To Generate Random Numbers in Ms Sql server [duplicate] - sql

I need a different random number for each row in my table. The following seemingly obvious code uses the same random value for each row.
SELECT table_name, RAND() magic_number
FROM information_schema.tables
I'd like to get an INT or a FLOAT out of this. The rest of the story is I'm going to use this random number to create a random date offset from a known date, e.g. 1-14 days offset from a start date.
This is for Microsoft SQL Server 2000.

Take a look at SQL Server - Set based random numbers which has a very detailed explanation.
To summarize, the following code generates a random number between 0 and 13 inclusive with a uniform distribution:
ABS(CHECKSUM(NewId())) % 14
To change your range, just change the number at the end of the expression. Be extra careful if you need a range that includes both positive and negative numbers. If you do it wrong, it's possible to double-count the number 0.
A small warning for the math nuts in the room: there is a very slight bias in this code. CHECKSUM() results in numbers that are uniform across the entire range of the sql Int datatype, or at least as near so as my (the editor) testing can show. However, there will be some bias when CHECKSUM() produces a number at the very top end of that range. Any time you get a number between the maximum possible integer and the last exact multiple of the size of your desired range (14 in this case) before that maximum integer, those results are favored over the remaining portion of your range that cannot be produced from that last multiple of 14.
As an example, imagine the entire range of the Int type is only 19. 19 is the largest possible integer you can hold. When CHECKSUM() results in 14-19, these correspond to results 0-5. Those numbers would be heavily favored over 6-13, because CHECKSUM() is twice as likely to generate them. It's easier to demonstrate this visually. Below is the entire possible set of results for our imaginary integer range:
Checksum Integer: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Range Result: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 2 3 4 5
You can see here that there are more chances to produce some numbers than others: bias. Thankfully, the actual range of the Int type is much larger... so much so that in most cases the bias is nearly undetectable. However, it is something to be aware of if you ever find yourself doing this for serious security code.

When called multiple times in a single batch, rand() returns the same number.
I'd suggest using convert(varbinary,newid()) as the seed argument:
SELECT table_name, 1.0 + floor(14 * RAND(convert(varbinary, newid()))) magic_number
FROM information_schema.tables
newid() is guaranteed to return a different value each time it's called, even within the same batch, so using it as a seed will prompt rand() to give a different value each time.
Edited to get a random whole number from 1 to 14.

RAND(CHECKSUM(NEWID()))
The above will generate a (pseudo-) random number between 0 and 1, exclusive. If used in a select, because the seed value changes for each row, it will generate a new random number for each row (it is not guaranteed to generate a unique number per row however).
Example when combined with an upper limit of 10 (produces numbers 1 - 10):
CAST(RAND(CHECKSUM(NEWID())) * 10 as INT) + 1
Transact-SQL Documentation:
CAST(): https://learn.microsoft.com/en-us/sql/t-sql/functions/cast-and-convert-transact-sql
RAND(): http://msdn.microsoft.com/en-us/library/ms177610.aspx
CHECKSUM(): http://msdn.microsoft.com/en-us/library/ms189788.aspx
NEWID(): https://learn.microsoft.com/en-us/sql/t-sql/functions/newid-transact-sql

Random number generation between 1000 and 9999 inclusive:
FLOOR(RAND(CHECKSUM(NEWID()))*(9999-1000+1)+1000)
"+1" - to include upper bound values(9999 for previous example)

Answering the old question, but this answer has not been provided previously, and hopefully this will be useful for someone finding this results through a search engine.
With SQL Server 2008, a new function has been introduced, CRYPT_GEN_RANDOM(8), which uses CryptoAPI to produce a cryptographically strong random number, returned as VARBINARY(8000). Here's the documentation page: https://learn.microsoft.com/en-us/sql/t-sql/functions/crypt-gen-random-transact-sql
So to get a random number, you can simply call the function and cast it to the necessary type:
select CAST(CRYPT_GEN_RANDOM(8) AS bigint)
or to get a float between -1 and +1, you could do something like this:
select CAST(CRYPT_GEN_RANDOM(8) AS bigint) % 1000000000 / 1000000000.0

The Rand() function will generate the same random number, if used in a table SELECT query. Same applies if you use a seed to the Rand function. An alternative way to do it, is using this:
SELECT ABS(CAST(CAST(NEWID() AS VARBINARY) AS INT)) AS [RandomNumber]
Got the information from here, which explains the problem very well.

Do you have an integer value in each row that you could pass as a seed to the RAND function?
To get an integer between 1 and 14 I believe this would work:
FLOOR( RAND(<yourseed>) * 14) + 1

If you need to preserve your seed so that it generates the "same" random data every time, you can do the following:
1. Create a view that returns select rand()
if object_id('cr_sample_randView') is not null
begin
drop view cr_sample_randView
end
go
create view cr_sample_randView
as
select rand() as random_number
go
2. Create a UDF that selects the value from the view.
if object_id('cr_sample_fnPerRowRand') is not null
begin
drop function cr_sample_fnPerRowRand
end
go
create function cr_sample_fnPerRowRand()
returns float
as
begin
declare #returnValue float
select #returnValue = random_number from cr_sample_randView
return #returnValue
end
go
3. Before selecting your data, seed the rand() function, and then use the UDF in your select statement.
select rand(200); -- see the rand() function
with cte(id) as
(select row_number() over(order by object_id) from sys.all_objects)
select
id,
dbo.cr_sample_fnPerRowRand()
from cte
where id <= 1000 -- limit the results to 1000 random numbers

select round(rand(checksum(newid()))*(10)+20,2)
Here the random number will come in between 20 and 30.
round will give two decimal place maximum.
If you want negative numbers you can do it with
select round(rand(checksum(newid()))*(10)-60,2)
Then the min value will be -60 and max will be -50.

try using a seed value in the RAND(seedInt). RAND() will only execute once per statement that is why you see the same number each time.

If you don't need it to be an integer, but any random unique identifier, you can use newid()
SELECT table_name, newid() magic_number
FROM information_schema.tables

You would need to call RAND() for each row. Here is a good example
https://web.archive.org/web/20090216200320/http://dotnet.org.za/calmyourself/archive/2007/04/13/sql-rand-trap-same-value-per-row.aspx

The problem I sometimes have with the selected "Answer" is that the distribution isn't always even. If you need a very even distribution of random 1 - 14 among lots of rows, you can do something like this (my database has 511 tables, so this works. If you have less rows than you do random number span, this does not work well):
SELECT table_name, ntile(14) over(order by newId()) randomNumber
FROM information_schema.tables
This kind of does the opposite of normal random solutions in the sense that it keeps the numbers sequenced and randomizes the other column.
Remember, I have 511 tables in my database (which is pertinent only b/c we're selecting from the information_schema). If I take the previous query and put it into a temp table #X, and then run this query on the resulting data:
select randomNumber, count(*) ct from #X
group by randomNumber
I get this result, showing me that my random number is VERY evenly distributed among the many rows:

It's as easy as:
DECLARE #rv FLOAT;
SELECT #rv = rand();
And this will put a random number between 0-99 into a table:
CREATE TABLE R
(
Number int
)
DECLARE #rv FLOAT;
SELECT #rv = rand();
INSERT INTO dbo.R
(Number)
values((#rv * 100));
SELECT * FROM R

select ABS(CAST(CAST(NEWID() AS VARBINARY) AS INT)) as [Randomizer]
has always worked for me

Use newid()
select newid()
or possibly this
select binary_checksum(newid())

If you want to generate a random number between 1 and 14 inclusive.
SELECT CONVERT(int, RAND() * (14 - 1) + 1)
OR
SELECT ABS(CHECKSUM(NewId())) % (14 -1) + 1

DROP VIEW IF EXISTS vwGetNewNumber;
GO
Create View vwGetNewNumber
as
Select CAST(RAND(CHECKSUM(NEWID())) * 62 as INT) + 1 as NextID,
'abcdefghijklmnopqrstuvwxyz0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ'as alpha_num;
---------------CTDE_GENERATE_PUBLIC_KEY -----------------
DROP FUNCTION IF EXISTS CTDE_GENERATE_PUBLIC_KEY;
GO
create function CTDE_GENERATE_PUBLIC_KEY()
RETURNS NVARCHAR(32)
AS
BEGIN
DECLARE #private_key NVARCHAR(32);
set #private_key = dbo.CTDE_GENERATE_32_BIT_KEY();
return #private_key;
END;
go
---------------CTDE_GENERATE_32_BIT_KEY -----------------
DROP FUNCTION IF EXISTS CTDE_GENERATE_32_BIT_KEY;
GO
CREATE function CTDE_GENERATE_32_BIT_KEY()
RETURNS NVARCHAR(32)
AS
BEGIN
DECLARE #public_key NVARCHAR(32);
DECLARE #alpha_num NVARCHAR(62);
DECLARE #start_index INT = 0;
DECLARE #i INT = 0;
select top 1 #alpha_num = alpha_num from vwGetNewNumber;
WHILE #i < 32
BEGIN
select top 1 #start_index = NextID from vwGetNewNumber;
set #public_key = concat (substring(#alpha_num,#start_index,1),#public_key);
set #i = #i + 1;
END;
return #public_key;
END;
select dbo.CTDE_GENERATE_PUBLIC_KEY() public_key;

Update my_table set my_field = CEILING((RAND(CAST(NEWID() AS varbinary)) * 10))
Number between 1 and 10.

Try this:
SELECT RAND(convert(varbinary, newid()))*(b-a)+a magic_number
Where a is the lower number and b is the upper number

If you need a specific number of random number you can use recursive CTE:
;WITH A AS (
SELECT 1 X, RAND() R
UNION ALL
SELECT X + 1, RAND(R*100000) --Change the seed
FROM A
WHERE X < 1000 --How many random numbers you need
)
SELECT
X
, RAND_BETWEEN_1_AND_14 = FLOOR(R * 14 + 1)
FROM A
OPTION (MAXRECURSION 0) --If you need more than 100 numbers

Related

Given a table of numbers, can I get all the rows which add up to less than or equal to a number?

Say I have a table with an incrementing id column and a random positive non zero number.
id
rand
1
12
2
5
3
99
4
87
Write a query to return the rows which add up to a given number.
A couple rules:
Rows must be "consumed" in order, even if a later row makes it a a perfect match. For example, querying for 104 would be a perfect match for rows 1, 2, and 4 but rows 1-3 would still be returned.
You can use a row partially if there is more available than is necessary to add up to whatever is leftover on the number E.g. rows 1, 2, and 3 would be returned if your max number is 50 because 12 + 5 + 33 equals 50 and 90 is a partial result.
If there are not enough rows to satisfy the amount, then return ALL the rows. E.g. in the above example a query for 1,000 would return rows 1-4. In other words, the sum of the rows should be less than or equal to the queried number.
It's possible for the answer to be "no this is not possible with SQL alone" and that's fine but I was just curious. This would be a trivial problem with a programming language but I was wondering what SQL provides out of the box to do something as a thought experiment and learning exercise.
You didn't mention which RDBMS, but assuming SQL Server:
DROP TABLE #t;
CREATE TABLE #t (id int, rand int);
INSERT INTO #t (id,rand)
VALUES (1,12),(2,5),(3,99),(4,87);
DECLARE #target int = 104;
WITH dat
AS
(
SELECT id, rand, SUM(rand) OVER (ORDER BY id) as runsum
FROM #t
),
dat2
as
(
SELECT id, rand
, runsum
, COALESCE(LAG(runsum,1) OVER (ORDER BY id),0) as prev_runsum
from dat
)
SELECT id, rand
FROM dat2
WHERE #target >= runsum
OR #target BETWEEN prev_runsum AND runsum;

SQL Server 2012 : update a row with unique number

I have a table with 50k records. Now I want to update one column of the table with a random number. The number should be 7 digits.
I don't want to do that with procedure or loop.
PinDetailId PinNo
--------------------
783 2722692
784 9888648
785 6215578
786 7917727
I have tried this code but not able to succeed. I need 7 digit number.
SELECT
FLOOR(ABS(CHECKSUM(NEWID())) / 2147483647.0 * 3 + 1) rn,
(FLOOR(2000 + RAND() * (3000 - 2000) )) AS rn2
FROM
[GeneratePinDetail]
Random
For a random number, you can use ABS(CHECKSUM(NewId())) % range + lowerbound:
(source: How do I generate random number for each row in a TSQL Select?)
INSERT INTO ResultsTable (PinDetailId, PinNo)
SELECT PinDetailId,
(ABS(CHECKSUM(NewId())) % 1000000 + 1000000) AS `PinNo`
FROM GeneratePinDetail
ORDER BY PinDetailId ASC;
Likely Not Unique
I cannot guarantee these will be unique; but it should be evenly distributed (equal chance of any 7 digit number). If you want to check for duplicates you can run this:
SELECT PinDetailId, PinNo
FROM ResultsTable result
INNER JOIN (
SELECT PinNo
FROM ResultsTable
GROUP BY PinNo
HAVING Count(1) > 1
) test
ON result.PinNo = test.PinNo;
You can create a sequence object and update your fields - it should automatically increment per update.
https://learn.microsoft.com/en-us/sql/t-sql/functions/next-value-for-transact-sql
Updated based on comment:
After retrieving the 'next value for' in the sequence, you can do operations on it to randomize. The sequence can basically be used then to create a unique seed for your randomization function.
If you don't want to create a function yourself, SQL Server has the RAND function build in already.
https://learn.microsoft.com/en-us/sql/t-sql/functions/rand-transact-sql

Giving Range to the SQL Column

I have SQL table in which I have column and Probability . I want to select one row from it with randomly but I want to give more chances to the more waighted probability. I can do this by
Order By abs(checksum(newid()))
But the difference between Probabilities are too much so it gives more chance to highest probability.Like After picking 74 times that value it pick up another value for once than again around 74 times.I want to reduce this .Like I want 3-4 times to it and than others and all. I am thinking to give Range to the Probabilies.Its Like
Row[i] = Row[i-1]+Row[i]
How can I do this .Do I need to create function?Is there any there any other way to achieve this.I am neewby.Any help will be appriciated.Thank You
EDIT:
I have solution of my problem . I have one question .
if I have table as follows.
Column1 Column2
1 50
2 30
3 20
can i get?
Column1 Column2 Column3
1 50 50
2 30 80
3 20 100
Each time I want to add value with existing one.Is there any Way?
UPDATE:
Finally get the solution after 3 hours,I just take square root of my probailities that way I can narrow the difference bw them .It is like I add column with
sqrt(sqrt(sqrt(Probability)))....:-)
I'd handle it by something like
ORDER BY rand()*pow(<probability-field-name>,<n>)
for different values of n you will distort the linear probabilities into a simple polynomial. Small values of n (e.g. 0.5) will compress the probabilities to 1 and thus make less probable choices more probable, big values of n (e.g. 2) will do the opposite and further reduce probability of already inprobable values.
Since the difference in probabilities is too great, you need to add a computed field with a revised weighting that has a more even probability distribution. How you do that depends on your data and preferred distribution. One way to do it is to "normalize" the weighting to an integer between 1 and 10 so that the lowest probability is never more than ten times smaller than the highest.
Answer to your recent question:
SELECT t.Column1,
t.Column2,
(SELECT SUM(Column2)
FROM table t2
WHERE t2.Column1 <= t.Column1) Column3
FROM table t
Here is a basic example how to select one row from the table with taking into account the assigned row weights.
Suppose we have table:
CREATE TABLE TableWithWeights(
Id int NOT NULL PRIMARY KEY,
DataColumn nvarchar(50) NOT NULL,
Weight decimal(18, 6) NOT NULL -- Weight column
)
Let's fill table with sample data.
INSERT INTO TableWithWeights VALUES(1, 'Frequent', 50)
INSERT INTO TableWithWeights VALUES(2, 'Common', 30)
INSERT INTO TableWithWeights VALUES(3, 'Rare', 20)
This is the query that returns one random row with taking into account given row weights.
SELECT * FROM
(SELECT tww1.*, -- Select original table data
-- Add column with the sum of all weights of previous rows
(SELECT SUM(tww2.Weight)- tww1.Weight
FROM TableWithWeights tww2
WHERE tww2.id <= tww1.id) as SumOfWeightsOfPreviousRows
FROM TableWithWeights tww1) as tww,
-- Add column with random number within the range [0, SumOfWeights)
(SELECT RAND()* sum(weight) as rnd
FROM TableWithWeights) r
WHERE
(tww.SumOfWeightsOfPreviousRows <= r.rnd)
and ( r.rnd < tww.SumOfWeightsOfPreviousRows + tww.Weight)
To check query results we can run it for 100 times.
DECLARE #count as int;
SET #count = 0;
WHILE ( #count < 100)
BEGIN
-- This is the query that returns one random row with
-- taking into account given row weights
SELECT * FROM
(SELECT tww1.*, -- Select original table data
-- Add column with the sum of all weights of previous rows
(SELECT SUM(tww2.Weight)- tww1.Weight
FROM TableWithWeights tww2
WHERE tww2.id <= tww1.id) as SumOfWeightsOfPreviousRows
FROM TableWithWeights tww1) as tww,
-- Add column with random number within the range [0, SumOfWeights)
(SELECT RAND()* sum(weight) as rnd
FROM TableWithWeights) r
WHERE
(tww.SumOfWeightsOfPreviousRows <= r.rnd)
and ( r.rnd < tww.SumOfWeightsOfPreviousRows + tww.Weight)
-- Increase counter
SET #count += 1
END
PS The query was tested on SQL Server 2008 R2. And of course the query can be optimized (it's easy to do if you get the idea)

How can I round a column in a single SQL request without changing the overall sum?

I've got a table defined like this :
create table #tbFoo
(bar float)
And I'm looking for a way to round every value contained in column bar without changing the total sum (which is known to be an integer, or very close to an integer because of float number precision).
Rounding every value to the nearest integer won't work (ex : 1,5;1,5 will be rounded to 1;1 or 2;2)
It's quite easy to do this using several requests (eg storing the original sum, rounding, computing the new sum, and updating as many rows as needed to go back to the original sum), but this is not a very elegant solution.
Is there a way to do this using a single SQL request?
I'm using SQL Server 2008, so solutions taking advantage of this specific vendor are welcome.
Edit : I'm looking for a request minimizing the differences between the old values and the new ones. In other words, a value should never be rounded up if a greater value has been rounded down, and vice-versa
Update:
See this solution explained in more details in the article in my blog:
Rounding numbers preserving their sum
You need to keep cumulative offset for each value:
1.2 (1 + 0.0) ~ 1 1 1.2 +0.2
1.2 (1 + 0.2) ~ 1 2 2.4 +0.4
1.2 (1 + 0.4) ~ 1 3 3.6 +0.6
1.2 (1 + 0.6) ~ 2 5 4.8 -0.2
1.2 (1 - 0.2) ~ 1 6 6.0 0.0
This is easily done in MySQL, but in SQL Server you will have to write a cursor or use cumulative subselects (which are less efficient).
Update:
The query below selects the difference between the sums of the values and of those rounded down to the nearest smaller integer.
This gives us the number (N) of values we should round up.
Then we order the values by their fractional part (ones that are closer to their ceiling go first) and round the first N up, the others down.
SELECT value,
FLOOR(value) + CASE WHEN ROW_NUMBER() OVER (ORDER BY value - FLOOR(value) DESC) <= cs THEN 1 ELSE 0 END AS nvalue
FROM (
SELECT cs, value
FROM (
SELECT SUM(value) - SUM(FLOOR(value)) AS cs
FROM #mytable
) c
CROSS JOIN
#mytable
) q
Here's the script for the test data:
SET NOCOUNT ON
GO
SELECT RAND(0.20090917)
DECLARE #mytable TABLE (value FLOAT NOT NULL)
DECLARE #cnt INT;
SET #cnt = 0;
WHILE #cnt < 100
BEGIN
INSERT
INTO #mytable
VALUES (FLOOR(RAND() * 100) / 10)
SET #cnt = #cnt + 1
END
INSERT
INTO #mytable
SELECT 600 - SUM(value)
FROM #mytable
If you have a list of n values whose elements are accurate only to within an integer value (+-0.5), then any sum of those elements will have a cumulative error or +-(n*0.5). If you have 6 elements in your list which should add up to some number, then your worst case scenario is that you're off by 3 if you just add the integer values.
If you find some way of showing 10.2 as 11 in order to make the sum work, you've changed the precision of that element from +-0.5 to +-0.8, which is counterintuitive when looking at integers?
One possible solution to think about is to round your number during display only (using some format string on your output), not already at the retrieval stage. Each number will be as close as possible to the actual value, but the sum will be more correct too.
Example: If you have 3 values of 1/3 each, displayed as whole-numbered percentages, then you should be showing 33, 33 and 33. To do anything else is to create a margin of error greater than +-0.5 for any individual value. Your total should still be displayed as 100%, because that is the best possible value (as opposed to working with sums of already rounded values)
Also, be aware that by using a float, you've already introduced a limitation on your precision because you have no way of accurately representing 0.1. For more on that, read
What Every Computer Scientist Should Know About Floating-Point Arithmetic
First get the difference between the rounded sum and the actual sum, and the number of records:
declare #Sum float, #RoundedSum float, #Cnt int
select #Sum = sum(bar), #RoundedSum = sum(round(bar)), #Cnt = count(*)
from #tbFoo
Then you spread the difference equally on all values before rounding:
declare #Offset float
set #Offset = (#Sum - #RoundedSum) / #Cnt
select bar = round(bar + #Offset)
from #tbFoo

Ranking in MySQL, how do I get the best performance with frequent updates and a large data set?

I want grouped ranking on a very large table, I've found a couple of solutions for this problem e.g. in this post and other places on the web. I am, however, unable to figure out the worst case complexity of these solutions. The specific problem consists of a table where each row has a number of points and a name associated. I want to be able to request rank intervals such as 1-4. Here are some data examples:
name | points
Ab 14
Ac 14
B 16
C 16
Da 15
De 13
With these values the following "ranking" is created:
Query id | Rank | Name
1 1 B
2 1 C
3 3 Da
4 4 Ab
5 4 Ac
6 6 De
And it should be possible to create the following interval on query-id's: 2-5 giving rank: 1,3,4 and 4.
The database holds about 3 million records so if possible I want to avoid a solution with complexity greater than log(n). There are constantly updates and inserts on the database so these actions should preferably be performed in log(n) complexity as well. I am not sure it's possible though and I've tried wrapping my head around it for some time. I've come to the conclusion that a binary search should be possible but I haven't been able to create a query that does this. I am using a MySQL server.
I will elaborate on how the pseudo code for the filtering could work. Firstly, an index on (points, name) is needed. As input you give a fromrank and a tillrank. The total number of records in the database is n. The pseudocode should look something like this:
Find median point value, count rows less than this value (the count gives a rough estimate of rank, not considering those with same amount of points). If the number returned is greater than the fromrank delimiter, we subdivide the first half and find median of it. We keep doing this until we are pinpointed to the amount of points where fromrank should start. then we do the same within that amount of points with the name index, and find median until we have reached the correct row. We do the exact same thing for tillrank.
The result should be log(n) number of subdivisions. So given the median and count can be made in log(n) time it should be possible to solve the problem in worst case complexity log(n). Correct me if I am wrong.
You need a stored procedure to be able to call this with parameters:
CREATE TABLE rank (name VARCHAR(20) NOT NULL, points INTEGER NOT NULL);
CREATE INDEX ix_rank_points ON rank(points, name);
CREATE PROCEDURE prc_ranks(fromrank INT, tillrank INT)
BEGIN
SET #fromrank = fromrank;
SET #tillrank = tillrank;
PREPARE STMT FROM
'
SELECT rn, rank, name, points
FROM (
SELECT CASE WHEN #cp = points THEN #rank ELSE #rank := #rn + 1 END AS rank,
#rn := #rn + 1 AS rn,
#cp := points,
r.*
FROM (
SELECT #cp := -1, #rn := 0, #rank = 1
) var,
(
SELECT *
FROM rank
FORCE INDEX (ix_rank_points)
ORDER BY
points DESC, name DESC
LIMIT ?
) r
) o
WHERE rn >= ?
';
EXECUTE STMT USING #tillrank, #fromrank;
END;
CALL prc_ranks (2, 5);
If you create the index and force MySQL to use it (as in my query), then the complexity of the query will not depend on the number of rows at all, it will depend only on tillrank.
It will actually take last tillrank values from the index, perform some simple calculations on them and filter out first fromrank values.
Time of this operation, as you can see, depends only on tillrank, it does not depend on how many records are there.
I just checked in on 400,000 rows, it selects ranks from 5 to 100 in 0,004 seconds (that is, instantly)
Important: this only works if you sort on names in DESCENDING order. MySQL does not support DESC clause in the indices, that means that the points and name must be sorted in one order for INDEX SORT to be usable (either both ASCENDING or both DESCENDING). If you want fast ASC sorting by name, you will need to keep negative points in the database, and change the sign in the SELECT clause.
You may also remove name from the index at all, and perform a final ORDER'ing without using an index:
CREATE INDEX ix_rank_points ON rank(points);
CREATE PROCEDURE prc_ranks(fromrank INT, tillrank INT)
BEGIN
SET #fromrank = fromrank;
SET #tillrank = tillrank;
PREPARE STMT FROM
'
SELECT rn, rank, name, points
FROM (
SELECT CASE WHEN #cp = points THEN #rank ELSE #rank := #rn + 1 END AS rank,
#rn := #rn + 1 AS rn,
#cp := points,
r.*
FROM (
SELECT #cp := -1, #rn := 0, #rank = 1
) var,
(
SELECT *
FROM rank
FORCE INDEX (ix_rank_points)
ORDER BY
points DESC
LIMIT ?
) r
) o
WHERE rn >= ?
ORDER BY rank, name
';
EXECUTE STMT USING #tillrank, #fromrank;
END;
That will impact performance on big ranges, but you will hardly notice it on small ranges.