The BigDecimal of accurate Java computing

The computation in Java is mainly double,float,int,long,bigdecimal

1, float and double major users scientific and engineering calculations , which perform binary floating-point operations, are designed to provide more accurate, fast approximation calculations over a wide range of numerical ranges. However, they do not provide completely accurate results, so they should not be used in situations where precise calculations are required.

eg

Double Double_price1 = 1.03;
Double double_price2 = 0.42;
SYSTEM.OUT.PRINTLN ("The computed value of double:" + (DOUBLE_PRICE1-DOUBLE_PRICE2));

Theoretical value: 0.61

Execution result: Double calculated value: 0.6100000000000001

Comparing theoretical values and execution results, it is found that a double calculation results in an error.

2, the accurate calculation mainly uses: BigDecimal, int or long.

2.1 Disadvantages of using BigDecimal:

A. Compared to the basic type, the calculation of BigDecimal is very inconvenient, need to use some methods of the BigDecimal class, can not directly use the +-*/and other mathematical operators;

B. The efficiency of implementation is slow;

2.2 Disadvantages of using int, long: You need to handle the decimal point yourself.

3, BigDecimal need to pay attention to the problem:

3.1 New BigDecimal (String val) is better than new BigDecimal (double val)

eg

BigDecimal Bigdecimal_price1 = new BigDecimal (1.03);
System.out.println ("bigdecimal_price1=" +bigdecimal_price1);

Execution Result: bigdecimal_price1=1.0300000000000000266453525910037569701671600341796875

Analysis: Because of the construction method used here: The new BigDecimal (double val) Double is affected by the precision, so the constructed BigDecimal object is biased from the theoretical value.

BigDecimal bigdecimal_price3 = new BigDecimal ("1.03");
System.out.println ("bigdecimal_price3=" +bigdecimal_price3);

Execution result: bigdecimal_price3=1.03 is not biased with theoretical expectation.

3.2 When using the Divede () method, you need to pay attention to precision.

Eg1:divide (BigDecimal divisor)

System.out.println (New BigDecimal ("ten"). Divide (New BigDecimal ("3"));

Execution Result: Throw exception

java.lang.arithmeticexception:non-terminating decimal expansion; No exact representable decimal result.

Analysis:

The result of 10/3 is an infinite decimal, exceeding the bigdecimal limit.

Eg2:divide (BigDecimal divisor, int roundingmode)

System.out.println ("0.08/12=" +new BigDecimal ("0.08"). Divide (New BigDecimal ("12"), 8));

Execution Result: Throw exception

java.lang.arithmeticexception:non-terminating decimal expansion; No exact representable decimal result.

System.out.println ("0.08/12=" +new BigDecimal ("0.08"). Divide (New BigDecimal ("12"), 6));

Execution result: 0.08/12=0.01 and expected result error

Why keep 8 decimal places will be abnormal, and retain 6 decimal places can not be abnormal, I because of limited experience, failed to find out the answer, I hope you are not hesitate to enlighten the veteran.

Eg3:divide (BigDecimal divisor, int scale, Roundingmode roundingmode)

System.out.println ("0.08/12=" +new BigDecimal ("0.08"). Divide (New BigDecimal ("n"), 8,bigdecimal.round_half_up));

Execution Result: 0.08/12=0.00666667 is consistent with expected results

The BigDecimal of accurate Java computing