may be easier:

(define ((n p) x)
  ...)

(pe ((D (n (up 'p_1 'p_2 'p_3 ...))) x))

Otherwise, you need *all* the arguments to go through a single vector
in order to differentiate it, because scmutils has no way to know
which arguments are scalars and which arguments are vectors.

___
MIT-Scheme-users mailing list
MIT-Scheme-users@gnu.org
https://lists.gnu.org/mailman/listinfo/mit-scheme-users

Re: [MIT-Scheme-users] Constructing a differential function with ScmUtils

[David Gray]

I’m going to deal with a few materials so I thought passing a vector directly 
would be better, but I didn’t manage to differentiate the resulting function 
e.g.: (define (n parameter-vector x)
 …….)
unless I explicitly passed each constant separately. 
___
MIT-Scheme-users mailing list
MIT-Scheme-users@gnu.org
https://lists.gnu.org/mailman/listinfo/mit-scheme-users

Re: [MIT-Scheme-users] Constructing a differential function with ScmUtils

[Taylor R Campbell]

> Date: Thu, 9 Mar 2017 10:21:10 +0200
> From: David Gray 
> 
> I'm doing some calculations for GVD in various dielectrics, and I
> thought to ask here if anyone knows if there is a less clunky method
> of constructing the differential function with a list of constants:

What do you find clunky about this?  Would you rather pass in a vector
of constants?  Would you rather use the autocurrying notation to avoid
naming n and nofx separately?

(define ((n p1 p2 p3 p4 p5 p6) l)
  (define (d t b l2)
(/ (* t l2) (- l2 b)))
  (let ((l2 (square l)))
(sqrt (+ 1 (d p1 p2 l2) (d p3 p4 l2) (d p5 p6 l2)
(define n_symbolic (n 'p1 'p2 'p3 'p4 'p5 'p6))
(define n_numeric (n 1.03961212 ...))
(pe ((D n_symbolic) 'x))
((D n_numeric) 0.8)

___
MIT-Scheme-users mailing list
MIT-Scheme-users@gnu.org
https://lists.gnu.org/mailman/listinfo/mit-scheme-users