Skip to content

Latest commit

 

History

History
2045 lines (1548 loc) · 70.5 KB

usage_command_line.md

File metadata and controls

2045 lines (1548 loc) · 70.5 KB

Stim command line reference

Index

Commands

stim analyze_errors

NAME
    stim analyze_errors

SYNOPSIS
    stim analyze_errors \
        [--allow_gauge_detectors] \
        [--approximate_disjoint_errors [probability]] \
        [--block_decompose_from_introducing_remnant_edges] \
        [--decompose_errors] \
        [--fold_loops] \
        [--ignore_decomposition_failures] \
        [--in filepath] \
        [--out filepath]

DESCRIPTION
    Converts a circuit into a detector error model.

OPTIONS
    --allow_gauge_detectors
        Allows non-deterministic detectors to appear in the circuit.

        Normally (without `--allow_gauge_detectors`), when a detector's
        detecting region anti-commutes with a reset or measurement, stim
        will raise an exception when analyzing the circuit. When
        `--allow_gauge_detectors` is set, stim will instead append an error
        mechanism into the detector error model that has a probability of
        50% and flips all the detectors that anticommute with the operation.

        This is potentially useful in situations where the layout of
        detectors is supposed to stay fixed despite variations in the
        circuit structure. Decoders can interpret the existence of the 50%
        error as a weight 0 edge saying that the detectors should be fused
        together.

        For example, in the following stim circuit, the two detectors each
        anticommute with the reset operation:

            R 0
            H 0
            CNOT 0 1
            M 0 1
            DETECTOR rec[-1]
            DETECTOR rec[-2]

        Without `--allow_gauge_detectors`, stim will raise an exception when
        analyzing this circuit. With `--allow_gauge_detectors`, stim will
        add `error(0.5) D1 D2` to the output detector error model.

        BEWARE that gauge detectors are very tricky to work with, and not
        necessarily supported by all tools (even within stim itself). For
        example, when converting from measurements to detection events,
        there isn't a single choice for whether or not each individual gauge
        detector produced a detection event. This means that it is valid
        behavior for one conversion from measurements to detection events
        to give different results from another, as long as the gauge
        detectors that anticommute with the same operations flip together in
        a consistent fashion that respects the structure of the circuit.


    --approximate_disjoint_errors
        Allows disjoint errors to be approximated during the conversion.

        Detector error models require that all error mechanisms be
        specified as independent mechanisms. But some of the circuit error
        mechanisms that Stim allows can express errors that don't correspond
        to independent mechanisms. For example, the custom error channel
        `PAULI_CHANNEL_1(0.1, 0.2, 0.0)` can't be expressed exactly as a set
        of independent error mechanisms. But it can be approximated as an
        `X_ERROR(0.1)` followed by a `Y_ERROR(0.2)`.

        This flag can be set to any probability between 0 (the default when
        not specified) and 1 (the default when specified without a value).
        When set to a value strictly between 0 and 1, this determines the
        maximum disjoint probability that is allowed to be approximated as
        an independent probability.

        Without `--approximate_disjoint_errors`, attempting to convert a
        circuit containing `PAULI_CHANNEL_1(0.1, 0.2, 0.0)` will fail with
        an error stating an approximation is needed. With
        `--approximate_disjoint_errors`, the conversion will succeed by
        approximating the error into an `X_ERROR(0.1)` followed by an
        independent `Y_ERROR(0.2)`.

        Note that, although `DEPOLARIZE1` and `DEPOLARIZE2` are often
        defined in terms of disjoint errors, they can be exactly converted
        into a set of independent errors (unless the probability of the
        depolarizing error occurring exceeds maximum mixing, which is 75%
        for `DEPOLARIZE1` and 93.75% for `DEPOLARIZE2`). So the
        `--approximate_disjoint_errors` flag isn't needed for depolarizing
        errors that appear in practice.

        The error mechanisms that require approximations are:
        - PAULI_CHANNEL_1
        - PAULI_CHANNEL_2
        - ELSE_CORRELATED_ERROR

        In principle some custom Pauli channels can be converted exactly,
        but Stim does not currently contain logic that attempts to do this.


    --block_decompose_from_introducing_remnant_edges
        Prevents A*B from being decomposed unless A,B BOTH appear elsewhere.

        Irrelevant unless `--decompose_errors` is specified.

        When `--decompose_errors` is specified, any circuit error that
        causes more than two detection events must be decomposed into a
        set of errors with at most two detection events. The main constraint
        on this process is that it must not use errors that couldn't
        otherwise occur, since introducing such errors could violate
        important properties that are used for decoding. For example, in the
        normal surface code, it is very important that the decoding graphs
        for X errors and Z errors are disjoint in the bulk, and decomposing
        an error into a set of errors that violated this property would be
        disastrous.

        However, a corner case in this logic occurs if an error E1 that
        produces detection events A*B needs to be decomposed when an error
        E2 that produces detection events A appears elsewhere but no error
        producing detection events B appears elsewhere. The detection events
        B can be produced by both E1 and E2 occurring, but this a
        combination of two errors and so treating it as one error can cause
        problems. For example, it can result in the code distance appearing
        to be smaller than it actually is. Introducing B is referred to as
        introducing a "remnant edge" because B *only* appears in the
        detector error model as a remnant of removing A from A*B.

        By default, Stim does allow remnant edges to be introduced. Stim
        will only do this if it is absolutely necessary, but it *will* do
        it. And there are in fact QEC circuits where the decomposition
        requires these edges to succeed. But sometimes the presence of a
        remnant edge is a hint that the DETECTOR declarations in the circuit
        are subtly wrong. To cause the decomposition process to fail in
        this case, the `--block_decompose_from_introducing_remnant_edges`
        can be specified.


    --decompose_errors
        Decomposes errors with many detection events into "graphlike" parts.

        When `--decompose_errors` is specified, Stim will suggest how errors
        that cause more than 2 detection events (non-graphlike errors) can
        be decomposed into errors with at most 2 detection events (graphlike
        errors). For example, an error like `error(0.1) D0 D1 D2 D3` may be
        instead output as `error(0.1) D0 D1 ^ D2 D3` or as
        `error(0.1) D0 D3 ^ D1 ^ D2`.

        The purpose of this feature is to make matching apply to more cases.
        A common decoding strategy is "matching", where detection events are
        paired up in order to determine which errors occurred. Matching only
        works when the Tanner graph of the problem is a graph, not a
        hypergraph. In other words, it requires all errors to produce at
        most two detection events. This is a problem, because in practice
        there are essentially always circuit error mechanisms that produce
        more than two detection events. For example, in a CSS surface code,
        Y type errors on the data qubits will produce four detection events.
        For matching to work in these cases, non-graphlike errors (errors
        with more than two detection events) need to be approximated as a
        combination of graphlike errors.

        When Stim is decomposing errors, the main guarantee that it provides
        is that it will not introduce error mechanisms with symptoms that
        are otherwise impossible or that would require a combination of
        non-local errors to actually achieve. Informally, Stim guarantees it
        will preserve the "structure" of the detector error model when
        suggesting decompositions.

        It's also worth noting that the suggested decompositions are
        information preserving: the undecomposed model can always be
        recovered by simply filtering out all `^` characters splitting the
        errors into suggested components.

        Stim uses two strategies for decomposing errors: intra-channel and
        inter-channel.

        The *intra-channel* strategy is always applied first, and works by
        looking at the various detector/observable sets produced by each
        case of a single noise channel. If some cases are products of other
        cases, that product is *always* decomposed. For example, suppose
        that a single qubit depolarizing channel has a `Y5` case that
        produces four detection events `D0 D1 D2 D3`, an `X5` case that
        produces two detection events `D0 D1`, and a `Z5` case that produces
        two detection events `D2 D3`. Because `D0 D1 D2 D3` is the
        combination of `D0 D1` and `D2 D3`, the `Y5` case will be decomposed
        into `D0 D1 ^ D2 D3`. An important corner case here is the corner of
        the CSS surface code, where a Y error has two symptoms which is
        graphlike but because the intra-channel strategy is aggressive the
        Y error will still be decomposed into X and Z pieces. This can keep
        the X and Z decoding graphs disjoint.

        The *inter-channel* strategy is used when an error component is
        still not graphlike after the intra-channel strategy was applied.
        This strategy searches over all other error mechanisms looking for a
        combination that explains the error. If
        `--block_decompose_from_introducing_remnant_edges` is specified then
        this must be an exact match, otherwise the match can omit up to two
        of the symptoms in the error (resulting in the producing of a
        "remnant edge").

        Note that the code implementing these strategies does not special
        case any Pauli basis. For example, it does not prefer to decompose
        Y into X*Z as opposed to X into Y*Z. It also does not prefer to
        decompose YY into IY*YI as opposed to IY into YY*YI. The code
        operates purely in terms of the sets of symptoms produced by the
        various cases, with little regard for how those sets were produced.

        If these strategies fail to decompose error into graphlike pieces,
        Stim will throw an error saying it failed to find a satisfying
        decomposition.


    --fold_loops
        Allows the output to contain `repeat` blocks.

        This flag substantially improves performance on circuits with
        `REPEAT` blocks with large repetition counts. The analysis will take
        less time and the output will be more compact. This option is only
        OFF by default to maintain strict backwards compatibility of the
        output.

        When a circuit contains a `REPEAT` block, the structure of the
        detectors often settles into a form that is identical from iteration
        to iteration. Specifying the `--fold_loops` option tells Stim to
        watch for periodicity in the structure of detectors by using a
        "tortoise and hare" algorithm (see
        https://en.wikipedia.org/wiki/Cycle_detection ).
        This improves the asymptotic complexity of analyzing the loop from
        O(total_repetitions) to O(cycle_period).

        Note that, although logical observables can "cross" from the end of
        the loop to the start of the loop without preventing loop folding,
        detectors CANNOT. If there is any detector introduced after the
        loop, whose sensitivity region extends to before the loop, loop
        folding will fail and the code will fall back to flattening the
        loop. This is disastrous for loops with huge repetition counts (e.g.
        in the billions) because in that case loop folding is the difference
        between the error analysis finishing in seconds instead of in days.


    --ignore_decomposition_failures
        Allows non-graphlike errors into the output when decomposing errors.

        Irrelevant unless `--decompose_errors` is specified.

        Without `--ignore_decomposition_failures`, circuit errors that fail
        to decompose into graphlike detector error model errors will cause
        an error and abort the conversion process.

        When `--ignore_decomposition_failures` is specified, circuit errors
        that fail to decompose into graphlike detector error model errors
        produce non-graphlike detector error models. Whatever processes
        the detector error model is then responsible for dealing with the
        undecomposed errors (e.g. a tool may choose to simply ignore them).


    --in
        Chooses the stim circuit file to read the circuit to convert from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --out
        Chooses where to write the output detector error model.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is a stim detector error model. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_dem_detector_error_model.md


EXAMPLES
    Example #1
        >>> cat example_circuit.stim
        R 0 1
        X_ERROR(0.125) 0 1
        CNOT 0 1
        M 0 1
        DETECTOR rec[-1]
        DETECTOR rec[-2]

        >>> stim analyze_errors --in example_circuit.stim
        error(0.125) D0
        error(0.125) D0 D1


    Example #2
        >>> stim gen \
                --code repetition_code \
                --task memory \
                --distance 3 \
                --rounds 1000 \
                --after_reset_flip_probability 0.125 \
                > rep_code.stim
        >>> stim analyze_errors --fold_loops --in rep_code.stim
        error(0.125) D0
        error(0.125) D0 D1
        error(0.125) D0 D2
        error(0.125) D1 D3
        error(0.125) D1 L0
        error(0.125) D2 D4
        error(0.125) D3 D5
        detector(1, 0) D0
        detector(3, 0) D1
        repeat 998 {
            error(0.125) D4 D6
            error(0.125) D5 D7
            shift_detectors(0, 1) 0
            detector(1, 0) D2
            detector(3, 0) D3
            shift_detectors 2
        }
        shift_detectors(0, 1) 0
        detector(1, 0) D2
        detector(3, 0) D3
        detector(1, 1) D4
        detector(3, 1) D5

stim convert

NAME
    stim convert

SYNOPSIS
    stim convert \
        --bits_per_shot int \
        [--circuit filepath] \
        [--in filepath] \
        [--in_format 01|b8|r8|ptb64|hits|dets] \
        --num_detectors int \
        --num_measurements int \
        --num_observables int \
        [--obs_out filepath] \
        [--obs_out_format 01|b8|r8|ptb64|hits|dets] \
        [--out filepath] \
        [--out_format 01|b8|r8|ptb64|hits|dets] \
        --types M|D|L

DESCRIPTION
    Convert data between result formats.

    See the various formats here:
    https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md

    To read and write data, the size of the records must be known.
    If writing to a dets file, then the number of measurements, detectors
    and observables per record must also be known.

    Both of these pieces of information can either be given directly, or
    inferred from various data sources, such as circuit or dem files.


OPTIONS
    --bits_per_shot
        Specifies the number of bits per shot in the input/output files.

        This argument is required if the circuit, dem or num_* flags
        are not given, and not supported when writing to a dets file.

        In this case we just treat the bits aas arbitrary data. It is up
        to the user to interpert it correctly.


    --circuit
        Specifies where the circuit that generated the data is.

        This argument is optional, but can be used to infer the number of
        measurements, detectors and observables to use per record.

        The circuit file should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --in
        Chooses the file to read data from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input's format is specified by `--in_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --in_format
        Specifies the data format to use when reading data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --num_detectors
        Specifies the number of detectors in the input/output files.

        This argument is required if writing to a dets file and the circuit
        or dem is not given.


    --num_measurements
        Specifies the number of measurements in the input/output files.

        This argument is required if writing to a dets file and the circuit
        is not given.


    --num_observables
        Specifies the number of observables in the input/output files.

        This argument is required if writing to a dets file and the circuit
        or dem is not given.


    --obs_out
        Specifies the file to write observable flip data to.

        When producing detection event data, the goal is typically to
        predict whether or not the logical observables were flipped by using
        the detection events. This argument specifies where to write that
        observable flip data.

        If this argument isn't specified, the observable flip data isn't
        written to a file.

        The output is in a format specified by `--obs_out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --obs_out_format
        Specifies the data format to use when writing observable flip data.

        Irrelevant unless `--obs_out` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out
        Chooses where to write the data to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output's format is specified by `--out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out_format
        Specifies the data format to use when writing output data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --types
        Specifies the types of events in the files.

        This argument is required if a circuit is given as the circuit can
        give the number of each type of event, but not which events are
        contained within an input file.

        Note that in most cases, a file will have either measurements only,
        detections only, or detections and observables.

        The type values (M, D, L) correspond to the value prefix letters
        in dets files. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md#dets


EXAMPLES
    Example #1
        >>> cat example.01
        10000
        11001
        00000
        01001

        >>> stim convert \
            --in example.01 \
            --in_format 01 \
            --out_format dets
            --num_measurements 5
        shot M0
        shot M0 M1 M4
        shot
        shot M1 M4


    Example #2
        >>> cat example.dem
        detector D0
        detector D1
        logical_observable L2

        >>> cat example.dets
        shot D0
        shot D0 D1 L2
        shot
        shot D1 L2

        >>> stim convert \
            --in example.dets \
            --in_format dets \
            --out_format 01
            --dem example.dem
        10000
        11001
        00000
        01001


    Example #3
        >>> cat example_circuit.stim
        X 0
        M 0 1
        DETECTOR rec[-2]
        DETECTOR rec[-1]
        OBSERVABLE_INCLUDE(2) rec[-1]

        >>> cat example_measure_data.01
        00
        01
        10
        11

        >>> stim convert \
            --in example_measure_data.01 \
            --in_format 01 \
            --out_format dets
            --circuit example_circuit.stim \
            --types M
        shot
        shot M1
        shot M0
        shot M0 M1


    Example #4
        >>> cat example.01
        0010
        0111
        1000
        1110

        >>> stim convert \
            --in example.01 \
            --in_format 01 \
            --out_format hits
            --bits_per_shot 4
        2
        1,2,3
        0
        0,1,2

stim detect

NAME
    stim detect

SYNOPSIS
    stim detect \
        [--append_observables] \
        [--in filepath] \
        [--obs_out filepath] \
        [--obs_out_format 01|b8|r8|ptb64|hits|dets] \
        [--out filepath] \
        [--out_format 01|b8|r8|ptb64|hits|dets] \
        [--seed int] \
        [--shots int]

DESCRIPTION
    Sample detection events and observable flips from a circuit.

OPTIONS
    --append_observables
        Appends observable flips to the end of samples as extra detectors.

        PREFER --obs_out OVER THIS FLAG. Mixing the observable flip data
        into detection event data tends to require simply separating them
        again immediately, creating unnecessary work. For example, when
        testing a decoder, you do not want to give the observable flips to
        the decoder because that is the information the decoder is supposed
        to be predicting from the detection events.

        This flag causes observable flip data to be appended to each sample,
        as if the observables were extra detectors at the end of the
        circuit. For example, if there are 100 detectors and 10 observables
        in the circuit, then the output will contain 110 detectors and the
        last 10 are the observables.

        Note that, when using `--out_format dets`, this option is implicitly
        activated but observables are not appended as if they were
        detectors (because `dets` has type hinting information). For
        example, in the example from the last paragraph, the observables
        would be named `L0` through `L9` instead of `D100` through `D109`.


    --in
        Chooses the stim circuit file to read the circuit to sample from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --obs_out
        Specifies the file to write observable flip data to.

        When sampling detection event data, the goal is typically to predict
        whether or not the logical observables were flipped by using the
        detection events. This argument specifies where to write that
        observable flip data.

        If this argument isn't specified, the observable flip data isn't
        written to a file.

        The output is in a format specified by `--obs_out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --obs_out_format
        Specifies the data format to use when writing observable flip data.

        Irrelevant unless `--obs_out` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out
        Chooses where to write the sampled data to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is in a format specified by `--out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out_format
        Specifies the data format to use when writing output data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --seed
        Makes simulation results PARTIALLY deterministic.

        The seed integer must be a non-negative 64 bit signed integer.

        When `--seed` isn't specified, the random number generator is seeded
        using fresh entropy requested from the operating system.

        When `--seed #` is set, the exact same simulation results will be
        produced every time ASSUMING:

        - the exact same other flags are specified
        - the exact same version of Stim is being used
        - the exact same machine architecture is being used (for example,
            you're not switching from a machine that has AVX2 instructions
            to one that doesn't).

        CAUTION: simulation results *WILL NOT* be consistent between
        versions of Stim. This restriction is present to make it possible to
        have future optimizations to the random sampling, and is enforced by
        introducing intentional differences in the seeding strategy from
        version to version.

        CAUTION: simulation results *MAY NOT* be consistent across machines.
        For example, using the same seed on a machine that supports AVX
        instructions and one that only supports SSE instructions may produce
        different simulation results.

        CAUTION: simulation results *MAY NOT* be consistent if you vary
        other flags and modes. For example, `--skip_reference_sample` may
        result in fewer calls the to the random number generator before
        reported sampling begins. More generally, using the same seed for
        `stim sample` and `stim detect` will not result in detection events
        corresponding to the measurement results.


    --shots
        Specifies the number of samples to take from the circuit.

        Defaults to 1.
        Must be an integer between 0 and a quintillion (10^18).


EXAMPLES
    Example #1
        >>> cat example.stim
        H 0
        CNOT 0 1
        X_ERROR(0.1) 0 1
        M 0 1
        DETECTOR rec[-1] rec[-2]

        >>> stim detect --shots 5 --in example.stim
        0
        1
        0
        0
        0


    Example #2
        >>> cat example.stim
        # Single-shot X-basis rep code circuit.
        RX 0 1 2 3 4 5 6
        MPP X0*X1 X1*X2 X2*X3 X3*X4 X4*X5 X5*X6
        Z_ERROR(0.1) 0 1 2 3 4 5 6
        MPP X0 X1 X2 X3 X4 X5 X6
        DETECTOR rec[-1] rec[-2] rec[-8]   # X6 X5 now = X5*X6 before
        DETECTOR rec[-2] rec[-3] rec[-9]   # X5 X4 now = X4*X5 before
        DETECTOR rec[-3] rec[-4] rec[-10]  # X4 X3 now = X3*X4 before
        DETECTOR rec[-4] rec[-5] rec[-11]  # X3 X2 now = X2*X3 before
        DETECTOR rec[-5] rec[-6] rec[-12]  # X2 X1 now = X1*X2 before
        DETECTOR rec[-6] rec[-7] rec[-13]  # X1 X0 now = X0*X1 before
        OBSERVABLE_INCLUDE(0) rec[-1]

        >>> stim detect \
            --in example.stim \
            --out_format dets \
            --shots 10
        shot
        shot
        shot L0 D0 D5
        shot D1 D2
        shot
        shot L0 D0
        shot D5
        shot
        shot D3 D4
        shot D0 D1

stim diagram

NAME
    stim diagram

SYNOPSIS
    stim diagram \
        [--filter_coords (float.seperatedby(',') | L# | D#).seperatedby(':')] \
        [--in filepath] \
        [--out filepath] \
        [--remove_noise] \
        [--tick int | int:int] \
        --type name

DESCRIPTION
    Produces various kinds of diagrams.


OPTIONS
    --filter_coords
        Specifies coordinate filters that determine what appears in the diagram.

        A coordinate is a double precision floating point number.
        A point is a tuple of coordinates.
        The coordinates of a point are separate by commas (',').
        A filter is a set of points.
        Points are separated by colons (':').

        Filters can also be set to specific detector or observable indices,
        like D0 or L0.

        Example:
            --filter-coords 2,3:4,5,6
                In a detector slice diagram this means that only detectors whose
                first two coordinates are (2,3), or whose first three coordinate
                are (4,5,6), should be included in the diagram.
            --filter-coords L0
                In a detector slice diagram this means that logical observable 0
                should be included. Logical observables are only included if
                explicitly filtered in.


    --in
        Where to read the object to diagram from.

        By default, the object is read from stdin. When `--in $FILEPATH` is
        specified, the object is instead read from the file at $FILEPATH.

        The expected type of object depends on the type of diagram.


    --out
        Chooses where to write the diagram to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The type of output produced depends on the type of diagram.


    --remove_noise
        Removes noise from the input before turning it into a diagram.

        For example, if the input is a noisy circuit and you aren't
        interested in the details of the noise but rather in the structure
        of the circuit, you can specify this flag in order to filter out
        the noise.


    --tick
        Specifies that the diagram should apply to a specific TICK or range
        of TICKS from the input circuit.

        To specify a single tick, pass an integer like `--tick=5`.
        To specify a range, pass two integers separated by a colon like
        `--tick=start:end`. Note that the range is half open.

        In detector and time slice diagrams, `--tick` identifies which ticks
        to include in the diagram. Note that `--tick=0` is the very
        beginning of the circuit and `--tick=1` is the instant of the first
        TICK instruction.


    --type
        The type of diagram to make.

        The available diagram types are:

        "timeline-text": Produces an ASCII text diagram of the operations
            performed by a circuit over time. The qubits are laid out into
            a line top to bottom, and time advances left to right. The input
            object should be a stim circuit.

            INPUT MUST BE A CIRCUIT.

        "timeline-svg": Produces an SVG image diagram of the operations
            performed by a circuit over time. The qubits are laid out into
            a line top to bottom, and time advances left to right. The input
            object should be a stim circuit.

            INPUT MUST BE A CIRCUIT.

        "timeline-3d": Produces a 3d model, in GLTF format, of the
            operations applied by a stim circuit over time.

            GLTF files can be opened with a variety of programs, or
            opened online in viewers such as
            https://gltf-viewer.donmccurdy.com/ .

            INPUT MUST BE A CIRCUIT.

        "timeline-3d-html": A web page containing a 3d model
            viewer of the operations applied by a stim circuit
            over time.

            INPUT MUST BE A CIRCUIT.

        "matchgraph-svg": An image of the decoding graph of a detector
            error model. Red lines are errors crossing a logical observable.

            INPUT MUST BE A DETECTOR ERROR MODEL OR A CIRCUIT.

        "matchgraph-3d": A 3d model, in GLTF format, of the
            decoding graph of a detector error model. Red lines are
            errors crossing a logical observable.

            GLTF files can be opened with a variety of programs, or
            opened online in viewers such as
            https://gltf-viewer.donmccurdy.com/ .

            INPUT MUST BE A DETECTOR ERROR MODEL OR A CIRCUIT.

        "matchgraph-3d-html": A web page containing a 3d model
            viewer of the decoding graph of a detector error
            model or circuit.

            INPUT MUST BE A DETECTOR ERROR MODEL OR A CIRCUIT.

        "detslice-text": An ASCII diagram of the stabilizers
            that detectors declared by the circuit correspond to
            during the TICK instruction identified by the `tick`
            argument.

            INPUT MUST BE A CIRCUIT.

        "detslice-svg": An SVG image of the stabilizers
            that detectors declared by the circuit correspond to
            during the TICK instruction identified by the `tick`
            argument. For example, a detector slice diagram of a
            CSS surface code circuit during the TICK between a
            measurement layer and a reset layer will produce the
            usual diagram of a surface code. Uses the Pauli color convention
            XYZ=RGB.

            INPUT MUST BE A CIRCUIT.

        "timeslice-svg": An SVG image of the operations that a circuit
            applies during the specified tick or range of ticks.

            INPUT MUST BE A CIRCUIT.

        "detslice-with-ops-svg": An SVG image of the operations that a
            circuit applies during the specified tick or range of ticks,
            combined with the detector slices after those operations are
            applied.

            INPUT MUST BE A CIRCUIT.


EXAMPLES
    Example #1
        >>> cat example_circuit.stim
        H 0
        CNOT 0 1

        >>> stim diagram \
            --in example_circuit.stim \
            --type timeline-text
        q0: -H-@-
               |
        q1: ---X-


    Example #2
        >>> # Making a video of detector slices moving around

        >>> # First, make a circuit to animate.
        >>> stim gen \
                --code surface_code \
                --task rotated_memory_x \
                --distance 5 \
                --rounds 100 \
                > surface_code.stim

        >>> # Second, use gnu-parallel and stim diagram to make video frames.
        >>> parallel stim diagram \
            --filter_coords 2,2:4,2 \
            --type detector-slice-svg \
            --tick {} \
            --in surface_code.stim \
            --out video_frame_{}.svg \
            ::: {0050..0150}

        >>> # Third, use ffmpeg to turn the frames into a GIF.
        >>> # (note: the complex filter argument is optional; it turns the background white)
        >>> ffmpeg output_animation.gif \
            -framerate 5 \
            -pattern_type glob -i 'video_frame_*.svg' \
            -pix_fmt rgb8 \
            -filter_complex "[0]split=2[bg][fg];[bg]drawbox=c=white@1:t=fill[bg];[bg][fg]overlay=format=auto"

        >>> # Alternatively, make an MP4 video instead of a GIF.
        >>> ffmpeg output_video.mp4 \
            -framerate 5 \
            -pattern_type glob -i 'video_frame_*.svg' \
            -vf scale=1024:-1 \
            -c:v libx264 \
            -vf format=yuv420p \
            -vf "pad=ceil(iw/2)*2:ceil(ih/2)*2"

stim explain_errors

NAME
    stim explain_errors

SYNOPSIS
    stim explain_errors \
        [--dem_filter filepath] \
        [--in filepath] \
        [--out filepath] \
        [--single]

DESCRIPTION
    Find circuit errors that produce certain detection events.

    Note that this command does not attempt to explain detection events
    by using multiple errors. This command can only tell you how to
    produce a set of detection events if they correspond to a specific
    single physical error annotated into the circuit.

    If you need to explain a detection event set using multiple errors,
    use a decoder such as pymatching to find the set of single detector
    error model errors that are needed and then use this command to
    convert those specific errors into circuit errors.


OPTIONS
    --dem_filter
        Specifies a detector error model to use as a filter.

        If `--dem_filter` isn't specified, an explanation of every single
        set of symptoms that can be produced by the circuit.

        If `--dem_filter` is specified, only explanations of the error
        mechanisms present in the filter will be output. This is useful when
        you are interested in a specific set of detection events.

        The filter is specified as a detector error model file. See
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_dem_detector_error_model.md


    --in
        Chooses the stim circuit file to read the explanatory circuit from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --out
        Chooses where to write the explanations to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is in an arbitrary semi-human-readable format.


    --single
        Explain using a single simple error instead of all possible errors.

        When `--single` isn't specified, every single circuit error that
        produces a specific detector error model is output as a possible
        explanation of that error.

        When `--single` is specified, only the simplest circuit error is
        output. The "simplest" error is chosen by using heuristics such as
        "has fewer Pauli terms" and "happens earlier".


EXAMPLES
    Example #1
        >>> stim gen \
            --code surface_code \
            --task rotated_memory_z \
            --distance 5 \
            --rounds 10 \
            --after_clifford_depolarization 0.001 \
            > example.stim
        >>> echo "error(1) D97 D102" > example.dem

        >>> stim explain_errors \
            --single \
            --in example.stim \
            --dem_filter example.dem
        ExplainedError {
            dem_error_terms: D97[coords 4,6,4] D102[coords 2,8,4]
            CircuitErrorLocation {
                flipped_pauli_product: Z36[coords 3,7]
                Circuit location stack trace:
                    (after 25 TICKs)
                    at instruction #83 (a REPEAT 9 block) in the circuit
                    after 2 completed iterations
                    at instruction #12 (DEPOLARIZE2) in the REPEAT block
                    at targets #3 to #4 of the instruction
                    resolving to DEPOLARIZE2(0.001) 46[coords 2,8] 36[coords 3,7]
            }
        }

stim gen

NAME
    stim gen

SYNOPSIS
    stim gen \
        [--after_clifford_depolarization probability] \
        [--after_reset_flip_probability probability] \
        [--before_measure_flip_probability probability] \
        [--before_round_data_depolarization probability] \
        --code surface_code|repetition_code|color_code \
        --distance int \
        [--out filepath] \
        --rounds int \
        --task name

DESCRIPTION
    Generates example circuits.

    The generated circuits include annotations for noise, detectors, logical
    observables, the spatial locations of qubits, the spacetime locations
    of detectors, and the inexorable passage of TICKs.

    Note that the generated circuits are not intended to be sufficient for
    research. They are really just examples to make it easier to get started
    using Stim, so you can try things without having to first go through
    the entire effort of making a correctly annotated quantum error
    correction circuit.


OPTIONS
    --after_clifford_depolarization
        Specifies a depolarizing noise level for unitary gates.

        Defaults to 0 when not specified.
        Must be a probability (a number between 0 and 1).

        Adds a `DEPOLARIZE1(p)` operation after every single-qubit Clifford
        operation, and a `DEPOLARIZE2(p)` noise operation after every
        two-qubit Clifford operation. When set to 0 or not set, these noise
        operations are not inserted at all.


    --after_reset_flip_probability
        Specifies a reset noise level.

        Defaults to 0 when not specified.
        Must be a probability (a number between 0 and 1).

        Adds an `X_ERROR(p)` after `R` and `RY` operations, and a
        `Z_ERROR(p)` after `RX` operations. When set to 0 or not set,
        these noise operations are not inserted at all.


    --before_measure_flip_probability
        Specifies a measurement noise level.

        Defaults to 0 when not specified.
        Must be a probability (a number between 0 and 1).

        Adds an `X_ERROR(p)` before `M` and `MY` operations, and a
        `Z_ERROR(p)` before `MX` operations. When set to 0 or not set,
        these noise operations are not inserted at all.


    --before_round_data_depolarization
        Specifies a quantum phenomenological noise level.

        Defaults to 0 when not specified.
        Must be a probability (a number between 0 and 1).

        Adds a `DEPOLARIZE1(p)` operation to each data qubit at the start of
        each round of stabilizer measurements. When set to 0 or not set,
        these noise operations are not inserted at all.


    --code
        The error correcting code to use.

        The available error correcting codes are:

        `surface_code`
            The surface code. A quantum code with a checkerboard pattern of
            alternating X and Z stabilizers.
        `repetition_code`
            The repetition code. The simplest classical code.
        `color_code`
            The color code. A quantum code with a hexagonal pattern of
            overlapping X and Z stabilizers.


    --distance
        The minimum number of physical errors that cause a logical error.

        The code distance determines how spatially large the generated
        circuit has to be. Conventionally, the code distance specifically
        refers to single-qubit errors between rounds instead of circuit
        errors during rounds.

        The distance must always be a positive integer. Different
        codes/tasks may place additional constraints on the distance (e.g.
        must be larger than 2 or must be odd or etc).


    --out
        Chooses where to write the generated circuit to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --rounds
        The number of times the circuit's measurement qubits are measured.

        The number of rounds must be an integer between 1 and a quintillion
        (10^18). Different codes/tasks may place additional constraints on
        the number of rounds (e.g. enough rounds to have measured all the
        stabilizers at least once).


    --task
        What the generated circuit should do; the experiment it should run.

        Different error correcting codes support different tasks. The
        available tasks are:

        `memory` (repetition_code):
            Initialize a logical `|0>`,
            preserve it against noise for the given number of rounds,
            then measure.
        `rotated_memory_x` (surface_code):
            Initialize a logical `|+>` in a rotated surface code,
            preserve it against noise for the given number of rounds,
            then measure in the X basis.
        `rotated_memory_z` (surface_code):
            Initialize a logical `|0>` in a rotated surface code,
            preserve it against noise for the given number of rounds,
            then measure in the X basis.
        `unrotated_memory_x` (surface_code):
            Initialize a logical `|+>` in an unrotated surface code,
            preserve it against noise for the given number of rounds,
            then measure in the Z basis.
        `unrotated_memory_z` (surface_code):
            Initialize a logical `|0>` in an unrotated surface code,
            preserve it against noise for the given number of rounds,
            then measure in the Z basis.
        `memory_xyz` (color_code):
            Initialize a logical `|0>`,
            preserve it against noise for the given number of rounds,
            then measure. Use a color code that alternates between measuring
            X, then Y, then Z stabilizers.


EXAMPLES
    Example #1
        >>> stim gen \
            --code repetition_code \
            --task memory \
            --distance 3 \
            --rounds 100 \
            --after_clifford_depolarization 0.001
        # Generated repetition_code circuit.
        # task: memory
        # rounds: 100
        # distance: 3
        # before_round_data_depolarization: 0
        # before_measure_flip_probability: 0
        # after_reset_flip_probability: 0
        # after_clifford_depolarization: 0.001
        # layout:
        # L0 Z1 d2 Z3 d4
        # Legend:
        #     d# = data qubit
        #     L# = data qubit with logical observable crossing
        #     Z# = measurement qubit
        R 0 1 2 3 4
        TICK
        CX 0 1 2 3
        DEPOLARIZE2(0.001) 0 1 2 3
        TICK
        CX 2 1 4 3
        DEPOLARIZE2(0.001) 2 1 4 3
        TICK
        MR 1 3
        DETECTOR(1, 0) rec[-2]
        DETECTOR(3, 0) rec[-1]
        REPEAT 99 {
            TICK
            CX 0 1 2 3
            DEPOLARIZE2(0.001) 0 1 2 3
            TICK
            CX 2 1 4 3
            DEPOLARIZE2(0.001) 2 1 4 3
            TICK
            MR 1 3
            SHIFT_COORDS(0, 1)
            DETECTOR(1, 0) rec[-2] rec[-4]
            DETECTOR(3, 0) rec[-1] rec[-3]
        }
        M 0 2 4
        DETECTOR(1, 1) rec[-2] rec[-3] rec[-5]
        DETECTOR(3, 1) rec[-1] rec[-2] rec[-4]
        OBSERVABLE_INCLUDE(0) rec[-1]

stim help

NAME
    stim help

SYNOPSIS
    stim help

DESCRIPTION
    Prints helpful information about using stim.

stim m2d

NAME
    stim m2d

SYNOPSIS
    stim m2d \
        [--append_observables] \
        --circuit filepath \
        [--in filepath] \
        [--in_format 01|b8|r8|ptb64|hits|dets] \
        [--obs_out filepath] \
        [--obs_out_format 01|b8|r8|ptb64|hits|dets] \
        [--out filepath] \
        [--out_format 01|b8|r8|ptb64|hits|dets] \
        [--ran_without_feedback] \
        [--skip_reference_sample] \
        --sweep filepath \
        [--sweep_format 01|b8|r8|ptb64|hits|dets]

DESCRIPTION
    Convert measurement data into detection event data.

    When sampling data from hardware, instead of from simulators, it's
    necessary to convert the measurement data into detection event data that
    can be fed into a decoder. This is necessary both because of
    complexities in the exact sets of measurements being compared by a
    circuit to produce detection events and also because the expected parity
    of a detector's measurement set can vary due to (for example) spin echo
    operations in the circuit.

    Stim performs this conversion by simulating taking a reference sample
    from the circuit, in order to determine the expected parity of the
    measurements sets defining detectors and observables, and then comparing
    the sampled measurement data to these expectations.


OPTIONS
    --append_observables
        Appends observable flips to the end of samples as extra detectors.

        PREFER --obs_out OVER THIS FLAG. Mixing the observable flip data
        into detection event data tends to require simply separating them
        again immediately, creating unnecessary work. For example, when
        testing a decoder, you do not want to give the observable flips to
        the decoder because that is the information the decoder is supposed
        to be predicting from the detection events.

        This flag causes observable flip data to be appended to each sample,
        as if the observables were extra detectors at the end of the
        circuit. For example, if there are 100 detectors and 10 observables
        in the circuit, then the output will contain 110 detectors and the
        last 10 are the observables.


    --circuit
        Specifies where the circuit that generated the measurements is.

        This argument is required, because the circuit is what specifies how
        to derive detection event data from measurement data.

        The circuit file should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --in
        Chooses the file to read measurement data from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input's format is specified by `--in_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --in_format
        Specifies the data format to use when reading measurement data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --obs_out
        Specifies the file to write observable flip data to.

        When producing detection event data, the goal is typically to
        predict whether or not the logical observables were flipped by using
        the detection events. This argument specifies where to write that
        observable flip data.

        If this argument isn't specified, the observable flip data isn't
        written to a file.

        The output is in a format specified by `--obs_out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --obs_out_format
        Specifies the data format to use when writing observable flip data.

        Irrelevant unless `--obs_out` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out
        Chooses where to write the sampled data to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output's format is specified by `--out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out_format
        Specifies the data format to use when writing output detection data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --ran_without_feedback
        Converts the results assuming all feedback operations were skipped.

        Pauli feedback operations don't need to be performed on the quantum
        computer. They can be performed within the classical control system.
        As such, it often makes sense to skip them when sampling from the
        circuit on hardware, and then account for this later during data
        analysis. Turning on this flag means that the quantum computer
        didn't apply the feedback operations, and it's the job of the m2d
        conversion to read the measurement data, rewrite it to account for
        feedback effects, then convert to detection events.

        In the python API, the same effect can be achieved by using
        stim.Circuit.with_inlined_feedback().compile_m2d_converter().


    --skip_reference_sample
        Asserts the circuit can produce a noiseless sample that is just 0s.

        When this argument is specified, the reference sample (that the
        measurement data will be compared to) is generated by simply setting
        all measurements to 0 instead of by simulating the circuit without
        noise.

        Skipping the reference sample can significantly improve performance,
        because acquiring the reference sample requires using the tableau
        simulator. If the vacuous reference sample is actually a result that
        can be produced by the circuit, under noiseless execution, then
        specifying this flag has no observable outcome other than improving
        performance.

        CAUTION. When the all-zero sample isn't a result that can be
        produced by the circuit under noiseless execution, specifying this
        flag will cause incorrect output to be produced.


    --sweep
        Specifies a file to read sweep configuration data from.

        Sweep bits are used to vary whether certain Pauli gates are included
        in a circuit, or not, from shot to shot. For example, if a circuit
        contains the instruction "CX sweep[5] 0" then there is an X pauli
        that is included only in shots where the corresponding sweep data
        has the bit at index 5 set to True.

        If `--sweep` is not specified, all sweep bits default to OFF. If
        `--sweep` is specified, each shot's sweep configuratoin data is
        read from the specified file.

        The sweep data's format is specified by `--sweep_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --sweep_format
        Specifies the data format to use when reading sweep config data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


EXAMPLES
    Example #1
        >>> cat example_circuit.stim
        X 0
        M 0 1
        DETECTOR rec[-2]
        DETECTOR rec[-1]
        OBSERVABLE_INCLUDE(2) rec[-1]

        >>> cat example_measure_data.01
        00
        01
        10
        11

        >>> stim m2d \
            --append_observables \
            --circuit example_circuit.stim \
            --in example_measure_data.01 \
            --in_format 01 \
            --out_format dets
        shot D0
        shot D0 D1 L2
        shot
        shot D1 L2

stim repl

NAME
    stim repl

SYNOPSIS
    stim repl

DESCRIPTION
    Runs stim in interactive read-evaluate-print (REPL) mode.

    Reads operations from stdin while immediately writing measurement
    results to stdout.


EXAMPLES
    Example #1
        >>> stim repl
        ... M 0
        0
        ... X 0
        ... M 0
        1
        ... X 2 3 9
        ... M 0 1 2 3 4 5 6 7 8 9
        1 0 1 1 0 0 0 0 0 1
        ... REPEAT 5 {
        ...     R 0 1
        ...     H 0
        ...     CNOT 0 1
        ...     M 0 1
        ... }
        00
        11
        11
        00
        11

stim sample

NAME
    stim sample

SYNOPSIS
    stim sample \
        [--in filepath] \
        [--out filepath] \
        [--out_format 01|b8|r8|ptb64|hits|dets] \
        [--seed int] \
        [--shots int] \
        [--skip_reference_sample]

DESCRIPTION
    Samples measurements from a circuit.

OPTIONS
    --in
        Chooses the stim circuit file to read the circuit to sample from.

        By default, the circuit is read from stdin. When `--in $FILEPATH` is
        specified, the circuit is instead read from the file at $FILEPATH.

        The input should be a stim circuit. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_stim_circuit.md


    --out
        Chooses where to write the sampled data to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is in a format specified by `--out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out_format
        Specifies the data format to use when writing output data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --seed
        Makes simulation results PARTIALLY deterministic.

        The seed integer must be a non-negative 64 bit signed integer.

        When `--seed` isn't specified, the random number generator is seeded
        using fresh entropy requested from the operating system.

        When `--seed #` is set, the exact same simulation results will be
        produced every time ASSUMING:

        - the exact same other flags are specified
        - the exact same version of Stim is being used
        - the exact same machine architecture is being used (for example,
            you're not switching from a machine that has AVX2 instructions
            to one that doesn't).

        CAUTION: simulation results *WILL NOT* be consistent between
        versions of Stim. This restriction is present to make it possible to
        have future optimizations to the random sampling, and is enforced by
        introducing intentional differences in the seeding strategy from
        version to version.

        CAUTION: simulation results *MAY NOT* be consistent across machines.
        For example, using the same seed on a machine that supports AVX
        instructions and one that only supports SSE instructions may produce
        different simulation results.

        CAUTION: simulation results *MAY NOT* be consistent if you vary
        other flags and modes. For example, `--skip_reference_sample` may
        result in fewer calls the to the random number generator before
        reported sampling begins. More generally, using the same seed for
        `stim sample` and `stim detect` will not result in detection events
        corresponding to the measurement results.


    --shots
        Specifies the number of samples to take from the circuit.

        Defaults to 1.
        Must be an integer between 0 and a quintillion (10^18).


    --skip_reference_sample
        Asserts the circuit can produce a noiseless sample that is just 0s.

        When this argument is specified, the reference sample (that is used
        to convert measurement flip data from frame simulations into actual
        measurement data) is generated by simply setting all measurements to
        0 instead of by performing a stabilizer tableau simulation of the
        circuit without noise.

        Skipping the reference sample can significantly improve performance,
        because acquiring the reference sample requires using the tableau
        simulator. If the vacuous reference sample is actually a result that
        can be produced by the circuit, under noiseless execution, then
        specifying this flag has no observable outcome other than improving
        performance.

        CAUTION. When the all-zero sample isn't a result that can be
        produced by the circuit under noiseless execution, specifying this
        flag will cause incorrect output to be produced. Specifically, the
        output measurement bits will be whether each measurement was
        *FLIPPED* instead of the actual absolute value of the measurement.


EXAMPLES
    Example #1
        >>> cat example_circuit.stim
        H 0
        CNOT 0 1
        M 0 1

        >>> stim sample --shots 5 < example_circuit.stim
        00
        11
        11
        00
        11



    Example #2
        >>> cat example_circuit.stim
        X 2 3 5
        M 0 1 2 3 4 5 6 7 8 9

        >>> stim sample --in example_circuit.stim --out_format dets
        shot M2 M3 M5

stim sample_dem

NAME
    stim sample_dem

SYNOPSIS
    stim sample_dem \
        [--err_out filepath] \
        [--err_out_format 01|b8|r8|ptb64|hits|dets] \
        [--in filepath] \
        [--obs_out filepath] \
        [--obs_out_format 01|b8|r8|ptb64|hits|dets] \
        [--out filepath] \
        [--out_format 01|b8|r8|ptb64|hits|dets] \
        [--replay_err_in filepath] \
        [--replay_err_in_format 01|b8|r8|ptb64|hits|dets] \
        [--seed int] \
        [--shots int]

DESCRIPTION
    Samples detection events from a detector error model.

    Supports recording and replaying the errors that occurred.


OPTIONS
    --err_out
        Specifies a file to write a record of which errors occurred.

        For example, the errors that occurred can be analyzed, modified, and
        then given to `--replay_err_in` to see the effects of changes.

        The output is in a format specified by `--err_out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --err_out_format
        Specifies the data format to use when writing recorded error data.

        Irrelevant unless `--err_out` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --in
        Chooses the file to read the detector error model to sample from.

        By default, the detector error model is read from stdin. When
        `--in $FILEPATH` is specified, the detector error model is instead
        read from the file at $FILEPATH.

        The input should be a stim detector error model. See:
        https://github.com/quantumlib/Stim/blob/main/doc/file_format_dem_detector_error_model.md


    --obs_out
        Specifies the file to write observable flip data to.

        When sampling detection event data, the goal is typically to predict
        whether or not the logical observables were flipped by using the
        detection events. This argument specifies where to write that
        observable flip data.

        If this argument isn't specified, the observable flip data isn't
        written to a file.

        The output is in a format specified by `--obs_out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --obs_out_format
        Specifies the data format to use when writing observable flip data.

        Irrelevant unless `--obs_out` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out
        Chooses where to write the sampled data to.

        By default, the output is written to stdout. When `--out $FILEPATH`
        is specified, the output is instead written to the file at $FILEPATH.

        The output is in a format specified by `--out_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --out_format
        Specifies the data format to use when writing output data.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --replay_err_in
        Specifies a file to read error data to replay from.

        When replaying error information, errors are no longer sampled
        randomly but are instead driven by the file data. For example, this
        file data could come from a previous run that wrote error data using
        `--err_out`.

        The input is in a format specified by `--err_in_format`. See:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --replay_err_in_format
        Specifies the data format to use when reading error data to replay.

        Irrelevant unless `--replay_err_in` is specified.

        The available formats are:

            01 (default): dense human readable
            b8: bit packed binary
            r8: run length binary
            ptb64: partially transposed bit packed binary for SIMD
            hits: sparse human readable
            dets: sparse human readable with type hints

        For a detailed description of each result format, see the result
        format reference:
        https://github.com/quantumlib/Stim/blob/main/doc/result_formats.md


    --seed
        Makes simulation results PARTIALLY deterministic.

        The seed integer must be a non-negative 64 bit signed integer.

        When `--seed` isn't specified, the random number generator is seeded
        using fresh entropy requested from the operating system.

        When `--seed #` is set, the exact same simulation results will be
        produced every time ASSUMING:

        - the exact same other flags are specified
        - the exact same version of Stim is being used
        - the exact same machine architecture is being used (for example,
            you're not switching from a machine that has AVX2 instructions
            to one that doesn't).

        CAUTION: simulation results *WILL NOT* be consistent between
        versions of Stim. This restriction is present to make it possible to
        have future optimizations to the random sampling, and is enforced by
        introducing intentional differences in the seeding strategy from
        version to version.

        CAUTION: simulation results *MAY NOT* be consistent across machines.
        For example, using the same seed on a machine that supports AVX
        instructions and one that only supports SSE instructions may produce
        different simulation results.

        CAUTION: simulation results *MAY NOT* be consistent if you vary
        other flags and modes. For example, `--skip_reference_sample` may
        result in fewer calls the to the random number generator before
        reported sampling begins. More generally, using the same seed for
        `stim sample` and `stim detect` will not result in detection events
        corresponding to the measurement results.


    --shots
        Specifies the number of samples to take from the detector error model.

        Defaults to 1.
        Must be an integer between 0 and a quintillion (10^18).


EXAMPLES
    Example #1
        >>> cat example.dem
        error(0) D0
        error(0.5) D1 L0
        error(1) D2 D3

        >>> stim sample_dem \
            --shots 5 \
            --in example.dem \
            --out dets.01 \
            --out_format 01 \
            --obs_out obs_flips.01 \
            --obs_out_format 01

        >>> cat dets.01
        0111
        0011
        0011
        0111
        0111

        >>> cat obs_flips.01
        1
        0
        0
        1
        1