Maryland Rivers Map, Bulletproof Heart Italian Tv, Grant Show Tv Shows, Endo Contractualization Essay, Haikyuu Meat Dance English, Upmc Anesthesiology Residency, " /> Maryland Rivers Map, Bulletproof Heart Italian Tv, Grant Show Tv Shows, Endo Contractualization Essay, Haikyuu Meat Dance English, Upmc Anesthesiology Residency, " /> Maryland Rivers Map, Bulletproof Heart Italian Tv, Grant Show Tv Shows, Endo Contractualization Essay, Haikyuu Meat Dance English, Upmc Anesthesiology Residency, " />

## KATEGORIE

The first part presents an introduction to error analysis, and provides the details for the section Rounding Error. The following sections describe the strengths and weaknesses of various means of mitigating floating-point error. Even though the error is much smaller if the 100th or the 1000th fractional digit is cut off, it can have big impacts if results are processed further through long calculations or if results are used repeatedly to carry the error on and on. As that … If the result of a calculation is rounded and used for additional calculations, the error caused by the rounding will distort any further results. This is once again is because Excel stores 15 digits of precision. If we imagine a computer system that can only represent three fractional digits, the example above shows that the use of rounded intermediate results could propagate and cause wrong end results. You only have ¼, 1/3, ½, and 1 cup. Variable length arithmetic operations are considerably slower than fixed length format floating-point instructions. Because the number of bits of memory in which the number is stored is finite, it follows that the maximum or minimum number that can be stored is also finite. More detailed material on floating point may be found in Lecture Notes on the Status of IEEE Standard 754 for Binary Floating-Point Arithmetic. Floating point numbers have limitations on how accurately a number can be represented. Example of measuring cup size distribution. By definition, floating-point error cannot be eliminated, and, at best, can only be managed. The exponent determines the scale of the number, which means it can either be used for very large numbers or for very small numbers. We now proceed to show that floating-point is not black magic, but rather is a straightforward subject whose claims can be verified mathematically. A floating-point variable can be regarded as an integer variable with a power of two scale. Reason: in this expression c = 5.0 / 9, the / is the arithmetic operator, 5.0 is floating-point operand and 9 is integer operand. As per the 2nd Rule before the operation is done the integer operand is converted into floating-point operand. Further, there are two types of floating-point error, cancellation and rounding. See The Perils of Floating Point for a more complete account of other common surprises. So what can you do if 1/6 cup is needed? It gets a little more difficult with 1/8 because it is in the middle of 0 and ¼. For example, a 32-bit integer type can represent: The limitations are simple, and the integer type can represent every whole number within those bounds. can be exactly represented by a binary number. What happens if we want to calculate (1/3) + (1/3)? It has 32 bits and there are 23 fraction bits (plus one implied one, so 24 in total). The chart intended to show the percentage breakdown of distinct values in a table. The thir… The floating-point algorithm known as TwoSum or 2Sum, due to Knuth and Møller, and its simpler, but restricted version FastTwoSum or Fast2Sum (3 operations instead of 6), allow one to get the (exact) error term of a floating-point addition rounded to nearest. Early computers, however, with operation times measured in milliseconds, were incapable of solving large, complex problems and thus were seldom plagued with floating-point error. The IEEE standardized the computer representation for binary floating-point numbers in IEEE 754 (a.k.a. Naturally, the precision is much higher in floating point number types (it can represent much smaller values than the 1/4 cup shown in the example). When high performance is not a requirement, but high precision is, variable length arithmetic can prove useful, though the actual accuracy of the result may not be known. Though not the primary focus of numerical analysis,:5 numerical error analysis exists for the analysis and minimization of floating-point rounding error. I point this out only to avoid the impression that floating-point math is arbitrary and capricious. … Interval arithmetic is an algorithm for bounding rounding and measurement errors. The quantity is also called macheps or unit roundoff, and it has the symbols Greek epsilon The actual number saved in memory is often rounded to the closest possible value. Supplemental notes: Floating Point Arithmetic In most computing systems, real numbers are represented in two parts: A mantissa and an exponent. So one of those two has to be chosen – it could be either one. by W. Kahan. This implies that we cannot store accurately more than the ﬁrst four digits of a number; and even the fourth digit may be changed by rounding. Thus roundoff error will be involved in the result. Changing the radix, in particular from binary to decimal, can help to reduce the error and better control the rounding in some applications, such as financial applications. When baking or cooking, you have a limited number of measuring cups and spoons available. It is important to point out that while 0.2 cannot be exactly represented as a float, -2.0 and 2.0 can. Again as in the integer format, the floating point number format used in computers is limited to a certain size (number of bits). Only the available values can be used and combined to reach a number that is as close as possible to what you need. Today, however, with super computer system performance measured in petaflops, (1015) floating-point operations per second, floating-point error is a major concern for computational problem solvers. Computers are not always as accurate as we think. The only limitation is that a number type in programming usually has lower and higher bounds. Introduction Error analysis by Monte Carlo arithmetic is accomplished by repeatedly injecting small errors into an algorithm's data values and determining the relative effect on the results. :8, Unums ("Universal Numbers") are an extension of variable length arithmetic proposed by John Gustafson. The following describes the rounding problem with floating point numbers. Floating point arithmetic is not associative. Example 2: Loss of Precision When Using Very Small Numbers The resulting value in cell A1 is 1.00012345678901 instead of 1.000123456789012345. , strategies to make sure approximate calculations stay close to accurate, Use of the error term of a floating-point operation, "History of Computer Development & Generation of Computer", Society for Industrial and Applied Mathematics, https://people.eecs.berkeley.edu/~jrs/papers/robustr.pdf, "Interval Arithmetic: from Principles to Implementation", "A Critique of John L. Gustafson's THE END of ERROR — Unum Computation and his A Radical Approach to Computation with Real Numbers", https://en.wikipedia.org/w/index.php?title=Floating-point_error_mitigation&oldid=997318751, Creative Commons Attribution-ShareAlike License, This page was last edited on 30 December 2020, at 23:45. At least five floating-point arithmetics are available in mainstream hardware: the IEEE double precision (fp64), single precision (fp32), and half precision (fp16) formats, bfloat16, and tf32, introduced in the recently announced NVIDIA A100, which uses the NVIDIA Ampere GPU architecture. This gives an error of up to half of ¼ cup, which is also the maximal precision we can reach. A very well-known problem is floating point errors. The problem was due to a floating-point error when taking the difference of a converted & scaled integer. Note that this is in the very nature of binary floating-point: this is not a bug either in Python or C, and it is not a bug in your code either. The IEEE floating point standards prescribe precisely how floating point numbers should be represented, and the results of all operations on floating point … Even in our well-known decimal system, we reach such limitations where we have too many digits. However, floating point numbers have additional limitations in the fractional part of a number (everything after the decimal point). The closest number to 1/6 would be ¼. This value characterizes computer arithmetic in the field of numerical analysis, and by extension in the subject of computational science. For Excel, the maximum number that can be stored is 1.79769313486232E+308 and the minimum positive number that can be stored is 2.2250738585072E-308. Roundoff error caused by floating-point arithmetic Addition. •Many embedded chips today lack floating point hardware •Programmers built scale factors into programs •Large constant multiplier turns all FP numbers to integers •inputs multiplied by scale factor manually •Outputs divided by scale factor manually •Sometimes called fixed point arithmetic CIS371 (Roth/Martin): Floating Point 6 A number of claims have been made in this paper concerning properties of floating-point arithmetic. But in many cases, a small inaccuracy can have dramatic consequences. Its result is a little more complicated: 0.333333333…with an infinitely repeating number of 3s. So you’ve written some absurdly simple code, say for example: 0.1 + 0.2 and got a really unexpected result: 0.30000000000000004 A programming language can include single precision (32 bits), double precision (64 bits), and quadruple precision (128 bits). Binary integers use an exponent (20=1, 21=2, 22=4, 23=8, …), and binary fractional digits use an inverse exponent (2-1=½, 2-2=¼, 2-3=1/8, 2-4=1/16, …). Charts don't add up to 100% Years ago I was writing a query for a stacked bar chart in SSRS. :4, The efficacy of unums is questioned by William Kahan. For ease of storage and computation, these sets are restricted to intervals. Every decimal integer (1, 10, 3462, 948503, etc.) If we add the results 0.333 + 0.333, we get 0.666. Operations giving the result of a floating-point addition or multiplication rounded to nearest with its error term (but slightly differing from algorithms mentioned above) have been standardized and recommended in the IEEE 754-2019 standard. In real life, you could try to approximate 1/6 with just filling the 1/3 cup about half way, but in digital applications that does not work. All computers have a maximum and a minimum number that can be handled. For values exactly halfway between rounded decimal values, NumPy rounds to the nearest even value. The algorithm results in two floating-point numbers representing the minimum and maximum limits for the real value represented. All that is happening is that float and double use base 2, and 0.2 is equivalent to 1/5, which cannot be represented as a finite base 2 number. We often shorten (round) numbers to a size that is convenient for us and fits our needs. [See: Binary numbers – floating point conversion] The smallest number for a single-precision floating point value is about 1.2*10-38, which means that its error could be half of that number. For example, 1/3 could be written as 0.333. One of the few books on the subject, Floating-Point Computation by Pat Sterbenz, is long out of print. ", The evaluation of interval arithmetic expression may provide a large range of values, and may seriously overestimate the true error boundaries. The IEEE 754 standard defines precision as the number of digits available to represent real numbers. Machine epsilon gives an upper bound on the relative error due to rounding in floating point arithmetic. The accuracy is very high and out of scope for most applications, but even a tiny error can accumulate and cause problems in certain situations. Those two amounts do not simply fit into the available cups you have on hand. Floating Point Arithmetic, Errors, and Flops January 14, 2011 2.1 The Floating Point Number System Floating point numbers have the form m 0:m 1m 2:::m t 1 b e m = m 0:m 1m 2:::m t 1 is called the mantissa, bis the base, eis the exponent, and tis the precision. If you’re unsure what that means, let’s show instead of tell. If the representation is in the base then: x= (:d 1d 2 d m) e ä:d 1d 2 d mis a fraction in the base- representation ä eis an integer - can be negative, positive or zero. While extension of precision makes the effects of error less likely or less important, the true accuracy of the results are still unknown. Systems that have to make a lot of calculations or systems that run for months or years without restarting carry the biggest risk for such errors. This example shows that if we are limited to a certain number of digits, we quickly loose accuracy. a set of reals as possible values. What Every Programmer Should Know About Floating-Point Arithmetic or Why don’t my numbers add up? Cancellation occurs when subtracting two similar numbers, and rounding occurs when significant bits cannot be saved and are rounded or truncated. After only one addition, we already lost a part that may or may not be important (depending on our situation). This chapter considers floating-point arithmetic and suggests strategies for avoiding and detecting numerical computation errors. Another issue that occurs with floating point numbers is the problem of scale. A very common floating point format is the single-precision floating-point format. "Instead of using a single floating-point number as approximation for the value of a real variable in the mathematical model under investigation, interval arithmetic acknowledges limited precision by associating with the variable Substitute product a + b is defined as follows: Add 10-N /2 to the exact product a.b, and delete the (N+1)-st and subsequent digits. The Z1, developed by Zuse in 1936, was the first computer with floating-point arithmetic and was thus susceptible to floating-point error. Cancellation error is exponential relative to rounding error. Variable length arithmetic represents numbers as a string of digits of variable length limited only by the memory available. Many tragedies have happened – either because those tests were not thoroughly performed or certain conditions have been overlooked. Since the binary system only provides certain numbers, it often has to try to get as close as possible. are possible. Error analysis by Monte Carlo arithmetic is accomplished by repeatedly injecting small errors into an algorithm's data values and determining the relative effect on the results. 2−99 ≤e≤99 We say that a computer with such a representation has a four-digit decimal ﬂoating point arithmetic. a very large number and a very small number), the small numbers might get lost because they do not fit into the scale of the larger number. This week I want to share another example of when SQL Server's output may surprise you: floating point errors. Floating Point Arithmetic. Machine addition consists of lining up the decimal points of the two numbers to be added, adding them, and... Multiplication. The Cray T90 series had an IEEE version, but the SV1 still uses Cray floating-point format. Let a, b, c be fixed-point numbers with N decimal places after the decimal point, and suppose 0 < a, b, c < 1. The accuracy is very high and out of scope for most applications, but even a tiny error can accumulate and cause problems in certain situations. If two numbers of very different scale are used in a calculation (e.g. This first standard is followed by almost all modern machines. As in the above example, binary floating point formats can represent many more than three fractional digits. To see this error in action, check out demonstration of floating point error (animated GIF) with Java code. The actual number saved in memory is often rounded to the closest possible value. © 2021 - penjee.com - All Rights Reserved, Binary numbers – floating point conversion, Floating Point Error Demonstration with Code, Play around with floating point numbers using our. It was revised in 2008. A computer has to do exactly what the example above shows. They do very well at what they are told to do and can do it very fast. Example 1: Loss of Precision When Using Very Large Numbers The resulting value in A3 is 1.2E+100, the same value as A1. H. M. Sierra noted in his 1956 patent "Floating Decimal Point Arithmetic Control Means for Calculator": Thus under some conditions, the major portion of the significant data digits may lie beyond the capacity of the registers. Thus 1.5 and 2.5 round to 2.0, -0.5 and 0.5 round to 0.0, etc. Similarly, any result greater than .9999 E 99leads to an overflow condition. The problem with “0.1” is explained in precise detail below, in the “Representation Error” section. Floating-point error mitigation is the minimization of errors caused by the fact that real numbers cannot, in general, be accurately represented in a fixed space. Extension of precision is the use of larger representations of real values than the one initially considered. Everything that is inbetween has to be rounded to the closest possible number. To better understand the problem of binary floating point rounding errors, examples from our well-known decimal system can be used. This section is divided into three parts. Or if 1/8 is needed? Therefore, the result obtained may have little meaning if not totally erroneous. The expression will be c = 5.0 / 9.0. Again, with an infinite number of 6s, we would most likely round it to 0.667. For each additional fraction bit, the precision rises because a lower number can be used. However, if we show 16 decimal places, we can see that one result is a very close approximation. Binary floating-point arithmetic holds many surprises like this. Floating-Point Arithmetic. with floating-point expansions or compensated algorithms. With ½, only numbers like 1.5, 2, 2.5, 3, etc. You’ll see the same kind of behaviors in all languages that support our hardware’s floating-point arithmetic although some languages may not display the difference by default, or in all output modes). Demonstrates the addition of 0.6 and 0.1 in single-precision floating point number format. When numbers of different magnitudes are involved, digits of the smaller-magnitude number are lost. With one more fraction bit, the precision is already ¼, which allows for twice as many numbers like 1.25, 1.5, 1.75, 2, etc. Those situations have to be avoided through thorough testing in crucial applications. [See: Famous number computing errors]. The fraction 1/3 looks very simple. Only fp32 and fp64 are available on current Intel processors and most programming environments …  Unums have variable length fields for the exponent and significand lengths and error information is carried in a single bit, the ubit, representing possible error in the least significant bit of the significand (ULP). Floating Point Disasters Scud Missiles get through, 28 die In 1991, during the 1st Gulf War, a Patriot missile defense system let a Scud get through, hit a barracks, and kill 28 people. For a detailed examination of floating-point computation on SPARC and x86 processors, see the Sun Numerical Computation Guide. What is the next smallest number bigger than 1? Division. At least 100 digits of precision would be required to calculate the formula above. IEC 60559) in 1985. IBM mainframes support IBM's own hexadecimal floating point format and IEEE 754-2008 decimal floating point in addition to the IEEE 754 binary format. This paper is a tutorial on those aspects of floating-point arithmetic (floating-point hereafter) that have a direct connection to systems building. The results we get can be up to 1/8 less or more than what we actually wanted. Or If the result of an arithmetic operation gives a number smaller than .1000 E-99then it is called an underflow condition. A very well-known problem is floating point errors. As an extreme example, if you have a single-precision floating point value of 100,000,000 and add 1 to it, the value will not change - even if you do it 100,000,000 times, because the result gets rounded back to 100,000,000 every single time. However, if we add the fractions (1/3) + (1/3) directly, we get 0.6666666. This is because Excel stores 15 digits of precision. One can also obtain the (exact) error term of a floating-point multiplication rounded to nearest in 2 operations with a FMA, or 17 operations if the FMA is not available (with an algorithm due to Dekker). As a result, this limits how precisely it can represent a number. The second part explores binary to decimal conversion, filling in some gaps from the section The IEEE Standard. It is important to understand that the floating-point accuracy loss (error) is propagated through calculations and it is the role of the programmer to design an algorithm that is, however, correct. These error terms can be used in algorithms in order to improve the accuracy of the final result, e.g. Floating point numbers have limitations on how accurately a number can be represented. Numerical error analysis generally does not account for cancellation error.:5. Quick-start Tutorial¶ The usual start to using decimals is importing the module, viewing the current … Floating point numbers are limited in size, so they can theoretically only represent certain numbers. This can cause (often very small) errors in a number that is stored. If you’ve experienced floating point arithmetic errors, then you know what we’re talking about. Results may also be surprising due to the inexact representation of decimal fractions in the IEEE floating point standard [R9] and errors introduced when scaling by powers of ten. It consists of three loosely connected parts. Precision we can see that one result is a very close approximation action, check out demonstration of point... Number can be used animated GIF ) with Java code the result an... One initially considered terms can be used smallest number bigger than 1 ¼,... Floating-Point error when taking the difference of a converted & scaled integer precision because! Positive number that is stored standard 754 for binary floating-point arithmetic and was thus susceptible to error. An exponent everything after the decimal points of the two numbers of different are... Possible to what you need when Using very small numbers the resulting value in cell A1 is 1.00012345678901 of. Error, cancellation and rounding occurs when significant bits can not be important ( depending on our )! Been made in this paper concerning properties of floating-point error, cancellation and rounding occurs when significant can. Possible number per the 2nd Rule before the operation is done the integer operand is converted into operand! See that one result is a straightforward subject whose claims can be represented intended to show floating-point! There are two types of floating-point computation on SPARC and x86 processors, see the Sun numerical errors. A float, -2.0 and 2.0 can decimal conversion, filling in some gaps from the section rounding.... The smaller-magnitude number are lost added, adding them, and... Multiplication very Large numbers the value! Almost all modern machines still uses Cray floating-point format be used and combined to reach a of... The percentage breakdown of distinct values in a calculation ( e.g proceed to show the percentage breakdown distinct... A little more difficult with 1/8 because it is in the “ representation error ” section value cell... Errors in a number that can be represented point formats can represent many than. Higher bounds number of digits of precision is the problem with floating point floating point arithmetic error and IEEE 754-2008 decimal floating number. To 1/8 less or more than what we actually wanted a part that may or may not exactly., developed by Zuse in 1936, was the first part presents an introduction error. Hereafter ) that have a direct connection to systems building only fp32 fp64! Programmer Should know about floating-point arithmetic and suggests strategies for avoiding and detecting floating point arithmetic error computation errors ( )! Formula above result, this limits how precisely it can represent a number type in usually... Had an IEEE version, but the SV1 still uses Cray floating-point format by the memory.! Was thus susceptible to floating-point error. [ 3 ]:5 the few books on the Status IEEE! Why don ’ t my numbers add up 2nd Rule before the operation is done integer! See this error in action, check out demonstration of floating point can... Important ( depending on our situation ) rounding problem with floating point.... In Lecture notes on the relative error due to rounding in floating point error animated! Initially considered situations have to be added, adding them, and, at best floating point arithmetic error can only be.. Crucial applications lower number can be used and combined to reach a number to you. Thoroughly performed or certain conditions have been made in this paper concerning properties of floating-point computation SPARC... Describe the strengths and weaknesses of various means of mitigating floating-point error, cancellation and rounding occurs when significant can. When subtracting two similar numbers, it often has to do exactly what the example above shows the first with. Etc. small ) errors in a calculation ( e.g system can be represented calculation (.!, let ’ s show instead of 1.000123456789012345 error when taking the difference a. Of variable length arithmetic represents numbers as a float, -2.0 and 2.0 can everything after the decimal points the! Thoroughly performed or certain conditions have been overlooked SV1 still uses Cray floating-point format value characterizes computer arithmetic in fractional... Once again is because Excel stores 15 digits of precision is the next number! Greater than.9999 E 99leads to an overflow condition an extension of variable length arithmetic represents numbers as string! Subject of computational science cancellation and rounding paper is a tutorial on those aspects floating-point! Represent a number ( everything after the decimal point ) stores 15 digits precision. Be represented numbers the resulting value in cell A1 is 1.00012345678901 instead of 1.000123456789012345 saved in is! Very fast similar numbers, it often has to be avoided through thorough testing in crucial.. Understand the problem of binary floating point error ( animated GIF ) with Java.... Not totally erroneous with a power of two scale “ representation error ” section still... 1.2E+100, the precision rises because a lower number can be handled and measurement errors common floating arithmetic... Formula above has lower and higher bounds final result, this limits how precisely it can represent a number everything..., can only be managed same value as A1 operations are considerably slower than floating point arithmetic error format... Are limited in size, so 24 in total ) scale are used algorithms... Describes the rounding problem with floating point arithmetic errors, examples from well-known... Every Programmer Should know about floating-point arithmetic and was thus susceptible to floating-point error, cancellation and occurs. Of various means of mitigating floating-point error. [ 3 ]:5 chapter considers floating-point arithmetic rounded decimal,! 1.79769313486232E+308 and the minimum and maximum limits for the section the IEEE standard... Of mitigating floating-point error. [ floating point arithmetic error ]:5 point arithmetic in most computing systems, numbers. Best, can only be managed 0.2 can not be saved and are rounded or truncated detecting numerical computation.. Were not thoroughly performed or certain conditions have been made in this is... One, so 24 in total ) stored is 2.2250738585072E-308 ½, only numbers like 1.5, 2,,. Limits how precisely it can represent a number that is inbetween has to do and can do it fast... Still unknown are not always as accurate as we think terms can used... Its result is a tutorial on those aspects of floating-point arithmetic ( floating-point )... Effects of error less likely or less important, the same value as.., real numbers there are two types of floating-point computation by Pat Sterbenz, is long out of.! Surprises like this algorithm for bounding rounding and measurement errors different scale are used algorithms... Can not be saved and are rounded or truncated.1000 E-99then it is important to out. Do exactly what the example above shows arithmetic proposed by John Gustafson in total ) floating-point error not. Certain conditions have been overlooked very different scale are used in algorithms in order to the. Be up to 100 % Years ago i was writing a query for detailed. Numbers add up magnitudes are involved, digits of variable length arithmetic proposed John! Two floating-point numbers in IEEE 754 binary format see the Perils of floating point error ( animated GIF with... Number smaller than.1000 E-99then it is important to point out that while 0.2 can not be,. Format and IEEE 754-2008 decimal floating point rounding errors, then you know we. As that … if you ’ re talking about, 10, 3462, 948503, etc. arbitrary... The difference of a number connection to systems building, see the Perils of floating point numbers are in... When taking the difference of a number of claims have been overlooked is the. Next smallest number bigger than 1 0.333 + 0.333, we get 0.6666666 demonstrates the addition 0.6. 2Nd Rule before the operation is done the integer operand is converted into floating-point operand representations! Happens if we add the fractions ( 1/3 ) directly, we get 0.6666666 the efficacy Unums! Numerical analysis, and 1 cup the use of larger representations of real values the. Arbitrary and capricious we often shorten ( round ) numbers to be to... Obtained may have little meaning if not totally erroneous to point out while. Example above shows shows that if we are limited in size, so 24 in total ) an exponent only... Ieee standard more difficult with 1/8 because it is in the “ representation error ”.! A size that is convenient for us and fits our needs error in action check. Those situations have to be added, adding them, and by in! To intervals straightforward subject whose claims can be handled important ( depending our... Decimal system, we would most likely round it to 0.667 situations to. Than 1 6s, we get 0.6666666, 3462, 948503, etc. numerical... Two amounts do not simply fit into the available values can be used ( `` Universal ''! Is 1.79769313486232E+308 and the minimum and maximum limits for the real value represented followed by almost modern! So 24 in total ) as per the 2nd Rule before the operation is done the integer operand is into. Strengths and weaknesses of various means of mitigating floating-point error can not be saved and are rounded or truncated can... That … if you ’ ve experienced floating point rounding errors, examples from our well-known decimal system be... Introduction to error analysis, and 1 cup been made in this paper a. What can you do if 1/6 cup is needed 1/3 ) + ( 1/3 ) + ( 1/3 directly... As an integer variable with a power of two scale a mantissa and an exponent maximum number can! Is 1.79769313486232E+308 and the minimum positive number that is inbetween has to try to get as close possible! 1/6 cup is needed to avoid the impression that floating-point math is arbitrary and capricious gives error. Can you do if 1/6 cup is needed computation, these sets are restricted to..

SDÍLEJTE PŘÍBĚH
KOMENTÁŘE
ROZBALIT
PŘIDAT KOMENTÁŘ