ABAP linear/nonlinear programming.

[Former Member]

Hi Experts,

Is ABAP supporting linear/ nonlinear programming.

If yes, please tell me the method and how can I solve the below problem using linear/nonlinear programming.

I have solved it by normal ABAP but it is very slow.

I have below two table's values.

1st Table values.

51

82

143

2nd Tables value is

1007

I want to consume the full value 1007 based on multiple combinations of 1st table's values.

I want the set of combinations results as below:

Combination 1.            51 x  19 = 969(consume), 38 (Balance)

Combination 2.            51 x 18 + 82 x 1 = 1000(Consume), 7 (Balance)

Combination 3.            51 x 16 + 81 x 2 = 980(Consume), 27 (Balance)

Combination 4.            51 x 16 + 143 x 1 = 959(Consume), 48 (Balance)

Combination 5.            51 x 15 + 82 x 1 + 143 x 1 = 990(Consume), 17 (Balance)

And so on.....

Thanks,

[hendrik_brandes]

Hello Narendra,

there is a standard-framework for such operations-research algorithm.

Look in this very detailed blog: http://scn.sap.com/community/abap/blog/2011/10/04/operations-research-abap

Have fun!

Kind regards,

Hendrik

[Former Member]

Hello Hendrik,

Thanks for reply, I know about this research technique but I am not able to use it because I don't know about this programming.

I don't know how to use it in my scenario. will you please provide me some sample code based on my scenario with my data.

Thanks

Narendra.

[Former Member]

hi,

can anyone solve my problem. is there any way to solve this problem.

[Former Member]

who worked on abap linear/nonlinear programming, plz help.

[murtuza_chhatariya]

TYPES: BEGIN OF ty_values,
        count   TYPE i,
        value1  TYPE i,
        multi1  TYPE i,
        value2  TYPE i,
        multi2  TYPE i,
        value3  TYPE i,
        multi3  TYPE i,
        result  TYPE i,
        balance TYPE i,
       END OF ty_values.
PARAMETERS: p_1 TYPE i DEFAULT 51,
            p_2 TYPE i DEFAULT 82,
            p_3 TYPE i DEFAULT 143,
            p_4 TYPE i DEFAULT 1007.
DATA: lt_tab TYPE TABLE OF ty_values,
      wa_tab TYPE ty_values,
      lv_max_comb_1 TYPE i,
      lv_max_comb_2 TYPE i,
      lv_max_comb_3 TYPE i,
      lv_max_comb   TYPE i,
      lv_count      TYPE i,
      lv_count1     TYPE i,
      lv_count2     TYPE i,
      lv_count3     TYPE i,
      lv_result     TYPE i,
      lv_balance    TYPE i,
      lv_next       TYPE i.
START-OF-SELECTION.
  lv_count  = 0.
  lv_next   = 1.
  lv_count1 = 0.
  lv_count2 = 0.
  lv_count3 = 0.
  DO.
    CLEAR: lv_result, lv_balance.
    lv_result =   p_1 * lv_count1
                + p_2 * lv_count2
                + p_3 * lv_count3.
    lv_balance = p_4 - lv_result.
    IF lv_balance < 0.
      IF lv_next = 4.
        EXIT.
      ELSE.
        lv_next = lv_next + 1.
        CASE lv_next.
          WHEN 2.
            lv_count2 = lv_count2 + 1.
            lv_count1 = 0.
          WHEN 3.
            lv_count3 = lv_count3 + 1.
            lv_count2 = 0.
            lv_count1 = 0.
        ENDCASE.
        CONTINUE.
      ENDIF.
    ELSE.
      wa_tab-count  = lv_count.
      wa_tab-value1 = p_1.
      wa_tab-value2 = p_2.
      wa_tab-value3 = p_3.
      wa_tab-multi1 = lv_count1.
      wa_tab-multi2 = lv_count2.
      wa_tab-multi3 = lv_count3.
      wa_tab-result  = lv_result.
      wa_tab-balance = lv_balance.
      APPEND wa_tab TO lt_tab.
      CLEAR wa_tab.
      lv_count  = lv_count  + 1.
      lv_count1 = lv_count1 + 1.
    ENDIF.
  ENDDO.
  SORT lt_tab BY balance.
  lv_count = 0.
  LOOP AT lt_tab INTO wa_tab.
    lv_count = lv_count + 1.
    WRITE: / 'Combination ', lv_count, ': ',
       p_1, ' x ', wa_tab-multi1, ' + ',
       p_2, ' x ', wa_tab-multi2, ' + ',
       p_3, ' x ', wa_tab-multi3, ' = ', wa_tab-result,
       ' Balance: ', wa_tab-balance.
  ENDLOOP.

[Former Member]

We tried to do some of this about 3 years ago and could not find a good way to do it either.  At that time the APO landscape had some features for linear programming.  You may want to see if it has advanced since then.

Regards,

Steve

I was wrong it was four years ago.  Here's the thread in case it helps.

http://scn.sap.com/thread/752485