There is also a new experimental tf.data API for saving and loading datasets(https://www.tensorflow.org/versions/r2.3/api_docs/python/tf/data/experimental/save)
https://github.com/tensorflow/tensorflow/commit/4d58a67a9f19ab8d0cfbb2d8e461ebb73ce06db6
https://github.com/tensorflow/tensorflow/issues/38483#issuecomment-640963109

https://github.com/tensorflow/tensorflow/releases/tag/v2.3.0

Release 2.3.0

Major Features and Improvements

tf.data adds two new mechanisms to solve input pipeline bottlenecks and save resources:
- snapshot
- tf.data service.

In addition checkout the detailed guide for analyzing input pipeline performance with TF Profiler.

tf.distribute.TPUStrategy is now a stable API and no longer considered experimental for TensorFlow. (earlier tf.distribute.experimental.TPUStrategy ).
TF Profiler introduces two new tools: a memory profiler to visualize your model’s memory usage over time and a python tracer which allows you to trace python function calls in your model. Usability improvements include better diagnostic messages and profile options to customize the host and device trace verbosity level.
Introduces experimental support for Keras Preprocessing Layers API (tf.keras.layers.experimental.preprocessing.*) to handle data preprocessing operations, with support for composite tensor inputs. Please see below for additional details on these layers.
TFLite now properly supports dynamic shapes during conversion and inference. We’ve also added opt-in support on Android and iOS for XNNPACK, a highly optimized set of CPU kernels, as well as opt-in support for executing quantized models on the GPU.
Libtensorflow packages are available in GCS starting this release. We have also started to release a nightly version of these packages.
The experimental Python API tf.debugging.experimental.enable_dump_debug_info() now allows you to instrument a TensorFlow program and dump debugging information to a directory on the file system. The directory can be read and visualized by a new interactive dashboard in TensorBoard 2.3 called Debugger V2, which reveals the details of the TensorFlow program including graph structures, history of op executions at the Python (eager) and intra-graph levels, the runtime dtype, shape, and numerical composistion of tensors, as well as their code locations.

Breaking Changes

Increases the minimum bazel version required to build TF to 3.1.0.
tf.data
- Makes the following (breaking) changes to the tf.data .
- C++ API: - IteratorBase::RestoreInternal , IteratorBase::SaveInternal , and DatasetBase::CheckExternalState become pure-virtual and subclasses are now expected to provide an implementation.
- The deprecated DatasetBase::IsStateful method is removed in favor of DatasetBase::CheckExternalState .
- Deprecated overrides of DatasetBase::MakeIterator and MakeIteratorFromInputElement are removed.
- The signature of tensorflow::data::IteratorBase::SaveInternal and tensorflow::data::IteratorBase::SaveInput has been extended with SerializationContext argument to enable overriding the default policy for the handling external state during iterator checkpointing. This is not a backwards compatible change and all subclasses of IteratorBase need to be updated accordingly.
tf.keras
- Add a new BackupAndRestore callback for handling distributed training failures & restarts. Please take a look at this tutorial for details on how to use the callback.
tf.image.extract_glimpse has been updated to correctly process the case where centered=False and normalized=False . This is a breaking change as the output is different from (incorrect) previous versions. Note this breaking change only impacts tf.image.extract_glimpse and tf.compat.v2.image.extract_glimpse API endpoints. The behavior of tf.compat.v1.image.extract_glimpse does not change. The behavior of exsiting C++ kernel ExtractGlimpse does not change either, so saved models using tf.raw_ops.ExtractGlimpse will not be impacted.

Known Caveats

tf.lite
- Keras-based LSTM models must be converted with an explicit batch size in the input layer.

Bug Fixes and Other Changes

TF Core:

Set tf2_behavior to 1 to enable V2 for early loading cases.
Add execute_fn_for_device function to dynamically choose the implementation based on underlying device placement.
Eager:
- Add reduce_logsumexp benchmark with experiment compile.
- Give EagerTensor s a meaningful __array__ implementation.
- Add another version of defun matmul for performance analysis.
tf.function /AutoGraph:
- AutoGraph now includes into TensorFlow loops any variables that are closed over by local functions. Previously, such variables were sometimes incorrectly ignored.
- functions returned by the get_concrete_function method of tf.function objects can now be called with arguments consistent with the original arguments or type specs passed to get_concrete_function . This calling convention is now the preferred way to use concrete functions with nested values and composite tensors. Please check the guide for more details on concrete_ function .
- Update tf.function 's experimental_relax_shapes to handle composite tensors appropriately.
- Optimize tf.function invocation, by removing redundant list converter.
- tf.function will retrace when called with a different variable instead of simply using the dtype & shape .
- Improve support for dynamically-sized TensorArray inside tf.function .
tf.math :
- Narrow down argmin /argmax contract to always return the smallest index for ties.
- tf.math.reduce_variance and tf.math.reduce_std return correct computation for complex types and no longer support integer types.
- Add Bessel functions of order 0,1 to tf.math.special .
- tf.divide now always returns a tensor to be consistent with documentation and other APIs.
tf.image :
- Replaced tf.image.non_max_suppression_padded with a new implementation that supports batched inputs, which is considerably faster on TPUs and GPUs. Boxes with area=0 will be ignored. Existing usage with single inputs should still work as before.
tf.linalg
- Add tf.linalg.banded_triangular_solve .
tf.random :
- Add tf.random.stateless_parameterized_truncated_normal .
tf.ragged :
- Add tf.ragged.cross and tf.ragged.cross_hashed operations.
tf.RaggedTensor :
- RaggedTensor.to_tensor() now preserves static shape.
- Add tf.strings.format() and tf.print() to support RaggedTensors.
tf.saved_model :
- u/tf.function from SavedModel no longer ignores args after a RaggedTensor when selecting the concrete function to run.
- Fix save model issue for ops with a list of functions.
- Add tf.saved_model.LoadOptions with experimental_io_device as arg with default value None to choose the I/O device for loading models and weights.
- Update tf.saved_model.SaveOptions with experimental_io_device as arg with default value None to choose the I/O device for saving models and weights.
GPU
- No longer includes PTX kernels for GPU except for sm_70 to reduce binary size. On systems with NVIDIA® Ampere GPUs (CUDA architecture 8.0) or newer, kernels are JIT-compiled from PTX and TensorFlow can take over 30 minutes to start up. This overhead can be limited to the first start up by increasing the default JIT cache size with: export CUDA_CACHE_MAXSIZE=2147483648 .:
Others
- Retain parent namescope for ops added inside tf.while_loop /tf.cond /tf.switch_case .
- Update tf.vectorized_map to support vectorizing tf.while_loop and TensorList operations.
- tf.custom_gradient can now be applied to functions that accept nested structures of tensors as inputs (instead of just a list of tensors). Note that Python structures such as tuples and lists now won't be treated as tensors, so if you still want them to be treated that way, you need to wrap them with tf.convert_to_tensor .
- No lowering on gradient case op when input is DeviceIndex op.
- Extend the ragged version of tf.gather to support batch_dims and axis args.
- Update tf.map_fn to support RaggedTensors and SparseTensors.
- Deprecate tf.group . It is not useful in eager mode.
- Add CPU and GPU implementation of modified variation of FTRL/FTRLV2 that can triggerred by multiply_linear_by_lr allowing a learning rate of zero.

tf.data:

tf.data.experimental.dense_to_ragged_batch works correctly with tuples.
tf.data.experimental.dense_to_ragged_batch to output variable ragged rank.
tf.data.experimental.cardinality is now a method on tf.data.Dataset .
tf.data.Dataset now supports len(Dataset) when the cardinality is finite.

tf.distribute:

Expose experimental tf.distribute.DistributedDataset and tf.distribute.DistributedIterator to distribute input data when using tf.distribute to scale training on multiple devices.
- Added a get_next_as_optional method for tf.distribute.DistributedIterator class to return a tf.experimental.Optional instance that contains the next value for all replicas or none instead of raising an out of range error. Also see new guide on input distribution.
Allow var.assign on MirroredVariables with aggregation=NONE in replica context. Previously this would raise an error. We now allow this because many users and library writers find using .assign in replica context to be more convenient, instead of having to use Strategy.extended.update which was the previous way of updating variables in this situation.
tf.distribute.experimental.MultiWorkerMirroredStrategy adds support for partial batches. Workers running out of data now continue to participate in the training with empty inputs, instead of raising an error. Learn more about partial batches here.
Improve the performance of reading metrics eagerly under tf.distribute.experimental.MultiWorkerMirroredStrategy .
Fix the issue that strategy.reduce() inside tf.function may raise exceptions when the values to reduce are from loops or if-clauses.
Fix the issue that tf.distribute.MirroredStrategy cannot be used together with tf.distribute.experimental.MultiWorkerMirroredStrategy .
Add a tf.distribute.cluster_resolver.TPUClusterResolver.connect API to simplify TPU initialization.

tf.keras:

Introduces experimental preprocessing layers API (tf.keras.layers.experimental.preprocessing ) to handle data preprocessing operations such as categorical feature encoding, text vectorization, data normalization, and data discretization (binning). The newly added layers provide a replacement for the legacy feature column API, and support composite tensor inputs.
Added categorical data processing layers:
- IntegerLookup & StringLookup : build an index of categorical feature values
- CategoryEncoding : turn integer-encoded categories into one-hot, multi-hot, or tf-idf encoded representations
- CategoryCrossing : create new categorical features representing co-occurrences of previous categorical feature values
- Hashing : the hashing trick, for large-vocabulary categorical features
- Discretization : turn continuous numerical features into categorical features by binning their values
Improved image preprocessing layers: CenterCrop , Rescaling
Improved image augmentation layers: RandomCrop , RandomFlip , RandomTranslation , RandomRotation , RandomHeight , RandomWidth , RandomZoom , RandomContrast
Improved TextVectorization layer, which handles string tokenization, n-gram generation, and token encoding
- The TextVectorization layer now accounts for the mask_token as part of the vocabulary size when output_mode='int'. This means that, if you have a max_tokens value of 5000, your output will have 5000 unique values (not 5001 as before).
- Change the return value of TextVectorization.get_vocabulary() from byte to string . Users who previously were calling 'decode' on the output of this method should no longer need to do so.
Introduce new Keras dataset generation utilities :
- image_dataset_from_directory is a utility based on tf.data.Dataset , meant to replace the legacy ImageDataGenerator . It takes you from a structured directory of images to a labeled dataset, in one function call. Note that it doesn't perform image data augmentation (which is meant to be done using preprocessing layers).
- text_dataset_from_directory takes you from a structured directory of text files to a labeled dataset, in one function call.
- timeseries_dataset_from_array is a tf.data.Dataset -based replacement of the legacy TimeseriesGenerator . It takes you from an array of timeseries data to a dataset of shifting windows with their targets.
Added experimental_steps_per_execution arg to model.compile to indicate the number of batches to run per tf.function call. This can speed up Keras Models on TPUs up to 3x.
Extends tf.keras.layers.Lambda layers to support multi-argument lambdas, and keyword arguments when calling the layer.
Functional models now get constructed if any tensor in a layer call's arguments/keyword arguments comes from a keras input. Previously the functional api would only work if all of the elements in the first argument to the layer came from a keras input.
Clean up BatchNormalization layer's trainable property to act like standard python state when it's used inside tf.functions (frozen at tracing time), instead of acting like a pseudo-variable whose updates kind of sometimes get reflected in already-traced tf.function traces.
Add the Conv1DTranspose layer.
Refine the semantics of SensitivitySpecificityBase derived metrics. See the updated API docstrings for tf.keras.metrics.SensitivityAtSpecificity and tf.keras.metrics.SpecificityAtSensitivty.

tf.lite:

Converter
- Restored inference_input_type and inference_output_type flags in TF 2.x TFLiteConverter (backward compatible with TF 1.x) to support integer (tf.int8, tf.uint8) input and output types in post training full integer quantized models.
- Added support for converting and resizing models with dynamic (placeholder) dimensions. Previously, there was only limited support for dynamic batch size, and even that did not guarantee that the model could be properly resized at runtime.
- Enabled experimental support for a new quantization mode with 16-bit activations and 8-bit weights. See lite.OpsSet.EXPERIMENTAL_TFLITE_BUILTINS_ACTIVATIONS_INT16_WEIGHTS_INT8 .
CPU
- Fix an issue w/ dynamic weights and Conv2D on x86.
- Add a runtime Android flag for enabling XNNPACK for optimized CPU performance.
- Add a runtime iOS flag for enabling XNNPACK for optimized CPU performance.
- Add a compiler flag to enable building a TFLite library that applies XNNPACK delegate automatically when the model has a fp32 operation.
GPU
- Allow GPU acceleration starting with internal graph nodes
- Experimental support for quantized models with the Android GPU delegate
- Add GPU delegate whitelist.
- Rename GPU whitelist -> compatibility (list).
- Improve GPU compatibility list entries from crash reports.
NNAPI
- Set default value for StatefulNnApiDelegate::Options::max_number_delegated_partitions to 3.
- Add capability to disable NNAPI CPU and check NNAPI Errno.
- Fix crashes when using NNAPI with target accelerator specified with model containing Conv2d or FullyConnected or LSTM nodes with quantized weights.
- Fix ANEURALNETWORKS_BAD_DATA execution failures with sum /max /min /reduce operations with scalar inputs.
Hexagon
- TFLite Hexagon Delegate out of experimental.
- Experimental int8 support for most hexagon ops.
- Experimental per-channel quant support for conv in Hexagon delegate.
- Support dynamic batch size in C++ API.
CoreML
- Opensource CoreML delegate
Misc
- Enable building Android TFLite targets on Windows
- Add support for BatchMatMul .
- Add support for half_pixel_centers with ResizeNearestNeighbor .
- Add 3D support for BatchToSpaceND .
- Add 5D support for BroadcastSub , Maximum , Minimum , Transpose and BroadcastDiv .
- Rename kTfLiteActRelu1 to kTfLiteActReluN1To1 .
- Enable flex delegate on tensorflow.lite.Interpreter Python package.
- Add Buckettize , SparseCross and BoostedTreesBucketize to the flex whitelist.
- Add support for selective registration of flex ops.
- Add missing kernels for flex delegate whitelisted ops.
- Fix issue when using direct ByteBuffer inputs with graphs that have dynamic shapes.
- Fix error checking supported operations in a model containing HardSwish .

Packaging Support

Added tf.sysconfig.get_build_info() . Returns a dict that describes the build environment of the currently installed TensorFlow package, e.g. the NVIDIA CUDA and NVIDIA CuDNN versions used when TensorFlow was built.

Profiler

Fix a subtle use-after-free issue in XStatVisitor::RefValue() .

TPU Enhancements

Adds 3D mesh support in TPU configurations ops.
Added TPU code for FTRL with multiply_linear_by_lr .
Silently adds a new file system registry at gstpu .
Support restartType in cloud tpu client.
Depend on a specific version of google-api-python-client.
Fixes apiclient import.

Tracing and Debugging

Add a TFE_Py_Execute traceme.

XLA Support

Implement stable argmin and argmax

A guide on hitting Legend in Comp

Crossposting this from /crucibleplaybook, figured some people on here might find this helpful as well. A lot of this post applies to all Comp, not just hitting Legend.
I’ve been seeing a lot of posts lately asking for tips for hitting Legend in Comp so I figured I’d put together a brief guide for anyone that’s interested. I’m happy to see so many people interested in hitting Legend!
Intro
First off, a little bit about me. I never played D1 so I had a rough first few months of D2Y1 (and a rough first few weeks of Y2 with the new special weapon uptime and TTK) as it was my first time with the Destiny franchise. But even then had a blast in Crucible and I always wanted to get better. I’m also an extremely competitive person so that helped fuel my desire for improvement.
I play on Xbox and I just got my Unbroken title this season so I’ve been to Legend 3 times (S4, S6, S7). I’ve learned a ton along the way and hitting Legend each season has been easier and more enjoyable than the previous one for a variety of reasons that I’ll share in this post.
Improvement Mindset
While your end goal is to hit Legend, focusing on this binary goal isn’t a good idea. A better approach is to think of playing Comp with the main goal of improving both as a player and a team. With this more open and long-term mindset, you will improve rapidly as a player, win more often, and have a much more enjoyable experience as a result.
When you focus on something as binary as hitting a certain rank, every game or even decision within a game starts to feel tense and you put an immense amount of artificial pressure on yourself. This often builds over the course of a game. Even if it’s subconscious, it will effect your play. You’ll play too passive, too aggressive, and/or make bad decisions. Your brain will be too wrapped up playing out the consequences of failure to focus on what you should be doing to give yourself and your team the best chances of winning. It’s been scientifically proven both in real sports and in E-Sports that tension leads to poor performance.
Instead, take every engagement and every game as an opportunity to learn something and to improve. You WILL start getting your ass kicked at some point, it’s just a matter of when. It might be at 3k and it might not be until 5k, but at some point it’ll happen. And when it does, the best thing to do is to record your gameplay and watch it back.
Gameplay Review
You can easily record your gameplay via Twitch by streaming and having it save past broadcasts. Then you can watch your gameplay there, or you can take it a step further and download your gameplay and run it through a free video editing program such as DaVinci Resolve or iMovie. The advantage of doing it this way is you can better control the playback and even view it frame-by-frame.
I’d recommend picking a game that you performed poorly and watch it once all the way through and take some mental notes. Then I’d watch it again, noting each death with why you died and what you could have done better to either kill your opponent first or escape safely. Even if you died to a Wardcliff or a solo super, write down something you could have done differently to prevent dying. Then categorize and tally them the best you can. The most frequent ones are what you should focus on getting better at. This can be during your next Comp session or QP/Rumble.
The reason reviewing your gameplay is so important is it’ll help speed up your rate of improvement and help you get past your current plateau a bit faster. Games in high comp tend to be very fast paced so it can be hard to think about or remember exactly what happened. Or what you think happened in the moment wasn’t what really happened and the gameplay review will show you this.
While it’s certainly possible to improve naturally and over time, recording and reviewing your gameplay will make you improve faster.
Playing the meta
A lot of people seem reluctant to use meta loadouts for whatever reason. I think most of it boils down to either wanting to be unique, or having a superiority complex by refusing to use certain good or easy to use weapons and strategies because they’re “cheap” or too easy. Throw all of this out the window.
There’s nothing cheap in Comp (other than DDoSing which is actual cheating and we won’t discuss it). There’s nothing that takes “no skill” to use. If it’s in the game then it’s fair game to be used as much and as effectively as possible. Everything has a counter. If you don’t believe this then you probably have a scrub mentality and it’s going to hold you back. There are some great posts about scrub mentality on this very sub.
Meta loadouts or weapons are usually the perfect cross section of both lethality and ease-of-use - USE THEM. This is the time and the place. Your opponents are trying to win at all costs and so should you.
I don’t want to go too much into detail here or debate here, but in general these are the best options for high comp on Console (4k+). They’re ranked in terms of effectiveness, so it’s probably better to improve with something at the top of the list than use something at the bottom.
Primary Weapons: * Luna (NF if you have it already) * Adaptive or Aggressive pulse rifles * Ace/Thorn/TLW * Very well rolled Legendary HC * Jade Rabbit/Mida/Polaris Lance (large maps only)
Special Weapons: * Aggressive or Precision frame Shotgun (MindbendeToil/Imperial Decree/DRB/Retold Tale) * Erentil or Wizened Rebuke * Beloved/Twilight Oath/Supremacy/Revoker
Heavy Weapons: * Wardcliff * Truth * PotG * Any rocket launcher
Subclasses: * Hunter - middle void, middle or bottom arc * Titan - bottom void or bottom arc * Warlock - top arc or bottom solar
Exotics: * Stompees for Hunter * OEM or Antaeus Wards for Titan * Transversive Steps for Warlock
Mods: * 3+ super mods * 1-2 paragon mods for hunter if desired * 1-2 grenade mods for stormcaller or sentinel if desired * Otherwise 5 super mods
Stats: * Minimum of 1 resilience with as little as possible (Titans min is 3 or 4 I think). The rest goes to mobility and/or recovery. I’d recommend 6+ mobility for most people, but some prefer a lower mobility and higher recovery.
I don’t really want to debate what else is meta or what’s the best or other specifics. But in my experience both playing and watching others play high comp, this is the meta.
For weapons, Luna and a shotgun is still the best and most versatile loadout for most people and most maps. Consider swapping to a pulse or scout instead of Luna (or a sniper instead of a shotgun) for larger maps. Especially for countdown, consider having at least one sniper on your team as being able to get a pick and play 4v3 puts your team at a huge advantage. Fusion rifles are also incredibly strong right now. You can basically treat one like your primary weapon and just use your actual primary to clean people up or shoot people past ~30m.
In the current meta supers are incredibly important. You want to use them frequently and make orbs for your teammates for them to pick up and vice versa. Try to use your super when the enemy team doesn’t have any supers ready or heavy ammo is about to be up. Coordinate with your teammates on who’s popping a super and when so you don’t double pop and your teammates can get heavy, map control, and shoot the enemies running away from you.
I’ve gotten some questions on why so little resilience so I’ll answer it here. You’re going to die to supers, heavy ammo, and special weapons a lot more than primaries. Your resilience won’t really matter against those things. Plus the primaries you do see in high comp (mostly NF) don’t get effected by resilience. And even the other ones that you’ll occasionally see, resilience doesn’t really change the TTK, it only requires more headshots instead of body shots. At this level most players will be hitting their headshots anyways. Resilience was much more important in Y1 when there was a lot of primary weapon uptime.
The only time I’d recommend a higher resilience is if you’re on a Titan with OEM (to supplement recovery) and prefer low mobility. 7+ resilience will cause Erentil to take 5 bolts instead of 4 and might occasionally make a shotgun need to hit an extra pellet out of the spread to kill you (10 pellets of the 12, instead of 9 of 12 for example), among a couple of other minor advantages. I still wouldn’t really recommend it as I think you get more overall usage out of high recovery, but I’ve seen some people in high comp make it work.
Controlling heavy ammo wins games. Titans can use their barricade to pull heavy even while the other team is laning it. Prioritize getting the heavy and preventing your opponents from getting it. Once you get it, use it and don’t die with it. I’d recommend using it quickly but if you’re running Wardcliff it’s not a bad idea to save a rocket for an opponents super.
Finding Teammates
One of the most important parts of hitting Legend is having quality teammates. And by quality I mean both skill and temperament. Unless you already have a large friends list filled with quality teammates, you’ll need to network to find some. You can do this both in-game and using LFG. You can solo queue with a decent amount of success until about 3.5k or so, then you’ll want to start forming a team. If you seem to gel with teammates when solo queuing, shoot them a message and ask if they’d like to team up.
As far as LFG goes, there are lots of LFG websites these days. I’ve personally had a lot of success with Xbox’s built in LFG system. LFG can get a bad rep at times which is understandable. Some people are toxic, can tilt easily, blame teammates, complain all the time, not very skilled, etc. You obviously want to avoid these types of people and instead find teammates that are skilled, chill, encouraging and fun to play with. The best way to do this is to host the LFG group yourself by making the post and weeding people out. I’m not going to debate if/how important KD is to determine someone’s skill and if/what minimum you should ask for, use your own discretion here.
Once you get a team, just start playing. It might take a game or two for everyone to start to feel more comfortable with one another based on playstyles, tendencies, personalities, communication, etc. If things are going well after 4 or 5 games, keep playing. If they keep going well, add them to your friends list and ask them to do the same. If the games are not going well, you don’t seem to be playing together well as a team, and/or your personalities don’t seem to fit, consider politely excusing yourself and forming a new group. There’s absolutely nothing wrong with doing this. Sometimes the team is just not a good fit for whatever reason, it’s best for everyone to just move on with no hard feelings.
And by games going well I don’t necessarily mean winning. Are you guys teamshotting well? Baiting and switching effectively? Controlling the power ammo? Timing super usage? Moving together as a team? Playing complimentary angels and watching each other’s back? All of these things are good signs of a team working well. One of the best indicators is the number of assists you’re getting as a team (these can be looked up on any 3rd party website).
If your team is playing well together over a long session, like I said, add them and ask if they’ll do the same. Next time you get on, ask if they want to play before looking for a group via LFG. Sometimes they’ll even have friends that want to play as well which is great! Add anyone and everyone you play well with and seem to be on the same page with both in-game and personality wise. Rinse and repeat and you’ll have a solid list of friends to play Comp with. If you keep networking you can grow your friends list very quickly and effectively. You can also use Discord to schedule comp sessions.
The best way to attract good teammates is to be the best teammate you can. Be the teammate that you’d want on your team every single game and make things easy on your teammates. Hype them up for making good plays and encourage them if they make a bad one. Team shoot, make good callouts, don’t tilt, etc. Anything you’d look for in a good teammate, try to do that yourself and you’ll attract some great people to play with.
Always warm up before playing Comp and make sure your teammates have too. Rumble or QP is fine, but even a quick 10 minute private match rumble with your comp team can help warm up and build some camaraderie.
Closing Thoughts
Reaching Legend in Comp is seen by most as a daunting task and not how it should be seen - a huge accomplishment. Most people won’t even attempt to try for a variety of reasons ranging from pride to not enough reward to time and effort. High Comp is very challenging and honestly a much different game than QP or low Comp. It can be frustrating and stressful. But if you think of it as playing to improve and become the best player you can instead of just hitting Legend, then it’ll be very well worth it. Drastically improving as a player and as a result eventually hitting Legend is by far the best feeling in the entire game.
You might not even get there this season but that’s okay! But by having an improvement mindset and improving as a player, you’ll have a leg up next season - just stick with it and you’ll get there.
My final parting piece of advice is to just enjoy the journey. You’ll lose some close games and you’ll win some close games. You’ll get blown out by streamers or recovs and you’ll surprise yourself and beat some teams that are much better than you. Don’t sweat any of the losses, just enjoy playing the game. At the end of the day, this is a video game that we all play for fun.
One thing to keep in mind, especially once you get past 5k and are making that final push, you’re playing against some of the best players in the world and many of them play Comp for a living or it’s literally all they do. For most of us this is just one of many hobbies that we do for fun in our spare time, so don’t get too upset when you lose to these teams.
Thanks for reading - good luck and have fun! I’d be happy to answer any questions that anybody has.
Cheers!

A guide on hitting Legend rank in Comp

I’ve been seeing a lot of posts lately asking for tips for hitting Legend in Comp so I figured I’d put together a brief guide for anyone that’s interested. I’m happy to see so many people interested in hitting Legend! A lot of this post applies to comp in general, not necessarily just hitting Legend.
Intro
First off, a little bit about me. I never played D1 so I had a rough first few months of D2Y1 (and a rough first few weeks of Y2 with the new special weapon uptime and TTK) as it was my first time with the Destiny franchise. But even then had a blast in Crucible and I always wanted to get better. I’m also an extremely competitive person so that helped fuel my desire for improvement.
I play on Xbox and I just got my Unbroken title this season so I’ve been to Legend 3 times (S4, S6, S7). I’ve learned a ton along the way and hitting Legend each season has been easier and more enjoyable than the previous one for a variety of reasons that I’ll share in this post.
Improvement Mindset
While your end goal is to hit Legend, focusing on this binary goal isn’t a good idea. A better approach is to think of playing Comp with the main goal of improving both as a player and a team. With this more open and long-term mindset, you will improve rapidly as a player, win more often, and have a much more enjoyable experience as a result.
When you focus on something as binary as hitting a certain rank, every game or even decision within a game starts to feel tense and you put an immense amount of artificial pressure on yourself. This often builds over the course of a game. Even if it’s subconscious, it will effect your play. You’ll play too passive, too aggressive, and/or make bad decisions. Your brain will be too wrapped up playing out the consequences of failure to focus on what you should be doing to give yourself and your team the best chances of winning. It’s been scientifically proven both in real sports and in E-Sports that tension leads to poor performance.
Instead, take every engagement and every game as an opportunity to learn something and to improve. You WILL start getting your ass kicked at some point, it’s just a matter of when. It might be at 3k and it might not be until 5k, but at some point it’ll happen. And when it does, the best thing to do is to record your gameplay and watch it back.
Gameplay Review
You can easily record your gameplay via Twitch by streaming and having it save past broadcasts. Then you can watch your gameplay there, or you can take it a step further and download your gameplay and run it through a free video editing program such as DaVinci Resolve or iMovie. The advantage of doing it this way is you can better control the playback and even view it frame-by-frame.
I’d recommend picking a game that you performed poorly and watch it once all the way through and take some mental notes. Then I’d watch it again, noting each death with why you died and what you could have done better to either kill your opponent first or escape safely. Even if you died to a Wardcliff or a solo super, write down something you could have done differently to prevent dying. Then categorize and tally them the best you can. The most frequent ones are what you should focus on getting better at. This can be during your next Comp session or QP/Rumble.
The reason reviewing your gameplay is so important is it’ll help speed up your rate of improvement and help you get past your current plateau a bit faster. Games in high comp tend to be very fast paced so it can be hard to think about or remember exactly what happened. Or what you think happened in the moment wasn’t what really happened and the gameplay review will show you this.
While it’s certainly possible to improve naturally and over time, recording and reviewing your gameplay will make you improve faster.
Playing the meta
A lot of people seem reluctant to use meta loadouts for whatever reason. I think most of it boils down to either wanting to be unique, or having a superiority complex by refusing to use certain good or easy to use weapons and strategies because they’re “cheap” or too easy. Throw all of this out the window.
There’s nothing cheap in Comp (other than DDoSing which is actual cheating and we won’t discuss it). There’s nothing that takes “no skill” to use. If it’s in the game then it’s fair game to be used as much and as effectively as possible. Everything has a counter. If you don’t believe this then you probably have a scrub mentality and it’s going to hold you back. There are some great posts about scrub mentality on this very sub.
Meta loadouts or weapons are usually the perfect cross section of both lethality and ease-of-use - USE THEM. This is the time and the place. Your opponents are trying to win at all costs and so should you.
I don’t want to go too much into detail here or debate here, but in general these are the best options for high comp on Console (4k+). They’re ranked in terms of effectiveness, so it’s probably better to improve with something at the top of the list than use something at the bottom.
Primary Weapons: * Luna (NF if you have it already) * Adaptive or Aggressive pulse rifles * Ace/Thorn/TLW * Very well rolled Legendary HC * Jade Rabbit/Mida/Polaris Lance (large maps only)
Special Weapons: * Aggressive or Precision frame Shotgun (MindbendeToil/Imperial Decree/DRB/Retold Tale) * Erentil or Wizened Rebuke * Beloved/Twilight Oath/Supremacy/Revoker
Heavy Weapons: * Wardcliff * Truth * PotG * Any rocket launcher
Subclasses: * Hunter - middle void, middle or bottom arc * Titan - bottom void or bottom arc * Warlock - top arc or bottom solar
Exotics: * Stompees for Hunter * OEM or Antaeus Wards for Titan * Transversive Steps for Warlock
Mods: * 3+ super mods * 1-2 paragon mods for hunter if desired * 1-2 grenade mods for stormcaller or sentinel if desired * Otherwise 5 super mods
Stats: * Minimum of 1 resilience with as little as possible (Titans min is 3 or 4 I think). The rest goes to mobility and/or recovery. I’d recommend 6+ mobility for most people, but some prefer a lower mobility and higher recovery.
I don’t really want to debate what else is meta or what’s the best or other specifics. But in my experience both playing and watching others play high comp, this is the meta.
For weapons, Luna and a shotgun is still the best and most versatile loadout for most people and most maps. Consider swapping to a pulse or scout instead of Luna (or a sniper instead of a shotgun) for larger maps. Especially for countdown, consider having at least one sniper on your team as being able to get a pick and play 4v3 puts your team at a huge advantage. Fusion rifles are also incredibly strong right now. You can basically treat one like your primary weapon and just use your actual primary to clean people up or shoot people past ~30m.
In the current meta supers are incredibly important. You want to use them frequently and make orbs for your teammates for them to pick up and vice versa. Try to use your super when the enemy team doesn’t have any supers ready or heavy ammo is about to be up. Coordinate with your teammates on who’s popping a super and when so you don’t double pop and your teammates can get heavy, map control, and shoot the enemies running away from you.
I’ve gotten some questions on why so little resilience so I’ll answer it here. You’re going to die to supers, heavy ammo, and special weapons a lot more than primaries. Your resilience won’t really matter against those things. Plus the primaries you do see in high comp (mostly NF) don’t get effected by resilience. And even the other ones that you’ll occasionally see, resilience doesn’t really change the TTK, it only requires more headshots instead of body shots. At this level most players will be hitting their headshots anyways. Resilience was much more important in Y1 when there was a lot of primary weapon uptime.
The only time I’d recommend a higher resilience is if you’re on a Titan with OEM (to supplement recovery) and prefer low mobility. 7+ resilience will cause Erentil to take 5 bolts instead of 4 and might occasionally make a shotgun need to hit an extra pellet out of the spread to kill you (10 pellets of the 12, instead of 9 of 12 for example), among a couple of other minor advantages. I still wouldn’t really recommend it as I think you get more overall usage out of high recovery, but I’ve seen some people in high comp make it work.
Controlling heavy ammo wins games. Titans can use their barricade to pull heavy even while the other team is laning it. Prioritize getting the heavy and preventing your opponents from getting it. Once you get it, use it and don’t die with it. I’d recommend using it quickly but if you’re running Wardcliff it’s not a bad idea to save a rocket for an opponents super.
Finding Teammates
One of the most important parts of hitting Legend is having quality teammates. And by quality I mean both skill and temperament. Unless you already have a large friends list filled with quality teammates, you’ll need to network to find some. You can do this both in-game and using LFG. You can solo queue with a decent amount of success until about 3.5k or so, then you’ll want to start forming a team. If you seem to gel with teammates when solo queuing, shoot them a message and ask if they’d like to team up.
As far as LFG goes, there are lots of LFG websites these days. I’ve personally had a lot of success with Xbox’s built in LFG system. LFG can get a bad rep at times which is understandable. Some people are toxic, can tilt easily, blame teammates, complain all the time, not very skilled, etc. You obviously want to avoid these types of people and instead find teammates that are skilled, chill, encouraging and fun to play with. The best way to do this is to host the LFG group yourself by making the post and weeding people out. I’m not going to debate if/how important KD is to determine someone’s skill and if/what minimum you should ask for, use your own discretion here.
Once you get a team, just start playing. It might take a game or two for everyone to start to feel more comfortable with one another based on playstyles, tendencies, personalities, communication, etc. If things are going well after 4 or 5 games, keep playing. If they keep going well, add them to your friends list and ask them to do the same. If the games are not going well, you don’t seem to be playing together well as a team, and/or your personalities don’t seem to fit, consider politely excusing yourself and forming a new group. There’s absolutely nothing wrong with doing this. Sometimes the team is just not a good fit for whatever reason, it’s best for everyone to just move on with no hard feelings.
And by games going well I don’t necessarily mean winning. Are you guys teamshotting well? Baiting and switching effectively? Controlling the power ammo? Timing super usage? Moving together as a team? Playing complimentary angels and watching each other’s back? All of these things are good signs of a team working well. One of the best indicators is the number of assists you’re getting as a team (these can be looked up on any 3rd party website).
If your team is playing well together over a long session, like I said, add them and ask if they’ll do the same. Next time you get on, ask if they want to play before looking for a group via LFG. Sometimes they’ll even have friends that want to play as well which is great! Add anyone and everyone you play well with and seem to be on the same page with both in-game and personality wise. Rinse and repeat and you’ll have a solid list of friends to play Comp with. If you keep networking you can grow your friends list very quickly and effectively. You can also use Discord to schedule comp sessions.
The best way to attract good teammates is to be the best teammate you can. Be the teammate that you’d want on your team every single game and make things easy on your teammates. Hype them up for making good plays and encourage them if they make a bad one. Team shoot, make good callouts, don’t tilt, etc. Anything you’d look for in a good teammate, try to do that yourself and you’ll attract some great people to play with.
Always warm up before playing Comp and make sure your teammates have too. Rumble or QP is fine, but even a quick 10 minute private match rumble with your comp team can help warm up and build some camaraderie.
Closing Thoughts
Reaching Legend in Comp is seen by most as a daunting task and not how it should be seen - a huge accomplishment. Most people won’t even attempt to try for a variety of reasons ranging from pride to not enough reward to time and effort. High Comp is very challenging and honestly a much different game than QP or low Comp. It can be frustrating and stressful. But if you think of it as playing to improve and become the best player you can instead of just hitting Legend, then it’ll be very well worth it. Drastically improving as a player and as a result eventually hitting Legend is by far the best feeling in the entire game.
You might not even get there this season but that’s okay! But by having an improvement mindset and improving as a player, you’ll have a leg up next season - just stick with it and you’ll get there.
My final parting piece of advice is to just enjoy the journey. You’ll lose some close games and you’ll win some close games. You’ll get blown out by streamers or recovs and you’ll surprise yourself and beat some teams that are much better than you. Don’t sweat any of the losses, just enjoy playing the game. At the end of the day, this is a video game that we all play for fun.
One thing to keep in mind, especially once you get past 5k and are making that final push, you’re playing against some of the best players in the world and many of them play Comp for a living or it’s literally all they do. For most of us this is just one of many hobbies that we do for fun in our spare time, so don’t get too upset when you lose to these teams.
Thanks for reading - good luck and have fun! I’d be happy to answer any questions that anybody has.
Cheers!

FlowCards: A Declarative Framework for Development of Ergo dApps

Introduction
ErgoScript is the smart contract language used by the Ergo blockchain. While it has concise syntax adopted from Scala/Kotlin, it still may seem confusing at first because conceptually ErgoScript is quite different compared to conventional languages which we all know and love. This is because Ergo is a UTXO based blockchain, whereas smart contracts are traditionally associated with account based systems like Ethereum. However, Ergo's transaction model has many advantages over the account based model and with the right approach it can even be significantly easier to develop Ergo contracts than to write and debug Solidity code.
Below we will cover the key aspects of the Ergo contract model which makes it different:

Paradigm

The account model of Ethereum is imperative. This means that the typical task of sending coins from Alice to Bob requires changing the balances in storage as a series of operations. Ergo's UTXO based programming model on the other hand is declarative. ErgoScript contracts specify conditions for a transaction to be accepted by the blockchain (not changes to be made in the storage state as result of the contract execution).

Scalability

In the account model of Ethereum both storage changes and validity checks are performed on-chain during code execution. In contrast, Ergo transactions are created off-chain and only validation checks are performed on-chain thus reducing the amount of operations performed by every node on the network. In addition, due to immutability of the transaction graph, various optimization strategies are possible to improve throughput of transactions per second in the network. Light verifying nodes are also possible thus further facilitating scalability and accessibility of the network.

Shared state

The account-based model is reliant on shared mutable state which is known to lead to complex semantics (and subtle million dollar bugs) in the context of concurrent/ distributed computation. Ergo's model is based on an immutable graph of transactions. This approach, inherited from Bitcoin, plays well with the concurrent and distributed nature of blockchains and facilitates light trustless clients.

Expressive Power

Ethereum advocated execution of a turing-complete language on the blockchain. It theoretically promised unlimited potential, however in practice severe limitations came to light from excessive blockchain bloat, subtle multi-million dollar bugs, gas costs which limit contract complexity, and other such problems. Ergo on the flip side extends UTXO to enable turing-completeness while limiting the complexity of the ErgoScript language itself. The same expressive power is achieved in a different and more semantically sound way.
With the all of the above points, it should be clear that there are a lot of benefits to the model Ergo is using. In the rest of this article I will introduce you to the concept of FlowCards - a dApp developer component which allows for designing complex Ergo contracts in a declarative and visual way.

From Imperative to Declarative

In the imperative programming model of Ethereum a transaction is a sequence of operations executed by the Ethereum VM. The following Solidity function implements a transfer of tokens from sender to receiver . The transaction starts when sender calls this function on an instance of a contract and ends when the function returns.

// Sends an amount of existing coins from any caller to an address function send(address receiver, uint amount) public { require(amount <= balances[msg.sender], "Insufficient balance."); balances[msg.sender] -= amount; balances[receiver] += amount; emit Sent(msg.sender, receiver, amount); }

The function first checks the pre-conditions, then updates the storage (i.e. balances) and finally publishes the post-condition as the Sent event. The gas which is consumed by the transaction is sent to the miner as a reward for executing this transaction.
Unlike Ethereum, a transaction in Ergo is a data structure holding a list of input coins which it spends and a list of output coins which it creates preserving the total balances of ERGs and tokens (in which Ergo is similar to Bitcoin).
Turning back to the example above, since Ergo natively supports tokens, therefore for this specific example of sending tokens we don't need to write any code in ErgoScript. Instead we need to create the ‘send’ transaction shown in the following figure, which describes the same token transfer but declaratively.
https://preview.redd.it/sxs3kesvrsv41.png?width=1348&format=png&auto=webp&s=582382bc26912ff79114d831d937d94b6988e69f
The picture visually describes the following steps, which the network user needs to perform:

Select unspent sender's boxes, containing in total tB >= amount of tokens and B >= txFee + minErg ERGs.
Create an output target box which is protected by the receiver public key with minErg ERGs and amount of T tokens.
Create one fee output protected by the minerFee contract with txFee ERGs.
Create one change output protected by the sender public key, containing B - minErg - txFee ERGs and tB - amount of T tokens.
Create a new transaction, sign it using the sender's secret key and send to the Ergo network.

What is important to understand here is that all of these steps are preformed off-chain (for example using Appkit Transaction API) by the user's application. Ergo network nodes don't need to repeat this transaction creation process, they only need to validate the already formed transaction. ErgoScript contracts are stored in the inputs of the transaction and check spending conditions. The node executes the contracts on-chain when the transaction is validated. The transaction is valid if all of the conditions are satisfied.
Thus, in Ethereum when we “send amount from sender to recipient” we are literally editing balances and updating the storage with a concrete set of commands. This happens on-chain and thus a new transaction is also created on-chain as the result of this process.
In Ergo (as in Bitcoin) transactions are created off-chain and the network nodes only verify them. The effects of the transaction on the blockchain state is that input coins (or Boxes in Ergo's parlance) are removed and output boxes are added to the UTXO set.
In the example above we don't use an ErgoScript contract but instead assume a signature check is used as the spending pre-condition. However in more complex application scenarios we of course need to use ErgoScript which is what we are going to discuss next.

From Changing State to Checking Context

In the send function example we first checked the pre-condition (require(amount <= balances[msg.sender],...) ) and then changed the state (i.e. update balances balances[msg.sender] -= amount ). This is typical in Ethereum transactions. Before we change anything we need to check if it is valid to do so.
In Ergo, as we discussed previously, the state (i.e. UTXO set of boxes) is changed implicitly when a valid transaction is included in a block. Thus we only need to check the pre-conditions before the transaction can be added to the block. This is what ErgoScript contracts do.
It is not possible to “change the state” in ErgoScript because it is a language to check pre-conditions for spending coins. ErgoScript is a purely functional language without side effects that operates on immutable data values. This means all the inputs, outputs and other transaction parameters available in a script are immutable. This, among other things, makes ErgoScript a very simple language that is easy to learn and safe to use. Similar to Bitcoin, each input box contains a script, which should return the true value in order to 1) allow spending of the box (i.e. removing from the UTXO set) and 2) adding the transaction to the block.
If we are being pedantic, it is therefore incorrect (strictly speaking) to think of ErgoScript as the language of Ergo contracts, because it is the language of propositions (logical predicates, formulas, etc.) which protect boxes from “illegal” spending. Unlike Bitcoin, in Ergo the whole transaction and a part of the current blockchain context is available to every script. Therefore each script may check which outputs are created by the transaction, their ERG and token amounts (we will use this capability in our example DEX contracts), current block number etc.
In ErgoScript you define the conditions of whether changes (i.e. coin spending) are allowed to happen in a given context. This is in contrast to programming the changes imperatively in the code of a contract.
While Ergo's transaction model unlocks a whole range of applications like (DEX, DeFi Apps, LETS, etc), designing contracts as pre-conditions for coin spending (or guarding scripts) directly is not intuitive. In the next sections we will consider a useful graphical notation to design contracts declaratively using FlowCard Diagrams, which is a visual representation of executable components (FlowCards).
FlowCards aim to radically simplify dApp development on the Ergo platform by providing a high-level declarative language, execution runtime, storage format and a graphical notation.
We will start with a high level of diagrams and go down to FlowCard specification.

FlowCard Diagrams

The idea behind FlowCard diagrams is based on the following observations: 1) An Ergo box is immutable and can only be spent in the transaction which uses it as an input. 2) We therefore can draw a flow of boxes through transactions, so that boxes flowing in to the transaction are spent and those flowing out are created and added to the UTXO. 3) A transaction from this perspective is simply a transformer of old boxes to the new ones preserving the balances of ERGs and tokens involved.
The following figure shows the main elements of the Ergo transaction we've already seen previously (now under the name of FlowCard Diagram).
https://preview.redd.it/06aqkcd1ssv41.png?width=1304&format=png&auto=webp&s=106eda730e0526919aabd5af9596b97e45b69777
There is a strictly defined meaning (semantics) behind every element of the diagram, so that the diagram is a visual representation (or a view) of the underlying executable component (called FlowCard).
The FlowCard can be used as a reusable component of an Ergo dApp to create and initiate the transaction on the Ergo blockchain. We will discuss this in the coming sections.
Now let's look at the individual pieces of the FlowCard diagram one by one.

1. Name and Parameters

Each flow card is given a name and a list of typed parameters. This is similar to a template with parameters. In the above figure we can see the Send flow card which has five parameters. The parameters are used in the specification.

2. Contract Wallet

This is a key element of the flow card. Every box has a guarding script. Often it is the script that checks a signature against a public key. This script is trivial in ErgoScript and is defined like the def pk(pubkey: Address) = { pubkey } template where pubkey is a parameter of the type Address . In the figure, the script template is applied to the parameter pk(sender) and thus a concrete wallet contract is obtained. Therefore pk(sender) and pk(receiver) yield different scripts and represent different wallets on the diagram, even though they use the same template.
Contract Wallet contains a set of all UTXO boxes which have a given script derived from the given script template using flow card parameters. For example, in the figure, the template is pk and parameter pubkey is substituted with the `sender’ flow card parameter.

3. Contract

Even though a contract is a property of a box, on the diagram we group the boxes by their contracts, therefore it looks like the boxes belong to the contracts, rather than the contracts belong to the boxes. In the example, we have three instantiated contracts pk(sender) , pk(receiver) and minerFee . Note, that pk(sender) is the instantiation of the pk template with the concrete parameter sender and minerFee is the instantiation of the pre-defined contract which protects the miner reward boxes.

4. Box name

In the diagram we can give each box a name. Besides readability of the diagram, we also use the name as a synonym of a more complex indexed access to the box in the contract. For example, change is the name of the box, which can also be used in the ErgoScript conditions instead of OUTPUTS(2) . We also use box names to associate spending conditions with the boxes.

5. Boxes in the wallet

In the diagram, we show boxes (darker rectangles) as belonging to the contract wallets (lighter rectangles). Each such box rectangle is connected with a grey transaction rectangle by either orange or green arrows or both. An output box (with an incoming green arrow) may include many lines of text where each line specifies a condition which should be checked as part of the transaction. The first line specifies the condition on the amount of ERG which should be placed in the box. Other lines may take one of the following forms:

amount: TOKEN - the box should contain the given amount of the given TOKEN
R == value - the box should contain the given value of the given register R
boxName ? condition - the box named boxName should check condition in its script.

We discuss these conditions in the sections below.

6. Amount of ERGs in the box

Each box should store a minimum amount of ERGs. This is checked when the creating transaction is validated. In the diagram the amount of ERGs is always shown as the first line (e.g. B: ERG or B - minErg - txFee ). The value type ascription B: ERG is optional and may be used for readability. When the value is given as a formula, then this formula should be respected by the transaction which creates the box.
It is important to understand that variables like amount and txFee are not named properties of the boxes. They are parameters of the whole diagram and representing some amounts. Or put it another way, they are shared parameters between transactions (e.g. Sell Order and Swap transactions from DEX example below share the tAmt parameter). So the same name is tied to the same value throughout the diagram (this is where the tooling would help a lot). However, when it comes to on-chain validation of those values, only explicit conditions which are marked with ? are transformed to ErgoScript. At the same time, all other conditions are ensured off-chain during transaction building (for example in an application using Appkit API) and transaction validation when it is added to the blockchain.

7. Amount of T token

A box can store values of many tokens. The tokens on the diagram are named and a value variable may be associated with the token T using value: T expression. The value may be given by formula. If the formula is prefixed with a box name like boxName ? formula , then it is should also be checked in the guarding script of the boxName box. This additional specification is very convenient because 1) it allows to validate the visual design automatically, and 2) the conditions specified in the boxes of a diagram are enough to synthesize the necessary guarding scripts. (more about this below at “From Diagrams To ErgoScript Contracts”)

8. Tx Inputs

Inputs are connected to the corresponding transaction by orange arrows. An input arrow may have a label of the following forms:

[email protected] - optional name with an index i.e. [email protected] or u/2 . This is a property of the target endpoint of the arrow. The name is used in conditions of related boxes and the index is the position of the corresponding box in the INPUTS collection of the transaction.
!action - is a property of the source of the arrow and gives a name for an alternative spending path of the box (we will see this in DEX example)

Because of alternative spending paths, a box may have many outgoing orange arrows, in which case they should be labeled with different actions.

9. Transaction

A transaction spends input boxes and creates output boxes. The input boxes are given by the orange arrows and the labels are expected to put inputs at the right indexes in INPUTS collection. The output boxes are given by the green arrows. Each transaction should preserve a strict balance of ERG values (sum of inputs == sum of outputs) and for each token the sum of inputs >= the sum of outputs. The design diagram requires an explicit specification of the ERG and token values for all of the output boxes to avoid implicit errors and ensure better readability.

10. Tx Outputs

Outputs are connected to the corresponding transaction by green arrows. An output arrow may have a label of the following [email protected] , where an optional name is accompanied with an index i.e. [email protected] or u/2 . This is a property of the source endpoint of the arrow. The name is used in conditions of the related boxes and the index is the position of the corresponding box in the OUTPUTS collection of the transaction.

Example: Decentralized Exchange (DEX)

Now let's use the above described notation to design a FlowCard for a DEX dApp. It is simple enough yet also illustrates all of the key features of FlowCard diagrams which we've introduced in the previous section.
The dApp scenario is shown in the figure below: There are three participants (buyer, seller and DEX) of the DEX dApp and five different transaction types, which are created by participants. The buyer wants to swap ergAmt of ERGs for tAmt of TID tokens (or vice versa, the seller wants to sell TID tokens for ERGs, who sends the order first doesn't matter). Both the buyer and the seller can cancel their orders any time. The DEX off-chain matching service can find matching orders and create the Swap transaction to complete the exchange.
The following diagram fully (and formally) specifies all of the five transactions that must be created off-chain by the DEX dApp. It also specifies all of the spending conditions that should be verified on-chain.

https://preview.redd.it/piogz0v9ssv41.png?width=1614&format=png&auto=webp&s=e1b503a635ad3d138ef91e2f0c3b726e78958646
Let's discuss the FlowCard diagram and the logic of each transaction in details:

Buy Order Transaction

A buyer creates a Buy Order transaction. The transaction spends E amount of ERGs (which we will write E: ERG ) from one or more boxes in the pk(buyer) wallet. The transaction creates a bid box with ergAmt: ERG protected by the buyOrder script. The buyOrder script is synthesized from the specification (see below at “From Diagrams To ErgoScript Contracts”) either manually or automatically by a tool. Even though we don't need to define the buyOrder script explicitly during designing, at run time the bid box should contain the buyOrder script as the guarding proposition (which checks the box spending conditions), otherwise the conditions specified in the diagram will not be checked.
The change box is created to make the input and output sums of the transaction balanced. The transaction fee box is omitted because it can be added automatically by the tools. In practice, however, the designer can add the fee box explicitly to the a diagram. It covers the cases of more complex transactions (like Swap) where there are many ways to pay the transaction fee.

Cancel Buy, Cancel Sell Transactions

At any time, the buyer can cancel the order by sending CancelBuy transaction. The transaction should satisfy the guarding buyOrder contract which protects the bid box. As you can see on the diagram, both the Cancel and the Swap transactions can spend the bid box. When a box has spending alternatives (or spending paths) then each alternative is identified by a unique name prefixed with ! (!cancel and !swap for the bid box). Each alternative path has specific spending conditions. In our example, when the Cancel Buy transaction spends the bid box the ?buyer condition should be satisfied, which we read as “the signature for the buyer address should be presented in the transaction”. Therefore, only buyer can cancel the buy order. This “signature” condition is only required for the !cancel alternative spending path and not required for !swap .

Sell Order Transaction

The Sell Order transaction is similar to the BuyOrder in that it deals with tokens in addition to ERGs. The transaction spends E: ERG and T: TID tokens from seller's wallet (specified as pk(seller) contract). The two outputs are ask and change . The change is a standard box to balance transaction. The ask box keeps tAmt: TID tokens for the exchange and minErg: ERG - the minimum amount of ERGs required in every box.

Swap Transaction

This is a key transaction in the DEX dApp scenario. The transaction has several spending conditions on the input boxes and those conditions are included in the buyOrder and sellOrder scripts (which are verified when the transaction is added to the blockchain). However, on the diagram those conditions are not specified in the bid and ask boxes, they are instead defined in the output boxes of the transaction.
This is a convention for improved usability because most of the conditions relate to the properties of the output boxes. We could specify those properties in the bid box, but then we would have to use more complex expressions.
Let's consider the output created by the arrow labeled with [email protected] . This label tells us that the output is at the index 0 in the OUTPUTS collection of the transaction and that in the diagram we can refer to this box by the buyerOut name. Thus we can label both the box itself and the arrow to give the box a name.
The conditions shown in the buyerOut box have the form bid ? condition , which means they should be verified on-chain in order to spend the bid box. The conditions have the following meaning:

tAmt: TID requires the box to have tAmt amount of TID token
R4 == bid.id requires R4 register in the box to be equal to id of the bid box.
script == buyer requires the buyerOut box to have the script of the wallet where it is located on the diagram, i.e. pk(buyer)

Similar properties are added to the sellerOut box, which is specified to be at index 1 and the name is given to it using the label on the box itself, rather than on the arrow.
The Swap transaction spends two boxes bid and ask using the !swap spending path on both, however unlike !cancel the conditions on the path are not specified. This is where the bid ? and ask ? prefixes come into play. They are used so that the conditions listed in the buyerOut and sellerOut boxes are moved to the !swap spending path of the bid and ask boxes correspondingly.
If you look at the conditions of the output boxes, you will see that they exactly specify the swap of values between seller's and buyer's wallets. The buyer gets the necessary amount of TID token and seller gets the corresponding amount of ERGs. The Swap transaction is created when there are two matching boxes with buyOrder and sellOrder contracts.

From Diagrams To ErgoScript Contracts

What is interesting about FlowCard specifications is that we can use them to automatically generate the necessary ErgoTree scripts. With the appropriate tooling support this can be done automatically, but with the lack of thereof, it can be done manually. Thus, the FlowCard allows us to capture and visually represent all of the design choices and semantic details of an Ergo dApp.
What we are going to do next is to mechanically create the buyOrder contract from the information given in the DEX flow card.
Recall that each script is a proposition (boolean valued expression) which should evaluate to true to allow spending of the box. When we have many conditions to be met at the same time we can combine them in a logical formula using the AND binary operation, and if we have alternatives (not necessarily exclusive) we can put them into the OR operation.
The buyOrder box has the alternative spending paths !cancel and !swap . Thus the ErgoScript code should have OR operation with two arguments - one for each spending path.

/** buyOrder contract */ { val cancelCondition = {} val swapCondition = {} cancelCondition || swapCondition }

The formula for the cancelCondition expression is given in the !cancel spending path of the buyOrder box. We can directly include it in the script.

/** buyOrder contract */ { val cancelCondition = { buyer } val swapCondition = {} cancelCondition || swapCondition }

For the !swap spending path of the buyOrder box the conditions are specified in the buyerOut output box of the Swap transaction. If we simply include them in the swapCondition then we get a syntactically incorrect script.

/** buyOrder contract */ { val cancelCondition = { buyer } val swapCondition = { tAmt: TID && R4 == bid.id && @contract } cancelCondition || swapCondition }

We can however translate the conditions from the diagram syntax to ErgoScript expressions using the following simple rules

[email protected] ==> val buyerOut = OUTPUTS(0)
tAmt: TID ==> tid._2 == tAmt where tid = buyerOut.tokens(TID)
R4 == bid.id ==> R4 == SELF.id where R4 = buyerOut.R4[Coll[Byte]].get
script == buyer ==> buyerOut.propositionBytes == buyer.propBytes

Note, in the diagram TID represents a token id, but ErgoScript doesn't have access to the tokens by the ids so we cannot write tokens.getByKey(TID) . For this reason, when the diagram is translated into ErgoScript, TID becomes a named constant of the index in tokens collection of the box. The concrete value of the constant is assigned when the BuyOrder transaction with the buyOrder box is created. The correspondence and consistency between the actual tokenId, the TID constant and the actual tokens of the buyerOut box is ensured by the off-chain application code, which is completely possible since all of the transactions are created by the application using FlowCard as a guiding specification. This may sound too complicated, but this is part of the translation from diagram specification to actual executable application code, most of which can be automated.
After the transformation we can obtain a correct script which checks all the required preconditions for spending the buyOrder box.

/** buyOrder contract */ def DEX(buyer: Addrss, seller: Address, TID: Int, ergAmt: Long, tAmt: Long) { val cancelCondition: SigmaProp = { buyer } // verify buyer's sig (ProveDlog) val swapCondition = OUTPUTS.size > 0 && { // securing OUTPUTS access val buyerOut = OUTPUTS(0) // from [email protected] buyerOut.tokens.size > TID && { // securing tokens access val tid = buyerOut.tokens(TID) val regR4 = buyerOut.R4[Coll[Byte]] regR4.isDefined && { // securing R4 access val R4 = regR4.get tid._2 == tAmt && // from tAmt: TID R4 == SELF.id && // from R4 == bid.id buyerOut.propositionBytes == buyer.propBytes // from script == buyer } } } cancelCondition || swapCondition }

A similar script for the sellOrder box can be obtained using the same translation rules. With the help of the tooling the code of contracts can be mechanically generated from the diagram specification.

Conclusions

Declarative programming models have already won the battle against imperative programming in many application domains like Big Data, Stream Processing, Deep Learning, Databases, etc. Ergo is pioneering the declarative model of dApp development as a better and safer alternative to the now popular imperative model of smart contracts.
The concept of FlowCard shifts the focus from writing ErgoScript contracts to the overall flow of values (hence the name), in such a way, that ErgoScript can always be generated from them. You will never need to look at the ErgoScript code once the tooling is in place.
Here are the possible next steps for future work:

Storage format for FlowCard Spec and the corresponding EIP standardized file format (Json/XML/Protobuf). This will allow various tools (Diagram Editor, Runtime, dApps etc) to create and use *.flowcard files.
FlowCard Viewer, which can generate the diagrams from *.flowcard files.
FlowCard Runtime, which can run *.flowcard files, create and send transactions to Ergo network.
FlowCard Designer Tool, which can simplify development of complex diagrams . This will make designing and validation of Ergo contracts a pleasant experience, more like drawing rather than coding. In addition, the correctness of the whole dApp scenario can be verified and controlled by the tooling.

Escape from Tarkov - New Player Guide!

Introduction

NEW VERSION: https://www.reddit.com/EscapefromTarkov/comments/ffyynf/escape_from_tarkov_new_player_guide_20_75_pages/

Greetings, this is dumnem, also known as Theorchero, but you can call me Theo. I'm an experienced Tarkov player and I'm writing this guide to try and assist new Tarkov players learn the game, because it has one hell of a learning curve. We'll be going over a lot of different aspects of this guide, and it is going to be huge. Feel free to digest this in parts.
Additionally, this is a work in progress. I will write as much as I can in one Reddit post, but subsequent parts will be in additional comments. Google Docs Version (Note: Link is placeholder atm)
Disclaimer: I haven't played Tarkov regularly in a couple months. It's possible there has been extensive changes that I have not kept up with. If there is anything I have gotten wrong or may have omitted, please let me know.
This is Primarily directed towards Tarkov Novices. It hopefully includes everything you need to know to be able to go into a Raid equipped for success and to successfully extract with gear.
Want to play with friends? Want to have fun and learn Tarkov? Check out my discord here.

Changelog

WORKING ON UPDATING FOR .12, HANG TIGHT.
6/2:

Added Veritas chart
Fixed Vaseline/Star Balm stats, lab key card.
Adjusted formatting slightly, spelling adjustments.
Added additional resource, updated old ones.

Tarkov Overview - What is Escape from Tarkov?
Tarkov Resources - Useful links
Tarkov's Maps
Tarkov's Health System
Tarkov's Quest System and Progression
Tarkov's Hotkeys to Know
Getting Started
Player Scavs
New Player's loadouts - LL1 Traders
What to Loot - How to get the most money per slot
Stash Management - How to combat Gear Fear
What now?

Tarkov Overview - What is Escape from Tarkov?

Escape from Tarkov is a tactical, realistic, FPS with MMO elements developed by Battlestate Games. It is currently in closed Beta. The game features several maps in which your primary character, your PMC, goes into Raids in order to find and salvage loot and useful equipment to survive and thrive in Tarkov. Death is very punishing in Tarkov. If you die you lose everything you had on you when you die (with the exception of what's inside your Container and your melee weapon) including any equipment you brought with you or what you found inside the Raid.
Enemies can be players (PMCs) or 'Scavs' (Scavengers) that are either controlled by AI or by players. Unlike many shooters, AI enemies in Tarkov are deadly - they can and will kill you on sight.
It features beautiful and immersive environments, intricate and in-depth weapon modification system, a complex health system, attention to detail with loot placement, and options for combat. Do you want to play slow and stealthy, to avoid fights, or set up a deadly ambush on an unwary foe? Or do you prefer to raw combat, where only your quick wit, placements of shots, and tenaciousness determines who gets out alive? It's your Tarkov. You make the rules.

Tarkov Resources - Useful links

I take no credit or responsibility for any of the content in these links. To the best of my knowledge, these are updated consistently and are accurate, but user beware.

Quick-Reference Ammo Chart

Created by SirKilljoy, you can find it here. Note: As of 6/2/19, Outdated. Searching for updated file. Send if you have one!
An updated ammo chart can be found here. It's not fancy or quick reference though. Sorry!

Tarkov Wiki

Absolutely fantastic resource. You can visit them here.
It is a massive collection of everything that we players have been able to find.
They contain trades, user-created maps, lists of ammo, parts, weapons, loot, etc. If it's in the game, it's on the Wiki, somewhere.
I highly recommend opening the wiki page for the Map that you plan on raiding in.
Factory
Customs
Woods
Shoreline
Interchange
The Lab ('Labs')

Map Keys and You

Huge collection of all the keys in the game. These are also on the wiki, but this page has them all on one page, and tries to inform the user if the key is worth keeping or using.
Check it out here.
This section is open to revision. Mention me in a thread (or in the comments below) about a resource and I'll see about adding it here.

Tarkov's Weapon Compatibility Guide

Pretty self explanatory. Also includes a Key guide and a Mod guide.
Check it out here.

HUGE Reference Bible by Veritas

Courtesy of Veritas (Send me his reddit username?), It's located here. (Open in new tab.)
Contains: Detailed information about: Ammunition, Health, Firearms, Body Armor, Helmets, Rigs & Backpacks, Labs & Quest keys.

Tarkov's Maps

Tarkov features several maps - ranging from wide, beautiful vistas to ruined factory districts, to an abandoned laboratory where illegal experiments were being conducted. It is important to learn the maps you intend to play. In order to keep your gear, you must 'extract' at one of your designated exfiltration points. Not all exfils will be active every game, and some are conditional.

To see what extracts are available to you, double tap 'O' to show raid time and your exfils. If it has a ???? it might not be open.

You can load Raids in an OFFLINE status, which allows you to explore the map or practice against AI without losing gear.
You do not keep any EXP or gear you find in the OFFLINE Raid, though.
To access OFFLINE Raids, head into a Raid normally until you see this screen.
Simply check 'Enable OFFLINE mode for this Raid' and you're good to go! You even have a choice on whether or not to add AI. You can also control how many AI enemies spawn, fewer than normal or a great deal more! You can even make Scavs fight each other. (Framerates beware.)

Factory

Gate 3 Extract
A small, fast-paced map that was primarily created for PvP. Scavs spawn in all the time. Very close quarters, shotguns and SMGs tend to dominate here. PMCs can only access one Exit (Gate 3) without the Factory Exit Key. Good place to go if you need PMC kills as action is pretty much guaranteed. It is recommend NOT to bring in a lot of gear to Factory until you are experienced.

Customs

Extract map
A fairly large map that was recently expanded. Essentially, players spawn either on 'warehouse' or 'boiler (stacks)' side. If you see a large red warehouse near you (Customs Warehouse), then you spawned on the warehouse side. If you don't, you likely spawned near Boiler side.
The location for most quests in the game. Finding geared players here is very easy, so if you are low level attempting to complete early quests (like The Bronze Pocketwatch) it's recommend to AFK in the raid for 20 mins or so, as most players will have moved out of spawns and hit the 'hot' areas already, so it's less likely to be contested, so you may grab your quest item(s) and proceed immediately to Exfil.
Contains a Scav Boss, which is a group of scavs with above-average to high-tier gear that has a chance to spawn in Dorms or Gas Station.

Woods

Woods Map with Exfil
A very large map that is mostly just a large forest, with the occasional bunker, and the Lumber Mill in the center. The Lumber Mill is the primary point of interest, as it contains a couple quest locations and is the primary location to farm Scavs, as Scavs killed on woods is the only known location to find the Arsenal Key, which is the rarest and most valuable key in the game at the moment.
Since the map is so large and open, sniper rifles with scopes usually reign king here. You will see a lot of players with Mosin rifles as they are a cheap way to train the Sniper skill (for a quest later on) and are capable of killing geared players and scavs alike.
Overall, not usually very populated. An early quest from Prapor sends you here to kill a number of Scavs. A good map to learn the game, as although the loot is not fantastic, you can get experience with how the game runs and operates while fighting AI and possibly getting lucky with a key find off a scav.

Shoreline

Shoreline Map, with Loot, Exfil, etc
A very large map, notorious for its FPS hit. Generally speaking, one of the better maps for loot.
The primary point of interest is the Resort, but scavs spawn there, and is primarily occupied by hatchlings (players only with hatchet, ie melee weapon) and geared players. Resort has great loot, but requires keys to access most of it.
A great map to learn though from new players as the outskirts still contains plenty of loot and combat opportunities with AI scavs. You can hit Villa, Scav Island, Weather station, Docks, etc and come out with a backpack full of valuable gear fairly easily.
Location of many quests, including a large quest chain where players are required to kill many, many, scavs on Shoreline.

Interchange

Detailed map
Great, great loot area, but very complex map. Doesn't run very well on old computers. Features a mostly-binary exfil system like Shoreline, but.. kinda worse. Exfil camping is fairly common on this map. Huge map with multiple floors and many many different stores. Communication with teammates is a challenge on this map, but the map is also fantastically detailed.
This map features a lot of loot that depends on the kind of store you're in. It's a great place to farm rare barter materials which are valuable to sell on the Flea market or to use for quests. An early quest (from Ragman) sends you here to kill a large amount of Scavs. I'd recommend getting Ragman to level 2 and accepting his quest before going to Interchange, as getting this quest done can take a while as it is and you want all scav kills to count towards progress.

The Lab ('Labs')

Here's a map.
This is a very complex map, so I highly recommend you read the Wiki article and look at all the maps to get an idea of what it's like.

INSURANCE DOES NOT WORK ON THIS MAP.

If you purchase insurance and lose your items on this map, doesn't matter from whom, you will not get them back.

LABS IS NOT LIKE OTHER MAPS. READ THIS SECTION CAREFULLY.

Labs is by far the most lucrative map to play at the moment. You can easily earn several million roubles per hour by killing Raiders (Juiced-up scavs that are ONLY on Labs), looting their equipment, then running to one of the many, many extracts. Extract camping is hard to pull off and pretty much not viable on Labs.

DISCLAIMER: Labs, like much of Tarkov, is under constant development, so issues may be fixed or created without warning. Always check patch notes!

Raiders

Raiders are the avatars of Death in Tarkov, clad in USEC and BEAR hats, high-end armor, and plenty of firepower to boot. Do not fight Raiders directly. They WILL kill you. Raiders are absolutely broken this patch, and are getting fixed next patch to help fix their exploits of both poor AI and how they're a little.. insane. At the moment, the only way to kill raiders is to camp a hallway, room, or door, and attract the scavs to you, headshotting them the moment they enter before they have a chance to lock onto you.
Raiders can see you through and shoot you through surfaces you cannot.
This means you have to be very careful when engaging them. They are also often equipped with very high-end ammo, meaning that most faceshields (even Killa helmets) can be useless vs a Scav who spawned with 'big boy' ammo, 7N39. They can shoot your head if it's even slightly visible. They can prone instantly, as they have no ping. If they drop without slumping over, get to cover immediately.
Typically, strategies to farm Labs (barring rushing certain rooms for static loot) involve rushing a camp-spot and baiting raiders to your location and taking them out quickly, efficiently, and with no mercy. There are many locations to camp, and since there's so many exfiltrations, it's ultimately up to personal preference.
Raiders often spawn with armor (often Troopers and Gen4s) a rig (Sniper or Ana Alpha rig, usually) and a variety helmets and weapons. Always check the ammo the raiders spawn with. If they spawn with ANYTHING other than PS (and 12 gauge) LOOT IT! You can right click their magazine and hit 'unload ammo' to get the ammo without having to grab the mags, which saves space.
BS, BT, BP, 7N39, etc can be worth several hundred roubles a round on the market. They're extremely valuable.
Additionally, Raiders spawn IFAKS, Morphine, and grenades (F-1 & Flashbangs (Zarya)) with regularity. They can also spawn with random consumables and large clumps of cash in their pockets as well. They can spawn several backpacks, most of them being rather large, if relatively uncommon.
Additionally, Raiders can have American names, breach doors, and mumble as if they are a USEC PMC because some of the Raiders are actually USECs. You will learn with experience what the Scavs will or will not say.

Experience Farming on Labs

Labs is perhaps the best place to farm experience on the current patch.
Killing a Raider with a headshot awards 1100 Experience.
This does not include any looting, inspection (searching bodies), examine, streak, or other experience.
Killing a large sequence of Raiders gives additional bonus experience in the form of Streak rewards, usually 100 bonus exp per additional kill.
Surviving the raid multiplies all of these sources of experience by 1.5x

Changes coming to Labs

Disclaimer: I am not a BSG developer or employee. This is what I have seen on this subreddit and heard elsewhere. Some might be purely rumor, but other points are confirmed by Nikita.
Labs is currently undergoing an overhaul. At the moment, you require consumable Keycards to enter Labs, which may be purchased from Therapist or bartered Mechanic in exchange for 1 Bitcoin starting at Loyalty Level II. They can also be found in drawers and jackets. Scavs can drop permanent keycards that replace most keys used in the previous iteration of labs.
The full extent of the changes coming is not known.
Remember, you can load a map in OFFLINE mode to practice against bots or to learn the map without fear of losing gear.

Tarkov's Health System

Tarkov Wiki Article
Tarkov has a very advanced health system, and while it might seem overwhelming at first, you'll get the hang of it rather quickly. It features a very wide variety of effects and injury, including hydration, energy, blood pressure, blood loss, fractures, contusion, intoxication, exhaustion, tremors and more.
Not all of the Health System is implemented yet. Expect changes!
Your character (PMC, or otherwise) has a combined Health of 435. Each of his limbs have separate health. Taking damage to a limb that reduces it to 0 'blacks' that limb. Blacked limbs are a problem. They greatly impair the activities your PMC performs, and taking damage in a blacked limb amplifies the damage by a multiplier and spreads that damage among your other non-black limbs equally. You cannot heal a blacked limb.
Notes: Bloodloss applies damage to the affected limb and can be spread like other damage to a blacked limb. Treat immediately. Also causes significant dehydration!
Losing a limb applies additional effects. Fractures also apply these effects but not the damage amplification (Except for damage if running on fractured leg.)
Dehydration is what happens when your Hydration level reaches 0. You can view your Hydration level in your gear page, at the bottom left. Becoming dehydrated is extremely bad. You take constant damage. Taking dehydration damage can kill you if you have a black chest or head.
Head/Chest: Bullet damage resulting in losing your head or chest is instant death. Note: Bloodloss resulting in your Head/Chest being black does not result in death, but any damage to them beyond that point will! A back chest will causes you to cough (much like your stomach!)
Painkillers: Prevents coughing that comes from your chest. Doesn't help otherwise.
Stomach: Massively increased rate of dehydration and energy loss. You must find liquids or exit the Raid soon. Additionally, your PMC will cough sputter loudly, attracting attention.
Painkillers: Significantly reduces the frequency and volume of the coughs.
Arms: Makes activities like searching, reloading, etc, take additional time, as well as adding a sway, reducing accuracy. Arms have a .7x damage multiplier.
Painkillers: Reduces sway, removes debuff Pain.
Legs: Blacked legs cause your PMC to stumble and be unable to run. Blacked legs have a 1x damage multiplier.
Painkillers: Allows you to walk at full speed and to run.
WARNING: Running while your legs are blacked or fractured WILL DAMAGE YOU.

Health Items

Tarkov features many health items - 'Aid' items, which can be used to restore your characters health and to fix ailments or injuries he receives as the result of combat or mishaps. The two most important health conditions to consider are bloodloss and fractures, which have both been covered above. Some food items may have ancillary effects, such as losing hydration.
Since in the current patch the only ailments to worry about are bleeding and fractures, it changes which health items are most necessary. We'll go over them below.

Health Restoration

Medical Items on Wiki
AI-2 medkit
The newb's medical kit. You receive several of these when you start Tarkov - they'll already be in your stash. Available from Level I Therapist, they are cheap and effective way of healing early in the game. They will not stop bloodloss. Because of this, you also need to bring bandages or a higher-grade medical kit.
Affectionately called 'little cheeses' by the Tarkov community. Using it takes 2 seconds, and because of how cheap it is, it's often brought in by higher level players to supplement their healing without draining their main kit (which is capable of healing bloodloss or sometimes fractures). Due to its short use time, it's often very useful during combat as you can take cover and quickly recover damage taken to a vital limb.
Bandages
The newb's bloodloss solution. Available from Therapist at Level I. A better version, the Army Bandage is available at Level II, after a quest. Mostly obsolete after unlocking the Car Medical kit. Activating takes 4 seconds, and removes bloodloss to one limb.
Splint
The newb's solution to fractures. Cheap, takes five seconds to use, and takes up 1 slot. Not generally recommended to take because fractures effects can be greatly mitigated with the use of Painkillers. Available from Therapist at Level I, no quest needed.
Car Medical Kit
The newb's first real medical solution. Available LL1 as a barter (2 Duct Tape) and available for Roubles after completing Therapist's second quest. Has a larger health pool than AI-2's (220, vs AI-2's 100), and removes bloodloss. Takes up a 1x2 slot, so requires to be placed in a tactical rig in order to be used effectively. Cheap and fairly efficient, takes a standard 4 seconds to use. Rendered effectively obsolete when the Salewa is unlocked.
Salewa
Good medkit for use in mid and end-game. Contains 400 total health and can remove bloodloss. Relatively expensive at 13k roubles per kit, though. Same size as the Car medical kit, so requires a tactical rig to use effectively. Because Tarkov does not currently have effects like Toxication in the game at the moment, this kit is favored by most players who go into a raid with at least a moderate level of gear.
Unlocked at Therapist Level II after completing a level 10 Prapor quest, Postman Pat Part II.
IFAK
Fantastic medical kit, and is the one preferred by most players. Features 300 health and the ability to remove bloodloss and a host of other negative effects that are not yet implemented into the game. It does not, however, remove fractures. Taking up only a single slot, it is favored by players in all stages of gear, and it is recommend to carry one in your Secure Container in case of emergencies. Is available at Therapist Level II for a barter (Sugar + Sodium), and may be purchased for Roubles at Level III after completing Healthcare Privacy, Part I.
It is a fairly expensive kit, but due to its durability, its small size, and ability to remove bloodloss, it is a very common medical item used by players of all levels.
Grizzly
The 'big daddy' medical kit, boasting an impressive total health resource of 1800. It is also a very large kit, taking up 4 slots (2x2) - in order to be able to use this quickly, it would require specialized tactical rigs that feature a 2x2 slot. It removes all negative effects (some costing HP resource), including fractures.
Used by highly-geared players who intend on staying in raids for an extended period of time, or by players with additional Secure Container space available in case of emergencies. It is available for barter at Therapist Level II, and purchase at Therapist Level 4.

Pain Management

Using any of these items results in your character being 'On Painkillers' which allows you to sprint on fractured and blacked legs, as well as reducing effects of fractures and blacked limbs, and removing the debuff Pain. Essentially, the only difference between most of these items are the speed of use, price, availability, and duration of the effect.
Analgin Painkillers
The holy grail of pain medication. With the recent changes, "Painkillers" now have 4 total uses, not 1. The total duration is now greater than Morphine and less risk of waste. Takes a short time to use, and is available from Therapist Level 1 for both barter and Roubles.
Morphine
Quick application of painkillers. Favored by some highly geared players as it has greater usability in combat then it's typical counterpart, Painkillers. Has a longer duration, but only one use. Is required for a fairly early Therapist (and a late Peacekeeper) Quest, so it is recommend to hoard 10 of them, then sell the rest unless you intend on using them. They are worth a good amount to Therapist and take up little space so they are a valuable loot item. Available from Therapist for Roubles at Level 4, after completing Healthcare Privacy, Part 3.
Augmentin
Basically a cheaper Morphine. One use, 260s. Not recommended over Painkillers due to its cost. No current barter for this item, so usually it's just a fairly expensive, small loot item to sell to Therapist when found.
Ibuprofen
Powerful painkiller. Lasts 600 seconds and has 12 uses. However, it is not recommended to use it as a Painkiller. It is very valuable because it cannot be purchased from Dealers, it must be found, and it is a barter component to late-game containers, the Keytool and THICC Items Case.
Vaseline
Powerful medical item. Cannot be purchased from dealers. Has 10 uses. Removes Pain.
Golden Star Balm
Fairly useful medical item. It can remove Pain and Contusion (not a big deal of a debuff, goes away on its own shortly) and provides a small bonus to hydration and energy. However, because Hydration is usually easy to restore (Liquids are easy to find as 'common' or 'trash' tier loot), and Energy at the moment can't run to 0 within current Raid timers, it is Recommended to just to sell to Therapist as a Loot item.

Medical Injectors are not covered here. Essentially, they are powerful but niche items with strong side effects. Most recommended use is to store them in your Secure Container and sell them either on the Flea Market or to Therapist for roubles.

To be able to Hotkey a medicine item, they must be in a tactical rig or your pockets.

Tarkov's Quest, Progression, and Experience Systems

Tarkov features a very immersive progression system where your main character (PMC) is going into raids to acquire loot - goods that can be sold for a profit to other players, to Dealers (NPC Merchants), or used to fulfill quest requirements in order to complete them and receive your rewards. Additionally, your main character will increase their prowess in a number of skills, which increases everything from how much they can run, increases the ease of which recoil can be controlled, and even how far you can throw grenades. These are referred to 'soft skills.'
Additionally, your PMC is assigned a Level. You can increase your Level by earning Experience - which is rewarded by performing numerous tasks throughout the Raid, completing quests, examining new items, killing other players and Scavs, etc. Successfully extracting from a raid will increase the experience you earn from the raid via a multiplier.
Increasing your PMC's level will allow you to complete additional quests, which increases your Reputation with certain Dealers (and may reduce your Reputation with others) allowing you access to better equipment to purchase. Additionally, completing quests will often reward you with large sums of currency and sometimes equipment, and certain quests unlock items for purchase from that dealer.
A Dealer's arsenal of available weapons, ammo, mods, medication, containers, and etc to purchase by you is determined by their Loyalty Level - or LL, for short. Certain Dealers specialize in different kinds of equipment, and they will pay different rates or straight up not buy particular items. In a future release, eventually Dealers will offer discounts to the player based off their Loyalty Level.
Article on Dealers

Increasing Loyalty Level

Increasing your Trader's loyalty level is extremely important to your progression and overall success in Tarkov. Being able to purchase better Ammo and Equipment is essential to being able to fight other players and secure their loot for your own. Owning Prepare for Escape and Edge of Darkness (EoD) editions of Escape from Tarkov will increase your starting Reputations with Traders. It is unclear if this change will stay after the game's full release.
Typically though, you need three things to increase your Trader's level.

Reputation

This is accomplished via quests. Completing a quest will reward you with an increase in the quest givers' reputation, sometimes an increase in another trader's reputation, and sometimes will reduce the Reputation of another trader. Not all Traders need a certain level of Reputation to increase their loyalty Level to II. Peacekeeper and Ragman, for example, just need you to spend a certain amount of money with them.

Character Level and Experience Gain

The primary gate behind your trade level (and thus your overall economy and gear leverage) is your Character Level. You increase this by gaining Experience. The easiest way to gain experience is to Loot high value areas, fight players, and kill scavs while completing quests. Generally speaking, your level will advance as you play the game at a moderate pace. One way to farm experience though is to avoid looting all-together and just focus on killing a large number of scavs from a safe distance, after learning where they tend to spawn on any given map. This patch however, labs is fantastic for experience gain. (See above.) Another strategy (albeit one that takes longer) is to loot everything, then drop what you don't want. You gain experience for finding items and picking them up, so picking them up to drop them is technically the best way to gain the most exp per kill.
You can receive additional bonuses to Experience earned. Successfully extracting will increase your experience by a multiplier, typically 1.5x the experience gained during the raid, escaping also rewards a 300 exp Escape bonus which is added to your total before the multiplier is applied.
You can also receive experience bonuses for Exploration, so visiting different parts of the map will reward you with sums of experience, usually 100 to 300 or so. Killing multiple enemies in a row will reward you with Streaks, whose rewards increase as you get more kills. Getting a kill with a Headshot also significantly improves experience gain from kills. You also receive a (very small) bonus when you survive consecutive raids.
Note: Completing a Raid too early (via extraction) will cause you to receive a Run-Through status, which reduces experience earned in that raid by 50%.
Most Quests require you to be a certain level to unlock, and upon completion rewards you with a lot of exp and usually the ability to purchase specific equipment.

Money Spent

This is pretty self-explanatory. As mentioned above, Peacekeeper and Ragman can be increased easily just be selling and buying from them. If you need to artifically inflate the amount spent, a good idea is to purchase a large amount of cheap items from them and sell them back to the Trader. You still take a significant loss, usually around 50-60% per purchase, but since the money spent counts both items sold to the vendor and purchases, you get about 140-150% credit per item at about half the cost.

Selling Efficiency

Not all dealers pay the same for certain items. It is important to note that a lot of this is my personal experience, and prices can fluctuate as the Developers may change them for any reason. Use your own common sense and check various dealers before selling particularly lucrative items.

Prapor

Sells AKs, Magazines, many different Ammo types, Grenades, and weapon modifications. I don't tend to sell to him very often, as he doesn't pay the highest for any items that I have personally seen and because you tend to buy most Ammo and mags from him it's not a particular issue to level him up with money spent.

Therapist

Sells medical supplies, food and drink, and storage cases, which are items that effectively increase the size of your stash because they have more space inside than they take up. Most storage items are restricted to certain item types. Pays most for items like Keys, Statues, Rolers, Bitcoin, etc. Many of these items should be sold on the market instead of to her, but often times it's not worth the hassle.

Fence

Pays least for items, sells items for more than other Dealers. Items other players have sold will appear here. Only sell items to Fence that other dealers will not take! Basically a placeholder for an expanded Market.

Skier

Sells various weapons, mods, ammo, Euros, and containers. Pays most for items like Armor, backpacks, headgear, facemasks, flashlights, sights, etc. It is important to note, that Skier will not buy Weapons or most Mods. That means for things like flashlights, you have to take the flashlights/sights off the mount or rail in order for him to buy them, but he pays the best.

Peacekeeper

Deals entirely in Western equipment, UN armor, helmets, etc. Will buy most items, but will pay USD for them. Deals entirely in USD. One good way to get his money spent requirement is just to buy USD, which is used for a later quest from Skier, which unlocks his quest chain. He has a lot of good deals, experiment for yourself. At the moment his MP5 for 10 'bars' knives (scav knives) is an exceptionally good deal and easy to accomplish for new players.

Mechanic

Sells mostly completed weapons with various modifications and unique names, and mods. Sells magazines and some ammo. Offers containers as you progress. His quests are easy to complete, but often are money dumps in exchange for large sums of EXP more than anything else. Pays the most for modifications (except for sights and suppressors) and stripped guns.

Ragman

Sells backpacks, armor, tactical vests, and helmets mostly. Offers aesthetic clothing. Can obtain LL2 by just purchasing from him, does not require reputation. In fact, his first quest tasks you with that very objective. As far as I can see, he does not pay the most for any items in particular. But he is a very useful merchant once you have him at level 2. He will sell Scav Backpacks, which are an extremely efficient backpack to use as it's fairly large but very cheap.

Rule of Thumb for selling items at most value

Weapons: Strip the weapon! Take apart ALL pieces of it (including gas tubes, separating flashlights from ring mounts, etc), sell what you can to Skier. For the rest, sell to Mechanic.
Keys, Food, Medical Items, Statues, Bitcoin, Rolers, etc: Therapist or the Market. For items like this, ALWAYS check the Market first! A lot of these kinds of items are in valuable trades or are required for quests; this means that other players are often willing to pay more for them, above trader prices.

Continued below in a comment, due to character limit.

When trading 5-minute binary option contracts, Bollinger Bands can alert you to general volatility and opportunity when you are deciding which markets to trade. Moving averages. Moving averages can work across multiple time frames, so you can choose charts that give shorter-term signals. This is perfect for trading 5-minute binary options. There are a lot of ways to trade the 5 minute binary options expiry. This time frame is one of the most versatile in terms of the types of strategies you can use because it is inherently volatile yet at the same time can sustain a trend long enough to be useful to us binary options traders. The risks involved in trading binary options are high and may not be suitable for all investors. Binary Options Edge doesn't retain responsibility for any trading losses you might face as a result of using the data hosted on this site. The data and quotes contained in this website are not provided by exchanges but rather by market makers. Actually, there are many different strategies to trade binary options on the 5-minute time limit expiration. For example, the main focus of the profitability of the «High Range» system are trading signals that, in this strategy, are obtained from technical indicators. They make this process profitable. Binary Today 5 Provides Guaranteed 81% ITM Trading Signals Binary Today 5 is a binary options trading software for every binary trader. The system is easy to use, install and provides consistent gains with little to no risk. Download the software,

5 MIN ATM SMALL ACCOUNT

1. Practice in demo until you are consistently profitable. B: Practice by doing nothing other than watching your trades win or lose and keeping notes in live... The road to success through trading IQ option Best Bot Reviews Iq Option 2020 ,We make videos using this softwhere bot which aims to make it easier for you t... Subscribe for more videos like this one! Today we go over one of my most successful strategies- and yet it is perhaps the simplest. Patience is crucial howev... I made this short video step by step with the 5 minute strategy and explaining how to follow its signals. FOREX & BINARY SIGNALS http://nextwavetrading.com/S... Available Now My 5 minute ATM Binary Options Trading Strategy!!! You can earn $50-150 an hr or SCALP. Questions, Comments,Concerns email me @ [email protected]

5 min binary options trading strategy - FX141.COM

vfxAlert - Signals for binary options

[N] Tensorflow 2.3.0 Released!

Release 2.3.0

Major Features and Improvements

Breaking Changes

Known Caveats

Bug Fixes and Other Changes

TF Core:

tf.data:

tf.distribute:

tf.keras:

tf.lite:

Packaging Support

Profiler

TPU Enhancements

Tracing and Debugging

XLA Support

A guide on hitting Legend in Comp

A guide on hitting Legend rank in Comp

FlowCards: A Declarative Framework for Development of Ergo dApps

FlowCards: A Declarative Framework for Development of Ergo dApps

Paradigm

Scalability

Shared state

Expressive Power

From Imperative to Declarative

From Changing State to Checking Context

FlowCard Diagrams

1. Name and Parameters

2. Contract Wallet

3. Contract

4. Box name

5. Boxes in the wallet

6. Amount of ERGs in the box

7. Amount of T token

8. Tx Inputs

9. Transaction

10. Tx Outputs

Example: Decentralized Exchange (DEX)

Buy Order Transaction

Cancel Buy, Cancel Sell Transactions

Sell Order Transaction

Swap Transaction

From Diagrams To ErgoScript Contracts

Conclusions

Escape from Tarkov - New Player Guide!

Introduction

NEW VERSION: https://www.reddit.com/EscapefromTarkov/comments/ffyynf/escape_from_tarkov_new_player_guide_20_75_pages/

Changelog

Table of Contents

Tarkov Overview - What is Escape from Tarkov?

Tarkov Resources - Useful links

Quick-Reference Ammo Chart

Tarkov Wiki

Map Keys and You

Tarkov's Weapon Compatibility Guide

HUGE Reference Bible by Veritas

Tarkov's Maps

INSURANCE DOES NOT WORK ON THIS MAP.

LABS IS NOT LIKE OTHER MAPS. READ THIS SECTION CAREFULLY.

Tarkov's Health System

Health Restoration

Pain Management

Tarkov's Quest, Progression, and Experience Systems

Increasing Loyalty Level

Selling Efficiency

Continued below in a comment, due to character limit.

5 MIN ATM SMALL ACCOUNT