Float

Name	Description	Storage Space	Value Range	Literal
REAL	Real number	32 bits	1.40129846432481707e-45 to 3.40282346638528860e+38, positive or negative	REAL
DOUBLE	Double-precision floating point number with 15 to 17 valid digits, depending on the application scenario. The number of valid digits does not depend on the decimal point.	64 bits	4.94065645841246544e-324 to 1.79769313486231570e+308, positive or negative	DOUBLE
FLOAT	Single-precision floating point number with 6 to 9 valid digits, depending on the application scenario. The number of valid digits does not depend on the decimal point.	32 bits	1.40129846432481707e-45 to 3.40282346638528860e+38, positive or negative	FLOAT

Usage:

When a distributed query uses high-performance hardware instructions to perform single-precision or double-precision computing, the computing result may be slightly different because the execution sequence is different each time when an aggregate function, such as SUM() or AVG(), is invoked. Especially when the data volume is large (tens of millions or even billions of records), the computing result may be slightly different. In this case, you are advised to use the DECIMAL data type.

An alias can be used to specify the data type.

Example:

--Create a table that contains float data.
CREATE TABLE float_t1 (float_col1 FLOAT) ;
--Insert data of the float type.
insert into float_t1 values (float '3.50282346638528862e+38');
--View data.
SELECT * FROM float_t1;
float_col1
------------
Infinity
(1 row)
--Drop the table.
DROP TABLE float_t1;

When the decimal part is 0, you can use cast() to convert the decimal part to an integer of the corresponding range. The decimal part is rounded off.
Example:
```
select CAST(1000.0001 as INT);
_col0
-------
1000
(1 row)
select CAST(122.5001 as TINYINT);
_col0
-------
123
(1 row)
```

When an exponential expression is used, the string can be converted to the corresponding type.
Example:
```
select CAST(152e-3 as double);
_col0
-------
0.152
(1 row)
```

Parent topic: Data Type

Previous topic: Fixed Precision

Next topic: Character