Maybe you could help me design that?

https://godbolt.org/z/cqcsGbqvc

The initial state is a null state and it is not a final state. On encountering an alphabet a or a b you transfer to corresponding state. Design your fsm such that when the character encountered last is an a or b and the character before that is also the same character then you reach the final state. Otherwise you reach an error state.

What is an error state like? I have not been able to understand this topic well

Ask your professor. This btw is a C/C++ community. Questions outside of that are not encouraged here.

How to generate a random double between DBL_MIN and DBL_MAX in C?

OOpsie, fixed a few optimization mistakes https://godbolt.org/z/8jfja5M8n @ollirz if you have an opinion on this (it is based on your code to some extend)

How is this a uniform distribution? Maybe I missed something but you are setting only 32 bits out of the 52 bits reserved for the mantissa You are missing out on denormalized numbers because the chance of generating an exponent value exactly equal to -1023 is less. Your program doesn't generate denormalized numbers with the same probability as that of normalised numbers.

although in the beginning user hasn't been prompted to give the value card_name[0], but in the condition , we check its value while(card_name[0]!='X'),if card_name doesn't have a value how can it check against some value https://dpaste.org/yd2Y

Fixed in the blogpost

I will go through the blogpost later.

It is Undefined Behavior. Reading a local variable which is not explicitly assigned an initial value is wrong.

OOpsie, fixed a few optimization mistakes https://godbolt.org/z/8jfja5M8n @ollirz if you have an opinion on this (it is based on your code to some extend)

https://archivist.nekoit.xyz/ieee754-doubles-and-uniform-random-distribution-in-c/ Here you go, this should be complete and explain all the ins and outs

It seems I can produce results outside of [-1,1] and can create results such as nan

It seems I can produce results outside of [-1,1] and can create results such as nan

Appreciated

"An exponent e starts at 1022 (which is actually -1) and has 50% chance of being 1022, 25% chance to be 1021, 12.5% chance to be 1020..." This statement seems wrong because X.e will be 1022 only if bit_divider is all 0 bits. The probability of this happening is 1/2^32. How then can you say that e being 1022 is 50% likely? It will be 1021 only if there is only 1 one bit in bit_divider. This is not 25% likely either. This is not a uniform distribution still. First of all denormalized numbers still are not equally likely. Second, the fact that you are adding a 32 bit number and a 64 bit number and ignoring the top 12 bits makes this a non uniform distribution.

hello, Homework Write a program that describes integers -9999 to +9999 with their pronunciation. use switch and case. How can I find a shortcut instead of typing one by one?

hello, Homework Write a program that describes integers -9999 to +9999 with their pronunciation. use switch and case. How can I find a shortcut instead of typing one by one?

but I can’t. I’m new at C

is there a shortcut for this?

so Do I have to write them all one by one?

so what is the logic of this

I know but I need a answet

do you mean that the exponent will be 1022 if the bottom bit of bit_divider is 0? I take bits from bit_divider 1 by one

Was the return statement in the else case there before? I seem to have missed it. Let me tell you why this does not generate numbers with equally likely behavior. The first problem is adding a 64 bit number and a 32 bit number and ignoring the top 12 bits. This makes some numbers for the mantissa more likely than the others. The second problem is denormalized numbers. You won't generate them at all. The third problem will be evident if you consider the probability of a number being generated that is in an uneven range -0.75 and +0.25 (I.e where the left end and right end are not equui distant from 0). You concern yourself with only the exponent and not the mantissa. The problem is that for a uniform distribution,the probability of numbers being generated in the range 0.5 to 1 should be the same as the probability of numbers being in the range -0.3 to 0.2. This won't happen in your case.

1/ How so, given that the 12 random bits are ignored, we still have the 52 random bits what here could make some numbers more likely than other? 2/ Do I need to generate denormalized numbers to cover the range uniformly? I argue not just like Clang and GCC developers 3/ I actually do, it changes nothing because I don't place numbers in the range according to a random exponent, but to an exponent with a ponderated random that follows not the amount of numbers in those ranges of ieee754 but the amount of real numbers in those ranges

1) Here is an example. Assuming X.m is 3 bits long and and decent_rand() generates a 2 bit number. So your code is equivalent to X.m = decent_rand() + decent_rand() << 2. Assuming decent_rand generates a 2 bit number with equal likelihood, your program should generate all buts from 000 to 111 with equal likelihood. 000 will be generated if decent_rand 1 produced 00 and decent_rand 2 produced 00 as well. 111 will be generated if decwnt_rand 1 produced 11 and decent_rand 2 produced 01 or 11. So the chance of 111 being generated is higher than the chance of 000 being generated 2) Denormalized numbers must be generated and if you take uniform_real_distribution in the standard C++ library, you will see that they are indeed generates. 3) The problem is the same as in situation 1

1) Here is an example. Assuming X.m is 3 bits long and and decent_rand() generates a 2 bit number. So your code is equivalent to X.m = decent_rand() + decent_rand() << 2. Assuming decent_rand generates a 2 bit number with equal likelihood, your program should generate all buts from 000 to 111 with equal likelihood. 000 will be generated if decent_rand 1 produced 00 and decent_rand 2 produced 00 as well. 111 will be generated if decwnt_rand 1 produced 11 and decent_rand 2 produced 01 or 11. So the chance of 111 being generated is higher than the chance of 000 being generated 2) Denormalized numbers must be generated and if you take uniform_real_distribution in the standard C++ library, you will see that they are indeed generates. 3) The problem is the same as in situation 1

Design a finite-state automaton for the alphabet Σ = {a, b} that accepts any sequence ending with at least two a or two b. Draw the state diagram, remember to show the start state and the state (s) acceptable.

2/ Nope, I looked at the code and can tell you that no they are not generated for a std::uniform_real_distribution(0.0,1.0)

http://coliru.stacked-crooked.com/a/585be0cdece94068 Denormalized numbers are indeed generated

2/ Nope, I looked at the code and can tell you that no they are not generated for a std::uniform_real_distribution(0.0,1.0)

Re read what I said

Ok let me check the standard library code. I am not sure how denormalized numbers can be left out when it is a uniform distribution

Is it possible for GNU make to enforce compilation using C++20 without makefile ? I could achieve the same result with g++ by using g++ -std=c++2a, it worked but wasn't practical, so I made an alias in my bashrc alias g++='g++ -std=c++2a', it didn't work. My code was still compiled against C++17, and due to the nature of my project, using make is better approach than g++ anyway. Is there any system-wide solution to enforce C++20 ? Machine : Linux (Ubuntu) GNU make version : 4.2.1 g++ version : 11.1.0

how do you intend to use make without a makefile?

If you want to move a line of code up or down, then you need to check your IDE settings (the code editor your use) to find what is the shortcut for that. Most of the IDEs use Ctrl+Shift+Up or Ctrl+Shift+Down combinations. Or you can just cut that code line, and paste it where you need it. Anyway if I understanding this question correctly, it's not really related to C/C++

Can gcc compile and run a source code on one command, just like go run xxx

Hey @Tazmikar would you want to take a look at this? you are far better at C than I am (I am more a C++ person) but I think this is decent at making a uniform distribution of doubles

Not loading.

I will investigate the issue with that, maybe a dns not propagated properly