Audit Software limitations

I have received numerous inquiries about any system limitations. First, the design intent is for use in business applications, as opposed to science or engineering (although it might be used for those areas). There is no intentional "crippling" by system design, i.e. a limitation designed to limit the program's usefulness. I am currently in the process of testing, and removing where possible, any system limitations found which can be removed at a reasonable cost. The design target is to be able to accurately handle dollar amounts into the trillions. However, currently there can be "rounding errors" or other limitations. The table below shows the current system limitations, as best I have been able to determine. This page is a "work in progress" and will be updated as I obtain more information.

I have not been informed by anyone that limitations have been encountered on the more typical situations such as spreadsheets and ordinary size files. I have tested some of the applications as documented in the "benchmarks" section containing up to 80,000,000 rows with no exceptions noted.

**Known system limitations**
Description	Known Limitation	Intention to fix?	Fix Date
Population statistics	Amounts less than -920 trillion and more than +920 trillion are not processed (i.e. ignored). Any amounts containing fractional pennies are rounded to the nearest 1/100 of a cent. For amounts without fractional pennies, results should be accurate to the penny for population total amounts of up to $10 trillion. Between 10 trillion and 100 trillion, rounding errors totaling 5 cents can occur. For amounts between 100 trillion and 920 trillion the extent of rounding errors is unknown, but thought to be less than $1 total. However, amounts with fractional pennies could result in rounding errors of up to $5 per 100,000 transactions. The maximum number of records which can be processed in a single run is approximately 500,000,000.	Yes	2-3-2008
Ageing, summarization and data stratification	These all have the same limitations as described above, except the only limit on the number of rows is that imposed by the operating system.	Yes	2-4-2008
Benford's Law, Weekdays, holidays	Counts have a maximum of approximately 2,000,000,000 each.	No	N/A
Numeric rounding - other (e.g. CMA sampling, Benford array)	Use of floating point arithmetic results in rounding errors in some instances.	Yes	3-22-2008
Date handling	Only handles dates between AD 100 and AD 9999.	No	N/A
Maximum number of rows of data	Operating system limits only - should handle files at least 2 GB in size. For population statistics, limit is 500,000,000.	No	N/A
Maximum number of ageing bins	80	Yes
Maximum number of stratification bins	100	Yes
Largest number of zeros tested on a round number	14, e.g. 900,000,000,000,000	Yes	1-29-2008
Holiday testing	Limited to U.S. federal holidays only	No	N/A
Sequential numbers	Limit to nine digit numbers (intent is to expand to 14)	Yes
Top/Bottom "N" Report (extremes)	Extremes for up to the top 100 and bottom 100 items are provided (limited to 100).	No	N/A
Sorting	Maximum number of rows which can be sorted is approximately 2 billion in a single file. Sorting is done in "chunks" of up to 1,000,000 rows. Sort speed is dependent upon the number of sort keys and type (e.g. numeric, character or date). Sort method used is "heap sort". Sort speed of 5-10,000 rows per second is common.	No	N/A
Sorting	Maximum number of columns which can be specified as part of the sort criteria is 200. Numeric column values of less than -920 trillion or greater than +920 trillion are not handled. Date column values must be between AD 100 and AD 9999.	No	N/A
Merging	Limitations of merge criteria are the same limitations as the sort criteria above.	No	N/A
Sub-totals	Same limitations as mentioned in "population statistics" above, except file size limitation is 100 million records.	No	N/A
Fraud pattern detection	Same limitations on sort criteria as above.	No	N/A

Known system limitations

Description

Known Limitation

Intention to fix?

Fix Date

Population statistics

Amounts less than -920 trillion and more than +920 trillion are not processed (i.e. ignored). Any amounts containing fractional pennies are rounded to the nearest 1/100 of a cent. For amounts without fractional pennies, results should be accurate to the penny for population total amounts of up to $10 trillion. Between 10 trillion and 100 trillion, rounding errors totaling 5 cents can occur. For amounts between 100 trillion and 920 trillion the extent of rounding errors is unknown, but thought to be less than $1 total. However, amounts with fractional pennies could result in rounding errors of up to $5 per 100,000 transactions.

The maximum number of records which can be processed in a single run is approximately 500,000,000.

Yes

2-3-2008

Ageing, summarization and data stratification

These all have the same limitations as described above, except the only limit on the number of rows is that imposed by the operating system.

Yes

2-4-2008

Benford's Law, Weekdays, holidays

Counts have a maximum of approximately 2,000,000,000 each.

N/A

Numeric rounding - other (e.g. CMA sampling, Benford array)

Use of floating point arithmetic results in rounding errors in some instances.

Yes

3-22-2008

Date handling

Only handles dates between AD 100 and AD 9999.

N/A

Maximum number of rows of data

Operating system limits only - should handle files at least 2 GB in size. For population statistics, limit is 500,000,000.

N/A

Maximum number of ageing bins

Yes

Maximum number of stratification bins

100

Yes

Largest number of zeros tested on a round number

14, e.g. 900,000,000,000,000

Yes

1-29-2008

Holiday testing

Limited to U.S. federal holidays only

N/A

Sequential numbers

Limit to nine digit numbers (intent is to expand to 14)

Yes

Top/Bottom "N" Report (extremes)

Extremes for up to the top 100 and bottom 100 items are provided (limited to 100).

N/A

Sorting

Maximum number of rows which can be sorted is approximately 2 billion in a single file. Sorting is done in "chunks" of up to 1,000,000 rows. Sort speed is dependent upon the number of sort keys and type (e.g. numeric, character or date). Sort method used is "heap sort". Sort speed of 5-10,000 rows per second is common.

N/A

Sorting

Maximum number of columns which can be specified as part of the sort criteria is 200. Numeric column values of less than -920 trillion or greater than +920 trillion are not handled. Date column values must be between AD 100 and AD 9999.

N/A

Merging

Limitations of merge criteria are the same limitations as the sort criteria above.

N/A

Sub-totals

Same limitations as mentioned in "population statistics" above, except file size limitation is 100 million records.

N/A

Fraud pattern detection

Same limitations on sort criteria as above.

N/A