Originally posted by: Calvin Sun Hainan

When I do a calculation for Math.ceil(a/(b*c)) within OPL script, it rounds up the result incorrectly.

If I print the result for a/(b*c) - Math.floor(a/(b*c)), I can see this value: 2.273736754432321e-013.

Within OPL script, we can only define var type and I guess the exact value of my float number "b" has been modified before it is passed to Math.ceil function. Does anyone have similar problem before. How do I solve this problem?

I can write something like if a/(b*c) - Math.floor(a/(b*c)) < 1-e010, then do not use Math.ceil(x), but that is not so nice.

Hi,

can you post a small example ?

regards

Originally posted by: Calvin Sun Hainan

CRF data:

#SECURITY;Security_ID;Security_Description;Unit_Price_USD
SECURITY;"ISIN:EUR";"EUR";1.37340000

I used Double.parseDouble() to parse in the above Unit_Price_USD and then store it as a double value in java. In Java if I print the value out with ceil or floor it is OK. Then I use the java function provided by IBM IloOplDataHandler.addNumItem(Unit_Price_USD) to read the above value 1.3734 into OPL. It seems that the problem happens here when it converts the double value to float. In OPL, I have the following tuple:

tuple TSecurity {key string securityId;string securityDescription;float unitPriceUsd;};

When I try to do floor(unitPriceUsd * 10000) within OPL script, it gives me 13733.

If I do a writeln (Math.ceil(s.unitPriceUsd * 10000) - s.unitPriceUsd * 10000); in OPL script. It gives me the following strange value.

1.818989403545857e-012

 It looks to me OPL script is using 32-bit precision here.

Hi

yes you re right the scripting uses 32 bits. So you should try to do the computations in the modeling part.

int i=1234891261234;
int j=i+1;

execute
{
j;
i;
writeln("ok")
}

execute
{
j=i+1;
i;
writeln("ok2")
}

displays ok and then fails since i+1 is valid in OPL but not in scripting

regards

Originally posted by: Calvin Sun Hainan

OK. Thanks. I will try to find a way to deal with it. Is there any plan to enhance OPL script to use 64 bits? OPL script is really useful as it can help write logics in a very shor time.

Hi,

do not hesitate to log a Request For Enhancement at https://www.ibm.com/developerworks/rfe/execute?use_case=changeRequestLanding&BRAND_ID=401&PROD_ID=1290&x=13&y=7

regards

Originally posted by: Baby Bone

I'm having problems with OPL .mod file not only with Math.floor, (e.g., I tested even the most simple Math.floor(5.3), and it returned a 0. 

The reason I used Math.floor was actually because I needed the % function (to find the remainder), but it didn't work. It's returning a 0 no matter what the inputs are. Does any know why this might be the case? I'm using v12.8 on my Mac. 

Hi

execute
{
var a=Math.floor(5.3);
writeln(a);
}

gives

5

regards